Reading edition: all ten original pitches, agent-first assessments and cited sources. Scores are editorial judgments with low evidence confidence. No live agents are deployed by this pitchbook.
ShiftBrief: The best employee’s know-how, ready for the next hire.
Turn one narrated screen recording into three owner-approved training clips, a checklist and a knowledge check. Sell completed onboarding, then automate delivery.
Buyer: A 10–50-person agency or ecommerce team's owner, with a new hire starting next week.
Offer: $299 — 15 source minutes → three 45–90 sec clips, checklist, quiz and one correction round. Delivery in 48 hours.
Cash budget: $6,200, excluding founder salary.
Thesis
Small teams already have recordings and repeated questions. They lack time to edit, structure and approve useful training. A fixed deliverable addresses a real hiring deadline without waiting for enterprise procurement.
Every instruction links to the source; missing policy becomes an owner question. Deliver a usable pack and clear update path. Cheap DIY tools already make videos, so completion and accuracy must earn the premium.
75/100B · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot, measured conversion or timed delivery data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
An owner with a hire starting soon has a specific outcome and deadline.
8/10
Standard inputs and bounded delivery15% weight · 4/5
One 15-minute recording, three clips, a checklist and one revision bound the work.
12/15
Agents perform the production work20% weight · 4/5
Agents can transcribe, segment, draft and assemble; the founder must still verify procedure fidelity.
16/20
Errors can be detected before delivery15% weight · 4/5
Every claim can point to source timing, with owner approval for missing policy.
12/15
Small, bounded founder workload15% weight · 3/5
35 minutes is a target for QA, exceptions and support, down from the existing 90-minute labor assumption.
9/15
Room for labor, retries and acquisition10% weight · 4/5
The $299 offer has room above the existing $84 fulfillment allowance, fees and cash acquisition.
8/10
A founder can reach buyers10% weight · 3/5
Direct owner outreach is plausible; the founder still needs access to teams that are hiring.
6/10
Repeat work reuses approved assets5% weight · 4/5
The same team may need process updates and training for later hires.
4/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Intake agentValidated intake and missing-information list
Input
One authorized recording, desired learner and owner questionnaire
Owns
Check source duration and usable audio Request missing steps or access-safe footage Create a job record and scope checklist
Human approval
Founder rejects regulated/high-risk training and confirms scope.
2Procedure agentSource-linked step map and quiz draft
Input
Recording transcript with timestamps and owner facts
Owns
Extract the actual steps Link each instruction to source evidence Flag missing or contradictory policy
Human approval
Owner supplies missing policy; founder prevents guesses from becoming instructions.
3Production agentThree draft clips plus documents
Input
Approved step map, source footage and clip templates
Owns
Cut three 45–90-second clips from the real footage Generate captions, checklist and quiz Assemble downloadable files
Human approval
Founder spot-checks source alignment and readability; no invented procedural footage.
4QA and delivery agentReview packet, delivery package and correction log
Input
Draft files and approved step map
Owns
Check duration, captions, broken links and required steps Prepare an exception report and owner review link Package approved files and track one revision
Human approval
Founder resolves flagged issues; buyer approves before final export.
Your job as founder
Qualify buyers and keep the 15-minute source scope fixed.
Approve procedural accuracy and resolve ambiguous policy with the owner.
Review the exception report, authorize release and handle the one revision.
Required skill: The founder must understand the selected business workflow well enough to verify instructions; the customer remains the policy authority.
The shift to agents: Replace founder video editing with source-linked segmentation and template assembly. Keep founder release review and owner policy approval; a draft-writing assistant alone would leave this a manual editing service.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
The founder must understand the selected business workflow well enough to verify instructions; the customer remains the policy authority.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $161.08 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 24 orders/week at 35 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $6,200 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $150. Build three packs from permissioned screen recordings using one fixed template; time every founder touch. Then offer the scoped $299 pilot to 20 relevant owners. Pass: At least 2 paid pilots; no unsupported procedure; median founder handling at or below 35 minutes by the fifth pack. Stop: If accurate delivery still takes over 90 minutes or owners will not approve source facts, keep it a service experiment or narrow the niche.
First sale
Before launch: Make one useful sample Record your own workflow. Produce the complete pack and a $299 scope page.
Day 1: Sell against an actual hire date Use five pre-booked owner demos to sell one pilot. If there is no warm network, test 30 qualified prospects before building.
Next 48 hours: Fulfill and observe Owner approves every procedure. Time the work and ask which repeated questions the pack removes.
Sources
Guidde pricing: Pro lists $29 per creator/month ($19/month billed annually); Business $59/month ($39/month billed annually). Businesses can buy video documentation, voice and exports.
Model roles
Muse Voice Transcribe: Transcript and turn timing from the manager's recording.
GPT-6 Astra: Source-linked steps, teaching script, quiz and unresolved-policy questions.
MiniMax Speech: Narrates the owner-approved script over original footage.
Stop criteria
Fewer than 2 paid pilots from 20 qualified conversations in 14 days, or median fulfillment over 90 minutes after five pilots.
SKUStudio: Six localized ads. One product. One fixed price.
Turn a 90-second voice brief and approved product images into six 15-second vertical ads: three hooks, each in English and Spanish, plus source captions and a one-page testing plan. Deliver within 48 hours and include one consolidated revision. This is a narrow done-for-you service first, a self-serve app after repeatable demand.
Buyer: Owner-led Shopify brand selling simple physical accessories, home goods or stationery; one launch or seasonal promotion with a real deadline. Exclude regulated products and unsupported performance claims at launch.
Offer: $199 — Three 15s hooks × English/Spanish = 6 vertical ads; captions, one revision, 48-hour delivery. Ad spend excluded.
Cash budget: $6,500, excluding founder salary.
Thesis
A $199 creative production purchase is a smaller owner decision than agency retainers. Finished localized assets, product fidelity, and one accountable delivery deadline can command more than access to a generic video tool. Distribution and review quality matter more than the model.
Owner-led Shopify accessory brands launching one SKU; preserve the actual product photo and approved facts.
68/100B · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot, measured conversion or timed delivery data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
A small ecommerce owner can pay for a launch with a real deadline.
8/10
Standard inputs and bounded delivery15% weight · 4/5
OneSKU, approved facts, three hooks and six exports make scope repeatable.
12/15
Agents perform the production work20% weight · 3/5
Compositing and captions can run automatically, but visual selection and bilingual finishing remain material.
12/20
Errors can be detected before delivery15% weight · 3/5
Product fidelity is checkable against the real image; creative quality and translation need founder judgment.
9/15
Small, bounded founder workload15% weight · 3/5
40 minutes is an unmeasured target versus the existing 75-minute assumption.
9/15
Room for labor, retries and acquisition10% weight · 4/5
At $199 there is room above the existing $60 fulfillment allowance, but rerenders and revisions must stay capped.
8/10
A founder can reach buyers10% weight · 3/5
Owner outreach is feasible; cheap self-serve competition means the actual channel still needs testing.
6/10
Repeat work reuses approved assets5% weight · 4/5
Saved brand facts and previous winning hooks can support later campaigns.
4/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Brief agentLocked product fact sheet and missing-input requests
Input
90-second voice brief, approved product photos and brand facts
Owns
Transcribe the brief Separate authorized claims from suggestions Validate image quality and promotion dates
Human approval
Founder checks scope and obtains owner approval for claims.
2Creative plannerThree storyboard/script pairs
Input
Approved facts, brand kit and three-hook template
Owns
Draft three distinct 15-second scripts Prepare English/Spanish captions Specify background-only visual prompts and real-photo overlays
Human approval
Founder with bilingual competence approves tone, claims and translation.
3Production agentSix draft vertical ads
Input
Approved boards, real product cutouts and templates
Owns
Generate bounded background variations Composite the actual product image Render six versions with captions and audio
Human approval
Founder checks product fidelity and rejects misleading visual results.
4QA and delivery agentRelease checklist and six-file delivery package
Input
Exports, approved facts and scope limits
Owns
Check durations, dimensions, subtitles and spelling Compare claims against approved facts Assemble review sheet and process one consolidated revision
Human approval
Founder resolves flagged issues and approves delivery; buyer publishes.
Your job as founder
Find launch-stage buyers and enforce oneSKU/one revision.
Approve bilingual claims and visually inspect every ad.
Handle unusual products, billing and failed renders.
Required skill: The founder must be able to judge product advertising claims, visual fidelity and English/Spanish copy; otherwise narrow to one verified language.
The shift to agents: Build a real-photo compositing pipeline with reusable layouts, bounded background generation and one review batch. Do not assume agents can autonomously judge product fidelity or creative taste.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
The founder must be able to judge product advertising claims, visual fidelity and English/Spanish copy; otherwise narrow to one verified language.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $97.98 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 21 orders/week at 40 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $6,500 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $150. Use permissioned photos from three simple product types to run the same pipeline; log founder minutes and retries. Offer one $199 pack to 30 qualified owners. Pass: At least 2 paid pilots; six technically valid exports per order; founder time at or below 40 minutes after five packs. Stop: Stop the agent-first claim if median handling stays above 75 minutes or product distortion reaches a release candidate unchecked.
First sale
Before launch: Make two credible samples Create permissioned packs and a checkout. Identify 30 small brands with a real upcoming launch.
Launch day: Sell three pilot slots Founder shows 10 qualified owners a sample concept and offers $199 upfront. Target one sale; no guarantee.
After five orders: Measure the operation Track edits, sales time and repeat orders. Build Shopify integration only after repeat demand.
Sources
Creatify pricing: Starter lists $39/month with 100 credits; Pro $99/month with 300 credits. A serious low-cost substitute.
Billo managed service: Managed creator marketing starts at $2,000/month, billed monthly in advance. Broader service and human-creator sourcing, so not directly equivalent.
Model roles
GPT-6 Astra: Turns approved facts into hooks, bilingual scripts and shotlists; flags unsupported claims.
H3 Max: Creates short abstract background/title clips only. Real products or property features remain in unaltered source images.
Muse Voice Transcribe: Transcribes the owner’s brief. A separate TTS service voices approved text; Muse does not synthesize speech.
Stop criteria
Fewer than 2 paid pilots from 30 qualified owner conversations, or median fulfillment above 100 minutes after 5 packs.
Toast Together: One speech. Both families.
Turn your real stories into a bilingual wedding speech you can deliver with confidence.
Buyer: A parent, sibling or newlywed speaking at an English–Spanish wedding within a month.
Offer: $69 — A three-minute speech, bilingual review, printable cue cards and one revision.
Cash budget: $5,200, excluding founder salary.
Thesis
A wedding creates a deadline, an emotional reason to pay and a finite deliverable. Sell one event well, then earn planner referrals; monthly retention is unlikely.
A native-language reviewer preserves the anecdote, checks names and selects bilingual passages without doubling the speech. The buyer gets a reviewed speech and rehearsal cards, not another writing interface.
62/100C · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot, measured conversion or timed delivery data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
A scheduled wedding supplies a strong purchase deadline.
8/10
Standard inputs and bounded delivery15% weight · 3/5
A three-minute speech and one revision constrain output, but emotional intake varies.
9/15
Agents perform the production work20% weight · 4/5
Agents can organize real anecdotes and produce cards and rehearsal audio.
16/20
Errors can be detected before delivery15% weight · 2/5
Cultural tone, family sensitivities and meaning preservation depend on a skilled human.
6/15
Small, bounded founder workload15% weight · 4/5
20 minutes retains the original human allowance; agent-first framing cannot remove sensitive review.
12/15
Room for labor, retries and acquisition10% weight · 3/5
The $69 price leaves a base-case margin, but emotional revisions and paid acquisition can consume it.
6/10
A founder can reach buyers10% weight · 2/5
Planner referrals are plausible but require earned trust; repeat consumer acquisition is difficult.
4/10
Repeat work reuses approved assets5% weight · 1/5
Most customers need one wedding speech; repeated sales depend on referrals.
1/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Story intake agentSpeaker-approved anecdote and pronunciation sheet
Input
Speaker voice notes, relationship, names and language preferences
Owns
Transcribe the speaker’s own stories Ask structured questions for missing details Mark facts and anecdotes that were actually supplied
Human approval
Buyer confirms names, relationships and which stories are appropriate.
2Speechwriter agentSpeech draft with source-linked anecdotes
Input
Approved story sheet and three-minute structure
Owns
Draft a coherent speech using only supplied facts Create paired bilingual passages without doubling length Flag potentially sensitive humor
Human approval
Founder reviews emotional tone, cultural meaning and any sensitive material.
3Rehearsal agentCue cards and rehearsal pack
Input
Approved speech and pronunciation notes
Owns
Create cue cards and timed rehearsal sections Generate clearly labeled device/stock narration if wanted Assemble printable and editable files
Human approval
Founder verifies pronunciation guidance; buyer approves final speech.
4QA and delivery agentRelease checklist and private delivery package
Input
Approved facts and output pack
Owns
Check names, omissions and estimated duration Diff the single revision against approved facts Deliver files and apply the retention/deletion policy
Human approval
Founder resolves translation flags and authorizes final delivery.
Your job as founder
Earn planner referrals and set emotional/revision boundaries.
Perform native-level bilingual and cultural review.
Resolve sensitive family context and approve each delivery.
Required skill: The founder must provide native-level English/Spanish editing and cultural judgment; a separate paid reviewer would change the stated solo-founder operation.
The shift to agents: Make structured story intake, factual drafting and cue-card delivery an agent pipeline; keep a short founder sensitivity review. This is feasible only for a founder with the needed bilingual editorial skill.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
The founder must provide native-level English/Spanish editing and cultural judgment; a separate paid reviewer would change the stated solo-founder operation.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $35.25 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 43 orders/week at 20 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $5,200 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $75. Run five permissioned mock speeches from structured intake with a bilingual founder; time review and one requested revision. Offer the $69 pack to 25 qualified speech givers through an existing planner contact. Pass: At least 3 purchases; no invented anecdote or name error; average total founder handling at or below 20 minutes. Stop: Narrow the cultural/language scope or reprice if average handling exceeds 30 minutes; pause any workflow that releases invented stories.
First sale
Before launch: Recruit the channel Show a fictional or permissioned sample to five bilingual planners and one celebrant. Secure opt-in introductions before relying on launch traffic.
First selling day: Offer three paid slots Charge $69 with 48-hour delivery. One sale from ten qualified viewers is an experiment, not a forecast.
After delivery: Measure revision work Ask whether the speaker used the speech; log editing time and seek permission before showing any excerpt.
Sources
Speechy services: Advertises SpeechyAI at £49, editing at £145 and bespoke writing at £275.
Model roles
GPT-6 Astra: Structure the story and translate it while retaining only approved facts.
Muse Voice Transcribe: Transcribe the speaker’s story; a separate model translates and a licensed service supplies narration.
MiniMax H3 Max: Optional $19 add-on: a short labeled visual vignette from authorized material. No promised likeness or lip sync.
Stop criteria
Rework after fewer than 3 purchases from 25 qualified speech givers, or average editing over 30 minutes. Pause if invented anecdotes or meaning errors survive review.
ListingLocale: The real property. Clearly presented in two languages.
Eight actual listing photos and a short agent voice brief become a 30-second listing video in English and Spanish, vertical and horizontal, plus subtitles and approved captions. Delivery in 24–48 hours, one revision. Actual property imagery stays unchanged; use photo pans and simple cuts instead of inventing camera viewpoints.
Buyer: Independent residential agent or small real-estate photographer serving English/Spanish buyers; agent or photographer can approve a small one-off invoice.
Offer: $179 — One 30s listing video in EN/ES × vertical/horizontal = 4 exports; captions, one revision, 24–48 hours. No filming.
Cash budget: $6,400, excluding founder salary.
Thesis
A listing is a concrete transaction with a deadline and existing marketing spend. An approved multilingual deliverable can command a service fee above cheap self-serve software. The first sale can be made to a sole agent or photographer without enterprise procurement.
Independent agents and photographers with a listing going live this week and existing authorized photos.
84/100A · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot, measured conversion or timed delivery data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
A listing going live creates a clear near-term marketing purchase.
8/10
Standard inputs and bounded delivery15% weight · 5/5
Eight photos, one approved fact sheet and fixed exports allow deterministic production.
15/15
Agents perform the production work20% weight · 4/5
Agents can plan, caption and render photo pans while founder checks facts and release.
16/20
Errors can be detected before delivery15% weight · 4/5
Actual photos and a signed-off fact sheet provide strong references for checks.
12/15
Small, bounded founder workload15% weight · 4/5
20 minutes is a target for fact, translation and export review versus the original 60-minute assumption.
12/15
Room for labor, retries and acquisition10% weight · 4/5
The $179 offer has room above the original $45 fulfillment allowance; extra formats must remain excluded.
8/10
A founder can reach buyers10% weight · 4/5
Agents and photographers are concentrated, reachable buyers with listing-specific triggers.
8/10
Repeat work reuses approved assets5% weight · 5/5
A photographer or agent can reuse brokerage styles and glossary on every new listing.
5/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
Eight authorized photos, voice brief and structured property facts
Owns
Check photo count and image rights declaration Extract the provided address, features and approved language Request confirmation for every missing fact
Human approval
Founder checks permission and scope; agent/photographer approves factual content.
2Script and translation agentPaired scripts with fact references
Input
Approved property facts and brokerage wording rules
Owns
Draft a 30-second script using only approved facts Prepare English/Spanish captions and narration Flag ambiguous property terminology
Create deterministic pans and cuts Lay out captions and authorized stock narration Render both languages in two aspect ratios
Human approval
Founder confirms photos remain truthful and important details are not cropped misleadingly.
4Release agentFour exports and an approval record
Input
Exports and approved fact sheet
Owns
Check address, duration, dimensions and text consistency Prepare a side-by-side review checklist Deliver approved files and handle the scoped revision
Human approval
Founder approves final quality; buyer approves before publication.
Your job as founder
Build referral relationships with photographers and listing agents.
Approve property facts, translation and representation fidelity.
Keep one listing/eight photos/one revision boundaries and handle exceptions.
Required skill: The founder needs bilingual property-marketing review competence, and the listing professional must approve facts and applicable brokerage language.
The shift to agents: Use deterministic photo animation, a strict fact schema and reusable brokerage templates so the founder reviews rather than edits the video. Generated property imagery is outside the workflow.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
The founder needs bilingual property-marketing review competence, and the listing professional must approve facts and applicable brokerage language.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $89.56 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 43 orders/week at 20 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $6,400 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $100. Produce five permissioned listing packs from exactly eight photos each; time fact review, translation and revisions. Offer the $179 pilot to 20 agents or photographers with a current listing. Pass: At least 2 paid pilots; zero invented property features; median founder handling at or below 20 minutes. Stop: Pause release on any fabricated feature; redesign if routine deliveries still need more than 60 minutes of founder work.
First sale
Before launch: Prepare two approved examples Use authorized photos and contact 15 local independent agents or property photographers.
Launch day: Find one live deadline Offer a $179 pilot for a listing going live this week. Show a watermarked preview; target one paid order, not a forecast.
After five listings: Test a referral channel Track completion time and repeat listings. A photographer referral fee must fit the $30 cash CAC target.
Sources
PhotoAIVideo pricing: Creator displays $450 billed annually for 52 video projects ($37.50/month equivalent); free plan offers 1 watermarked project. A low-cost self-serve competitor.
Katie Caswell Creative real-estate price sheet: Provider PDF lists a basic video package at $225 including interior/exterior footage and a 20-second edited HD video. Includes shooting, so it is a broader service, not a like-for-like quote.
Model roles
GPT-6 Astra: Turns approved facts into hooks, bilingual scripts and shotlists; flags unsupported claims.
H3 Max: Creates short abstract background/title clips only. Real products or property features remain in unaltered source images.
Muse Voice Transcribe: Transcribes the owner’s brief. A separate TTS service voices approved text; Muse does not synthesize speech.
Stop criteria
Fewer than 2 paid pilots from 20 listing-specific conversations, or fulfillment above 90 minutes after 5 orders. Any fabricated property feature is a delivery failure.
PitchReel: Show the idea before anyone books the shoot.
Give a small agency three visual directions, a storyboard and short motion studies for tomorrow's client pitch. Sell a faster creative decision.
Buyer: A small agency owner or freelance producer with a client pitch in the next 48 hours.
Offer: $299 — Three directions → one six-frame board, three 5-second motion tests at 768p, pitch PDF and one revision.
Cash budget: $6,400, excluding founder salary.
Thesis
Fast H3 Max generation makes multiple motion studies economical. Astra converts the brief into distinct shot plans; human selection turns outputs into a coherent client conversation. The buyer pays for a usable decision pack.
Brand constraints, shot selection, presentation and a bounded revision earn the fee. This is a finished previsualization service in a deadline-driven niche, not another general video generator.
49/100D · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot, measured conversion or timed delivery data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
An agency pitching tomorrow has urgency and an accountable buyer.
8/10
Standard inputs and bounded delivery15% weight · 2/5
Three directions are bounded numerically, but taste and brand fit are highly variable.
6/15
Agents perform the production work20% weight · 2/5
Agents generate many options; coherent creative selection and presentation remain founder work.
8/20
Errors can be detected before delivery15% weight · 2/5
Technical checks are automatable, but clients judge originality and aesthetic coherence.
6/15
Small, bounded founder workload15% weight · 2/5
75 minutes is an aggressive target against the current 120-minute production allowance.
6/15
Room for labor, retries and acquisition10% weight · 3/5
The $299 price covers the stated base case, but bespoke revisions and long selection cycles consume headroom.
6/10
A founder can reach buyers10% weight · 3/5
Small agencies are reachable through focused relationships, though work is lumpy.
6/10
Repeat work reuses approved assets5% weight · 3/5
Agency taste libraries can help later briefs, but recurrence depends on winning another brief.
3/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Brief analystConstraint sheet and acceptance questions
Input
Client brief, approved references, exclusions and presentation deadline
Owns
Extract required messages and visual constraints Identify unresolved brand or factual requirements Separate concept exploration from production-ready claims
Human approval
Founder scopes the creative decision and resolves ambiguity with the agency.
2Concept agentThree concept routes and candidate boards
Input
Approved constraint sheet and reference library
Owns
Propose three distinct directions Draft six-frame board options Generate bounded motion prompts linked to each direction
Human approval
Founder selects and edits the creative direction; this is a substantive craft step.
3Motion production agentDraft concept deck and motion studies
Input
Founder-selected boards and capped generation budget
Owns
Generate three 5-second motion studies with bounded retries Track which assets are exploratory Assemble pitch PDF and labeled previews
Human approval
Founder evaluates coherence, taste and client suitability; regenerate only within cap.
4Delivery checkerTechnical release checklist and delivery bundle
Input
Selected deck, files and constraints
Owns
Check lengths, export quality and required disclosure labels Package files and a revision checklist Track client feedback against the one-revision scope
Human approval
Founder makes the final creative and rights decision before release.
Your job as founder
Direct creative selection and make the presentation coherent.
Maintain agency relationships, negotiate scope and protect the retry budget.
Approve each concept and personally manage subjective revisions.
Required skill: The founder needs creative-direction and visual production judgment; delegating it to another human violates the one-founder premise.
The shift to agents: To become agent-first, narrow to one repeatable pitch format and a well-curated agency reference library. Under the current broad creative offer, agents assist a solo creative producer; they do not yet remove the craft bottleneck.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
The founder needs creative-direction and visual production judgment; delegating it to another human violates the one-founder premise.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $125.08 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 11 orders/week at 75 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $6,400 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $150. Run three permissioned briefs through one fixed deck format and log selection time and retries. Offer the $299 pack to 20 small agencies. Pass: At least 2 paid pilots and median founder handling at or below 75 minutes; two of first five buyers submit another brief within 30 days. Stop: If production remains near two hours or taste revisions are unbounded, retain it as a paid creative service rather than claiming an agent-led business.
First sale
Before launch: Show two credible examples Build permissioned sample packs and pre-book five agency conversations.
Day 1: Sell against tomorrow's pitch Offer one $299 fixed-scope pilot tied to a real brief. State the revision limit and concept-only output.
Next 48 hours: Deliver a decision pack Approve direction before generating motion studies. Present the chosen board and request the next brief.
Sources
Katalist pricing: Pro lists $39/month with story tools and image/video credits; Unlimited lists $99/month.
MiniMax video generation: H3 Max supports 5–15-second text, image and reference video generation at 480P/768P.
Model roles
GPT-6 Astra: Creative directions, constraints, shot list and client-ready copy.
MiniMax H3 Max: Short motion studies; selected frames also support the storyboard.
Muse Voice Transcribe: Optional transcription of the client's spoken creative brief.
Stop criteria
Fewer than 2 paid pilots from 20 qualified agencies, median production over two hours, or fewer than 2 repeat briefs among the first 5 clients within 30 days.
LocalDrop: A voice note becomes this week’s promotion.
A voice note and five real photos become three 15-second vertical promos with approved price, dates and address; English/Spanish caption versions and one revision included. Owner publishes the files. One-off purchase, no retainer, no filming visit.
Buyer: Independent restaurant or cafe owner with current phone photos and a time-limited offer. Start with one local bilingual district, not every business category.
Offer: $149 — Three 15s reels for one offer, EN/ES captions, one revision, next-day delivery. Owner posts; no filming required.
Cash budget: $5,200, excluding founder salary.
Thesis
Urgency and a nearby owner decision-maker make a small paid pilot more reachable than a cold consumer subscription. The product takes coordination and editing off the owner’s plate. It can repeat weekly only if owners actually reorder; views and restaurant revenue are not promised.
Start with cafes and restaurants in one bilingual neighborhood; sell a deadline-driven offer, not ongoing social management.
70/100B · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot or timing data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
A dated restaurant promotion creates a concrete next-day deadline; willingness to pay remains untested.
8/10
Standard inputs and bounded delivery15% weight · 4/5
One offer, five genuine photos and three fixed templates can repeat with a saved brand kit.
12/15
Agents perform the production work20% weight · 4/5
Agents can transcribe, check intake completeness, draft captions, populate templates and render approved assets.
16/20
Errors can be detected before delivery15% weight · 3/5
Dates and prices permit deterministic checks, but bilingual phrasing and food fidelity still need founder and owner approval.
9/15
Small, bounded founder workload15% weight · 3/5
A 25-minute fulfillment target is plausible only with complete intake and one consolidated correction; local selling remains founder work.
9/15
Room for labor, retries and acquisition10% weight · 3/5
A $149 ticket can support capped review, but missing assets, revisions and unmeasured sales time could absorb the margin.
6/10
A founder can reach buyers10% weight · 3/5
Nearby owners are identifiable and reachable; a trusted bilingual neighborhood network is an acquisition assumption.
6/10
Repeat work reuses approved assets5% weight · 4/5
Weekly offers create a recurring use case, but paid reorders rather than views must establish demand.
4/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Intake coordinatorComplete fact sheet with source links and unresolved fields.
Input
Owner voice note, five authorized photos and saved brand details.
Owns
Transcribe the brief and extract offer facts. Flag missing price, address, date or photo permissions. Prepare one consolidated clarification request for founder review.
Human approval
Founder confirms completeness; owner explicitly approves offer facts before production.
2Bilingual promotion editorThree source-grounded scripts and a bilingual caption sheet.
Input
Approved fact sheet, real photos and bilingual style guide.
Owns
Draft three short hooks and English/Spanish captions. Check every numerical claim against the fact sheet. Avoid inferred dietary claims or invented food imagery.
Human approval
Bilingual founder approves language, claims and creative choices.
3Template producerDownloadable preview pack and technical check report.
Input
Approved scripts, photos and three fixed motion templates.
Owns
Composite genuine food photos without changing the dish. Render three 15-second vertical files. Check duration, legibility, dimensions and caption completeness.
Human approval
Founder watches all exports; owner approves the pack and publishes it.
4Revision and delivery coordinatorFinal files and an order log.
Input
One consolidated owner correction and approved previews.
Owns
Map corrections to source facts and affected exports. Regenerate affected files and assemble downloads. Record elapsed handling time and reorder eligibility.
Human approval
Founder approves revised language and final release; out-of-scope work requires a new quote.
Your job as founder
Sell locally and qualify an actual promotion deadline.
Review English and Spanish copy and watch every final export.
Resolve ambiguous facts with the owner and approve exceptions.
Track all coordination, correction and delivery minutes against the 25-minute target.
Required skill: Founder must be competent in English/Spanish copy review and short-form visual editing. An agent cannot substitute for bilingual approval; an external reviewer would add explicit labor and staffing dependence.
The shift to agents: Replace bespoke editing and scattered messages with gated intake, a reusable brand kit, three deterministic templates and one correction form. Agents run drafting, rendering and packaging; the bilingual founder supplies the skilled review formerly described as editorial labor. The 25-minute target allocates 5 minutes to intake, 10 to bilingual/visual review, 5 to correction allowance and 5 to delivery handling; sales time sits in the separate fixed weekly allowance.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
Founder must be competent in English/Spanish copy review and short-form visual editing. An agent cannot substitute for bilingual approval; an external reviewer would add explicit labor and staffing dependence.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $74.93 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 34 orders/week at 25 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $5,200 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $150. Produce two permissioned samples, then offer the exact $149 pack to 15 qualified nearby owners. Run the first three paid orders through the full pipeline and log founder minutes. Pass: At least three paid orders, median fulfillment at or below 25 minutes including corrections, and at least one paid reorder within 45 days. Stop: Stop building after zero paid orders from 15 qualified conversations; narrow scope if median handling exceeds 40 minutes or owners will not provide complete approved facts.
First sale
Before launch: Earn permission for two samples Founder talks to 10 nearby owners during quiet hours and prepares two previews from authorized photos.
Launch day: Sell one current offer Show 5 owners with a real upcoming promotion; ask $149 upfront for next-day files. No guaranteed sale.
After delivery: Watch paid reorders Measure whether owners post the content and buy again. Track sales effort; views alone are not proof of value.
Sources
Dining Reels: Lists a 15s vertical social reel for €39, or €99/month for 4 social videos. EUR pricing; no currency conversion assumed.
Fork 2 Camera pricing: Lists $499/month for daily restaurant social posting, a three-month commitment and a one-time $199 setup fee. More extensive service than this proposed pack.
Model roles
GPT-6 Astra: Turns approved facts into hooks, bilingual scripts and shotlists; flags unsupported claims.
H3 Max: Creates short abstract background/title clips only. Real products or property features remain in unaltered source images.
Muse Voice Transcribe: Transcribes the owner’s brief. A separate TTS service voices approved text; Muse does not synthesize speech.
Stop criteria
No paid order after 15 qualified owner conversations, fulfillment over 60 minutes, or fewer than 3 of first 10 owners reordering within 45 days.
KinTape: Their stories. In both your languages.
Turn a relative’s voice notes into a bilingual keepsake for the birthday already on your calendar.
Buyer: An adult in a bilingual family buying a parent’s birthday gift so younger relatives can understand the original stories.
Offer: $79 — 15 source minutes, three audio chapters, a private page, PDF and one revision; 48-hour delivery.
Cash budget: $6,800, excluding founder salary.
Thesis
A meaningful finished gift has a clearer purchase trigger than an open-ended memory app. Original recordings plus reviewed translation can bridge generations, but digital value must compete with printed books.
Preserve the source audio, verify names and dates, and deliver a ready-to-share artifact. Narration is labeled translated audio; missing details stay missing.
51/100C · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot or timing data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 3/5
A birthday supplies a deadline, but a private digital keepsake is discretionary and competes with printed gifts.
6/10
Standard inputs and bounded delivery15% weight · 3/5
The 15-minute, three-chapter format repeats; each family's names, dialect and sensitive memories differ.
9/15
Agents perform the production work20% weight · 3/5
Transcription, chapter drafts, labeled narration and page/PDF assembly can be orchestrated from approved text.
12/20
Errors can be detected before delivery15% weight · 2/5
Emotional meaning and family-specific language resist automated checking; source audio and family confirmation remain essential.
6/15
Small, bounded founder workload15% weight · 3/5
A 30-minute review target retains the entire existing review allowance; substantial skilled attention cannot responsibly be removed.
9/15
Room for labor, retries and acquisition10% weight · 2/5
The $79 price leaves limited room for lengthy corrections, support and specialist review despite cheap model calls.
4/10
A founder can reach buyers10% weight · 2/5
An opted-in bilingual family network and community partners are prerequisites rather than established channels.
4/10
Repeat work reuses approved assets5% weight · 1/5
Occasion-driven gifts may generate referrals, but repeat purchases by one buyer are naturally infrequent.
1/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Consent and source organizerTimestamped transcript, consent checklist and clarification queue.
Input
Up to 15 minutes of clean voice notes, sharing preferences and supplied names.
Owns
Check required permission and sharing fields. Transcribe audio with timestamps. Flag uncertain names, dates and unclear segments without filling gaps.
Human approval
Founder verifies permission records; storyteller or buyer confirms uncertain personal details.
2Memory chapter editorSide-by-side source and translated chapter draft.
Input
Confirmed transcript and a three-chapter structure.
Owns
Group existing memories into three short chapters. Translate English/Spanish while linking sentences to the source. Flag idioms and emotionally ambiguous wording.
Human approval
Bilingual founder listens to the source and approves every translated chapter; family resolves personal meaning.
3Keepsake producerPrivate preview page, three audio chapters and PDF.
Input
Approved chapter text, original audio and authorized stock voice.
Owns
Generate translated narration and label it clearly. Build a private page and PDF from fixed templates. Check chapter links, narration/text consistency and access settings.
Human approval
Founder verifies access settings and samples narration against approved text; buyer approves sharing.
4Correction and retention coordinatorCorrected package and retention record.
Input
One consolidated family correction, approved delivery and retention choice.
Owns
Identify all outputs affected by an accepted correction. Rebuild approved text and audio versions. Prepare deletion and access-expiry actions according to the selected retention policy.
Human approval
Founder approves semantic changes and confirms requested deletion was completed.
Your job as founder
Establish permission and private-sharing expectations.
Listen to the source and review all translated prose as a competent bilingual editor.
Handle emotionally sensitive or ambiguous memories with the family.
Approve final access settings, corrections and retention handling.
Required skill: A founder fluent in both source and target languages, with editorial sensitivity and basic privacy operations, is required. Family review confirms personal facts but does not replace competent bilingual translation review.
The shift to agents: Move coordination, transcription, source alignment and artifact assembly into a staged pipeline with explicit approvals. Retain 30 founder minutes per order: 15 for source listening, 10 for translation/chapter checking and 5 for handling or corrections. This is an agent-operated production workflow, but the baseline skilled review does not disappear; complex dialects or emotional corrections require repricing or refusal.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
A founder fluent in both source and target languages, with editorial sensitivity and basic privacy operations, is required. Family review confirms personal facts but does not replace competent bilingual translation review.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $36.46 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 28 orders/week at 30 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $6,800 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $120. Make one permissioned sample and show it to 20 opted-in bilingual families with a birthday within 60 days; offer three $79 pilot slots and time complete fulfillment. Pass: At least three full-price orders, median total handling at or below 30 minutes, and all three buyers approve factual and emotional accuracy before sharing. Stop: Stop the current offer if fewer than three pay after 40 qualified sample views, or if review repeatedly exceeds 45 minutes; pause delivery when a translation changes the meaning of a memory.
First sale
Before launch: Find the occasion Recruit 25 opted-in prospects through a bilingual family network and two community partners. Show one completed, permissioned sample.
First selling day: Sell five gift slots Offer $79 gifts with 48-hour delivery. One buyer from 25 qualified invitees is a test assumption, not assured demand.
After delivery: Check meaning and gift value Measure corrections and reviewer minutes. Ask whether the buyer shared the gift and would buy another for a different relative.
Storyworth book pricing: Lists $69 Basic, $99 Color and $199 Unlimited plans, including books.
Model roles
GPT-6 Astra: Organize approved memories into chapters and translate without adding biography.
Muse Voice Transcribe: Transcribe the original audio. Keep it alongside the translation; use a separate licensed narration service.
MiniMax H3 Max: Produce one six-second atmospheric illustration, labeled as inspired by the memory rather than historical footage.
Stop criteria
After 40 qualified sample views and 15 interviews, rework if fewer than 3 pay at least $59. Reduce scope if median review exceeds 45 minutes or corrections exceed 15% of orders.
Interview Bridge: Your experience. Clearly expressed.
Turn the experience you explain in your first language into truthful English answers for your next interview.
Buyer: A Spanish-speaking applicant with an English-language customer-success interview in three to seven days.
Offer: $49 — Five checked story cards, three 20-minute rehearsals and a seven-day preparation pack.
Cash budget: $5,800, excluding founder salary.
Thesis
The calendar invite creates urgency. A role-specific, reviewed pack may earn a one-time purchase; generic mock interviews already have strong free and paid alternatives.
A bilingual reviewer checks five answers against the applicant’s fact sheet. Native-language explanations and follow-ups expose missing evidence instead of inventing accomplishments.
66/100B · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot or timing data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
An interview booked in three to seven days creates a strong preparation deadline.
8/10
Standard inputs and bounded delivery15% weight · 4/5
One job family, five story cards and three fixed-length rehearsals form a narrow repeatable workflow.
12/15
Agents perform the production work20% weight · 4/5
Agents can structure supplied facts, translate answers and run timed rehearsals while escalating unsupported claims.
16/20
Errors can be detected before delivery15% weight · 3/5
Fact-to-answer comparisons are inspectable, but truthful nuance and useful bilingual coaching need founder judgment.
9/15
Small, bounded founder workload15% weight · 4/5
The existing 15-minute review allowance is already tight; agent orchestration should protect it, not assume it vanishes.
12/15
Room for labor, retries and acquisition10% weight · 2/5
A $49 ticket is sensitive to support overruns and competes with inexpensive or free practice tools.
4/10
A founder can reach buyers10% weight · 2/5
A trusted career-coach referral channel is proposed but has not been demonstrated.
4/10
Repeat work reuses approved assets5% weight · 1/5
Most candidates need a short preparation burst; successful customers may quickly leave the category.
1/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Evidence intake editorApplicant-confirmed evidence sheet with source references.
Input
Applicant's Spanish explanation, fact sheet and customer-success job description.
Owns
Extract only supplied responsibilities and outcomes. Map each potential story to a supporting statement. Queue questions for missing examples or unmeasured outcomes.
Human approval
Applicant confirms the facts; founder rejects unsupported qualifications or metrics.
2Bilingual answer editorFive draft cards with claim-to-source links.
Input
Confirmed evidence sheet and five standard story-card templates.
Owns
Draft concise English answers and Spanish explanations. Preserve uncertainty and qualitative outcomes. Check every accomplishment against the evidence sheet.
Human approval
Bilingual founder with relevant coaching competence approves all five cards before rehearsal.
3Rehearsal facilitatorPractice transcripts and concise practice priorities.
Input
Approved cards and a fixed role-specific practice rubric.
Owns
Run three capped 20-minute automated practice sessions. Ask follow-ups tied to the applicant's existing facts. Produce editable transcripts and flag answers that introduce new claims.
Human approval
Founder reviews flagged factual conflicts; new accomplishments require applicant confirmation before inclusion.
4Pack and support coordinatorFinal pack and exception log.
Input
Approved cards, rehearsal records and one support queue.
Owns
Assemble the seven-day preparation pack. Route factual or translation questions to the founder. Track handling minutes and prepare final downloads.
Human approval
Founder resolves substantive questions and approves final revisions.
Your job as founder
Obtain opt-in coach referrals and qualify an actual interview date.
Review five bilingual cards against the applicant's confirmed evidence.
Handle factual exceptions and substantive coaching questions.
Keep practice preparation separate from assistance during an actual interview.
Required skill: Founder must review English/Spanish accurately and have credible interview-coaching competence for the selected job family. Agents cannot supply missing language judgment or verify an applicant's real-world history independently.
The shift to agents: Agents coordinate intake, evidence-linked drafts, automated rehearsals and delivery. The founder retains 15 minutes per order: about 10 for five-card factual/language review and 5 for exceptions or support. The three rehearsals are automated sessions; adding human rehearsal coaching would require a different price and time model. No outsourced reviewer is silently assumed.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
Founder must review English/Spanish accurately and have credible interview-coaching competence for the selected job family. Agents cannot supply missing language judgment or verify an applicant's real-world history independently.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $23.83 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 57 orders/week at 15 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $5,800 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $100. Use one consenting career-coach partner to present a sample to 20 qualified applicants with dated interviews. Sell the exact $49 offer and time five-card review plus all support. Pass: At least three paid buyers, median total founder handling at or below 15 minutes, and zero unsupported accomplishments in approved cards. Stop: Rework after fewer than three purchases among 30 qualified applicants or median handling above 20 minutes; pause immediately if fabricated qualifications reach a buyer.
First sale
Before launch: Find dated interviews Recruit one bilingual career coach to share an opt-in preview with 20 applicants who already have interviews.
First selling day: Sell a scoped preparation pack Show a useful story-card sample and offer five $49 slots. One sale from 20 viewers is a hypothesis; the channel must exist first.
After delivery: Audit truth and handling time Check every answer against supplied facts. Track reviewer minutes and ask which cards the applicant actually used.
Sources
Yoodli individual pricing: Pro is $8/month and Advanced $20/month, both billed annually; a five-session free tier also exists.
GPT-6 Astra: Translate and structure evidence, then generate follow-ups grounded in the applicant’s real facts.
Muse Voice Transcribe: Create editable rehearsal transcripts. Spoken prompts need a separately selected speech-output service.
Stop criteria
Rework if fewer than 3 of 30 qualified applicants pay $49 or handling exceeds 20 minutes. Pause immediately if fabricated qualifications reach a buyer.
CourseDub: Your next market starts with the lesson you already made.
Sell a small, human-checked Spanish localization pack to course creators. Preserve teaching intent and terminology, then deliver upload-ready files.
Buyer: A paid-course creator receiving Spanish-language requests before an upcoming promotion.
Offer: $149 — Three clips, 10 source minutes total → one Spanish version, captions, glossary and one correction round.
Cash budget: $5,300, excluding founder salary.
Thesis
Creators can buy cheap dubbing software. Some will pay more to receive a finished lesson with native-language review and consistent terminology. Sell a bounded audience experiment before promising a full multilingual course.
A glossary and bilingual approval distinguish this from clicking Translate. Preserve the source visuals; offer one language pair and predictable delivery. This service fails if the creator prefers editing themselves.
64/100C · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot or timing data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 3/5
An upcoming creator promotion can supply urgency when Spanish audience requests already exist.
6/10
Standard inputs and bounded delivery15% weight · 4/5
Three clips, one language pair, original visuals and a saved glossary constrain production well.
12/15
Agents perform the production work20% weight · 4/5
Agents can transcribe, apply a glossary, generate approved narration, align captions and package exports.
16/20
Errors can be detected before delivery15% weight · 3/5
Source alignment and glossary checks help, but teaching intent and specialized language still require skilled review.
9/15
Small, bounded founder workload15% weight · 3/5
A 45-minute target retains substantial bilingual and production approval; complex lessons can exceed it quickly.
9/15
Room for labor, retries and acquisition10% weight · 2/5
The $149 ticket is margin-sensitive once skilled founder labor, acquisition and corrections are counted.
4/10
A founder can reach buyers10% weight · 2/5
Creators with visible language requests can be identified, but warm access and purchase intent are unproven.
4/10
Repeat work reuses approved assets5% weight · 4/5
A successful pilot could expand across lessons and releases using a reusable creator glossary.
4/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Lesson intake and terminology analystSource transcript, glossary and acceptance checklist.
Input
Three authorized clips totaling at most 10 minutes and creator terminology preferences.
Owns
Transcribe and timestamp the source. Flag text-heavy slides, specialized terms and ambiguous teaching instructions. Build a proposed glossary and scope exceptions.
Human approval
Bilingual founder accepts only familiar subject matter; creator approves preferred terminology and scope.
2Localization editorAligned Spanish script and unresolved issues.
Input
Approved glossary and timestamped source transcript.
Owns
Translate into Spanish with source-linked segments. Check numerical values, examples and glossary consistency. Flag expansions that prevent natural timing.
Human approval
Bilingual founder checks the entire translated script and teaching intent; creator resolves domain ambiguities.
3Narration and caption producerSpanish video previews, captions and glossary files.
Input
Approved Spanish script, original visuals and authorized stock voice.
Owns
Generate narration without voice cloning. Align narration and captions to the original visuals. Render the three clips and run technical export checks.
Human approval
Founder listens and watches final exports for meaning, pronunciation and synchronization.
4Revision and release coordinatorFinal localization package and reusable approved glossary.
Input
One consolidated correction set and reviewed exports.
Owns
Identify affected segments and regenerate them. Check consistency across captions, audio and glossary. Package upload-ready files and log elapsed labor.
Human approval
Founder approves all changed segments and final release; broader edits are quoted separately.
Your job as founder
Qualify creators with actual Spanish audience interest and constrain subject matter.
Approve translation and terminology as a skilled English/Spanish editor.
Listen to final audio and verify visual timing and captions.
Resolve creator corrections, price exceptions and track total delivery time.
Required skill: Founder must be a competent English/Spanish localization editor with enough subject knowledge to preserve teaching intent and enough production skill to inspect timing. Unknown terminology needs creator clarification or separately costed expert review; agents cannot replace those skills.
The shift to agents: Replace manual file shuffling, first-draft translation, narration assembly and routine caption timing with an approval-driven pipeline. In a true solo operation the founder performs the bilingual review previously assigned to a recruited reviewer. Target 45 minutes: 20 for source/translation comparison, 10 for final playback, 10 for correction allowance and 5 for intake/delivery. Reject text-heavy slides and unfamiliar domains instead of inventing expertise or hiding reviewer staffing.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
Founder must be a competent English/Spanish localization editor with enough subject knowledge to preserve teaching intent and enough production skill to inspect timing. Unknown terminology needs creator clarification or separately costed expert review; agents cannot replace those skills.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $41.93 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 19 orders/week at 45 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $5,300 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $180. Create a permissioned two-minute sample in a domain the bilingual founder understands. Offer $149 batches to 20 creators with visible Spanish audience requests and time the first three paid batches. Pass: At least two paid pilots, median total handling at or below 45 minutes, zero meaning-changing errors in final review and one repeat paid batch among the first five buyers. Stop: Stop the fixed offer after fewer than two paid pilots from 20 qualified creators; reprice or narrow scope if handling exceeds 60 minutes or creator corrections repeatedly require specialist input.
First sale
Before launch: Choose one language pair Create a permissioned two-minute demo and recruit one dependable bilingual reviewer.
Day 1: Sell a small localization test Contact warm creators or 20 qualified prospects with visible language requests. Sell one $149 batch against a real release.
Next 48 hours: Deliver an approved package Approve the glossary, review the audio and captions, and record delivery time before quoting the next batch.
Sources
HeyGen pricing: Creator is $29/month with 600 credits. Pro starts at $49/month and includes translation script editing.
HeyGen translation guide: Video translation is an existing workflow; on-screen text translation is a separate task.
Model roles
Muse Voice Transcribe: Turns the source recording into text.
GPT-6 Astra: Translates teaching intent and flags ambiguous terminology.
MiniMax Speech: Reads the approved translation using an authorized stock voice.
Stop criteria
Fewer than 2 paid pilots from 20 qualified creators, more than 75 combined review/production minutes per batch, or no repeat batches among the first five clients.
Speaking Sprint: Seven days. One priority at a time.
A focused IELTS speaking plan that explains recurring mistakes in your first language and gives you a better second attempt.
Buyer: A Spanish-speaking candidate with an IELTS test booked in seven to fourteen days, already near their target level.
Offer: $29 — Three 15-minute mocks, short drills, a seven-day plan and one teacher feedback review.
Cash budget: $6,400, excluding founder salary.
Thesis
A finite preparation purchase can sell on prioritization and teacher calibration. It cannot credibly promise a new proficiency level in a week or compete through generic simulated interviews alone.
A qualified teacher checks the final practice priorities, explained clearly in Spanish. Track better attempts and completed drills instead of displaying an unsupported precise band prediction.
59/100C · Agent-first fit
Validate the open gates first
Editorial baseline · original price and cost assumptions. Evidence confidence: Low.
Planning assumptions; no paid pilot or timing data supplied.
Unmeasured targets. Capacity = (30 hours − 12 for sales, admin and maintenance) × 60 × 80% buffer ÷ minutes per order, rounded down. It is a delivery ceiling, not a sales forecast. Fulfillment labor is charged separately in the pitch’s economics. The editorial fit score stays fixed when you explore a different price or acquisition cost; the economics gate follows your scenario.
See all 8 ratings and reasons
A reachable buyer with a deadline10% weight · 4/5
A test booked within two weeks creates a clear deadline for targeted practice.
8/10
Standard inputs and bounded delivery15% weight · 4/5
One exam, one feedback language and three capped mocks create a bounded delivery sequence.
12/15
Agents perform the production work20% weight · 3/5
Agents can run practice, transcribe attempts, organize examples and assemble drills, while teacher judgment governs priorities.
12/20
Errors can be detected before delivery15% weight · 2/5
Text alone cannot assess pronunciation, and plausible feedback can be pedagogically wrong; audio review and calibration are essential.
6/15
Small, bounded founder workload15% weight · 4/5
A qualified teacher must review each plan; a 12-minute target is already slightly above the current 10-minute cost assumption.
12/15
Room for labor, retries and acquisition10% weight · 2/5
The $29 ticket has little tolerance for support, paid referrals or longer teacher review.
4/10
A founder can reach buyers10% weight · 2/5
Trusted tutor referrals could work, but the offer depends on access to teachers and students that has not been shown.
4/10
Repeat work reuses approved assets5% weight · 1/5
A seven-day pre-exam sprint is naturally finite; repeat exams are not a dependable retention strategy.
1/5
The agent team
A proposed staffing blueprint. Validate each job and approval step before deployment.
1Practice session facilitatorRecordings, editable transcripts and session completion records.
Input
Confirmed test date, consented recordings and teacher-approved original prompts.
Owns
Run three capped 15-minute mock sessions. Record and transcribe attempts with timestamps. Flag unclear audio and transcription uncertainty.
Human approval
Teacher founder approves the prompt set and excludes recordings too poor to support useful feedback.
2Evidence and priority analystDraft priority sheet with audio evidence and uncertainty flags.
Input
Practice audio/transcripts and a teacher-calibrated feedback rubric.
Owns
Find repeated issues with linked audio excerpts. Compare attempts without presenting an official band score. Draft a small number of practice priorities and Spanish explanations.
Human approval
Qualified teacher founder listens to supporting audio and accepts, revises or rejects each priority.
3Drill and plan producerPersonalized practice plan and short drills.
Input
Teacher-approved priorities and reviewed exercise templates.
Owns
Select short drills matched to the approved issue. Assemble a seven-day plan with a second-attempt prompt. Check consistency between the explanation, example and exercise.
Human approval
Teacher founder approves pedagogical usefulness and Spanish explanation before release.
4Progress and exception coordinatorAttempt comparison, export and unresolved question queue.
Input
Completed drills, second attempts and learner questions.
Owns
Track practice completion and organize attempt comparisons. Route substantive learning questions or suspect advice to the teacher. Package the final plan and record handling time.
Human approval
Teacher founder handles misleading feedback and any advice that needs professional judgment.
Your job as founder
Be the qualified teacher responsible for calibration and each learner's approved priorities.
Listen to relevant audio; never infer pronunciation quality from transcript text alone.
Review Spanish explanations and handle substantive learner questions.
Maintain tutor referral relationships and monitor advice correction rates.
Required skill: A qualified IELTS speaking teacher able to review English audio and explain feedback accurately in Spanish is required for the solo-founder version. An agent cannot become that teacher; otherwise a named, costed human teacher is required and the operation is no longer solo fulfillment.
The shift to agents: Agents run the practice sequence, gather audio evidence and assemble drills, while a qualified bilingual teacher founder owns the final teaching judgment. Do not describe independent teachers as invisible fulfillment staff. Budget 12 minutes per order: 6 for focused audio review, 4 for priority/plan approval and 2 for routine handling. This exceeds the existing 10-minute labor estimate; ambiguous speech or greater support needs must be declined, scoped separately or repriced.
Readiness gates · 4 unknown · 0 blocked
A high score cannot override a failed gate. “Pass” can mean planning arithmetic only; read each reason.
Unproven
Required founder expertise
A qualified IELTS speaking teacher able to review English audio and explain feedback accurately in Spanish is required for the solo-founder version. An agent cannot become that teacher; otherwise a named, costed human teacher is required and the operation is no longer solo fulfillment.
Unproven
Input rights and factual approval
Consent, image/audio rights and customer fact approval must be verified for each order; no actual order evidence was supplied.
Unproven
Acceptance and failure path
The agent workflow defines human release gates, but retry limits, source checks and correction/refund handling still need a timed end-to-end pilot.
Plan passes
Positive modeled contribution
Planning arithmetic: $12.71 per order after the stated fulfillment labor, models, delivery, cash acquisition, 2.9% + $0.30 fee and 5% refund reserve. All costs and demand are hypotheses; sales/build time and fixed overhead are separate.
Unproven
Founder capacity measured
Planning ceiling: 72 orders/week at 12 founder minutes/order, 30 founder hours, 12 fixed hours and a 20% delivery buffer. Target time is unmeasured; a pilot must prove it before making volume promises.
Plan passes
Startup budget below $10,000
Current itemized plan totals $6,400 for this one business, excluding founder salary and living costs. Spending is not evidence of demand.
Earn the next score with evidence.
Budget up to $150. Have two independent tutors assess a small sample of founder-approved plans as a paid validation expense, then offer the $29 sprint to 20 qualified learners through opt-in referrals. Log audio review, plan approval and all support time. Pass: At least three paid learners, median total teacher handling at or below 12 minutes, at least two complete a second attempt, and independent tutors accept at least 90% of proposed priorities. Stop: Stop delivery if advice is materially misleading; rework after fewer than three paid buyers among 30 qualified learners, median handling above 12 minutes or tutor rejection above 10%.
First sale
Before launch: Calibrate with teachers Recruit two independent IELTS tutors. Have them review the diagnostic and recommend it only for identified between-lesson practice needs.
First selling day: Offer ten pilot places Sell the seven-day plan for $29. One buyer among 20 qualified students is a planning scenario, not observed conversion.
After delivery: Measure useful repetitions Track second attempts, teacher corrections and handling time. Ask whether the priority was clear and usable.
Sources
E2 Speaking Intensive: Purchase page displayed $99 list / $74.25 promotional pricing for teacher-led speaking preparation; promotions can change.
SmallTalk2Me simulator: Offers AI speaking simulation and a free version, making a generic simulator difficult to differentiate.
Model roles
GPT-6 Astra: Find recurring issues, explain them in Spanish and produce teacher-reviewed practice plans.
Muse Voice Transcribe: Transcribe practice attempts for editable comparisons. Transcripts alone cannot establish pronunciation quality.
Stop criteria
Rework after fewer than 3 of 30 qualified candidates pay, handling above 12 minutes or teachers reject more than 10% of priority advice. Calibrate before adding score estimates.
A business one person can actually run.
Score each dimension from 0–5, multiply by its weight, and add the results. Agent execution, repeatable delivery, quality checks and founder workload account for 65 of 100 points. These are judgment calls to test, not measured probabilities.
A · 80–100 Prioritize a pilotB · 65–79 Test narrow scopeC · 50–64 Redesign bottlenecksD · 0–49 Weak solo fit
Evidence is separate. Low = desk research or an unverified brief. Medium = observed delivery and buyer interviews. High = repeated paid pilots with measured labor, acquisition and retention. All original ten ideas remain Low confidence.
A reachable buyer with a deadline10 points
Who can approve payment, for what concrete outcome, by when?
No identifiable buyer or useful outcome.
Broad audience; no purchasing trigger.
Named segment, but problem is optional or payer unclear.
Identifiable buyer with a recurring inconvenience and a plausible budget.
Specific decision-maker, consequential deadline and bounded paid outcome.
Buyer already has a budgeted recurring workflow and an obvious deadline-linked purchase.
Standard inputs and bounded delivery15 points
Can the same intake, templates and acceptance checks handle each order?
Every project is open-ended bespoke work.
Scope and input requirements vary almost completely.
Some templates, but frequent incomplete inputs or branching requirements.
A narrow deliverable with reusable steps and a defined revision limit.
Structured inputs and reusable production with limited exceptions.
A fixed input contract, deterministic assembly and explicit exception/rejection rules.
Agents perform the production work20 points
What share of operational steps can agents and software complete without a founder doing the craft?
The founder must produce nearly the entire output.
Agents only brainstorm or draft notes; founder produces the deliverable.
Agents create drafts; founder still edits or assembles most outputs.
Agents perform most routine steps; founder handles a substantial finishing stage.
Agents run intake through assembled deliverable; founder approves and resolves exceptions.
Agents run bounded end-to-end jobs with retries, logs and clear handoffs; founder handles rare exceptions and release approval.
Errors can be detected before delivery15 points
Can source checks and explicit acceptance rules catch the expensive mistakes?
Output cannot be checked, or harmful mistakes may be invisible.
Quality is almost entirely subjective or needs unbounded expert judgment.
A skilled founder must inspect most details; automated checks miss important errors.
Structured facts, checks and a short specialist release review cover the main failure modes.
Reliable automated acceptance checks cover correctness; short human review covers a narrowly defined residual risk.
Small, bounded founder workload15 points
Include review, revisions, exception handling and delivery support per paid order.
More than 180 expected founder minutes per order, or no credible bound.
91–180 expected founder minutes per order.
46–90 expected founder minutes per order.
21–45 expected founder minutes per order.
11–20 expected founder minutes per order.
At most 10 expected founder minutes per order.
Room for labor, retries and acquisition10 points
Does the scoped price cover labor, models, delivery, fees, refunds and cash acquisition?
Nonpositive contribution with all stated costs included.
Positive only when founder labor or a necessary cost is omitted.
Thin contribution or unusually sensitive to one revision/retry.
Positive contribution under the stated base case, with limited stress headroom.
Positive contribution with meaningful room for one bounded exception.
Strong contribution remains after a plausible high-cost stress case; price and scope are easy to enforce.
A founder can reach buyers10 points
Can one founder directly reach a narrow segment through a feasible channel?
No identifiable acquisition route.
Requires mass paid reach, enterprise sales or a large audience the founder lacks.
Plausible broad channel, but trust/access or acquisition cost is a substantial obstacle.
Specific direct outreach or referral route with a bounded weekly sales workload.
Concentrated buyer network with a clear trigger and reachable decision-makers.
Founder already controls a relevant, permissioned channel with repeat purchasing opportunities.
Repeat work reuses approved assets5 points
Can subsequent orders reuse the same buyer, approved knowledge and production setup?
No reuse and almost no repeat use.
One-off purchase; limited referral value.
Mostly one-off purchase, but a credible referral or adjacent repeat route.
Some repeat occasions and reusable customer context.
Frequent plausible repeat orders that reuse an approved brand or knowledge base.
A naturally recurring operational workflow where maintained approved context materially lowers later effort.
Gates override enthusiasm: verify input rights, founder expertise, release checks, positive economics, timed founder capacity and a startup cash budget of $10,000 or less. Unknown means validate first. A blocked gate means do not sell that scope yet.
Rubric version: solo-agent-v1 · September 22, 2026. Changing a weight requires a new rubric version and rescoring every idea.
Model research
GPT-6 Astra · reasoning and translation Text and image input; text output. It plans, adapts, translates and checks content. It does not itself provide audio output. Standard short-context API rates: $10 per million input tokens and $50 per million output tokens; include reasoning output in the bill. Official OpenAI model documentation
MiniMax H3 Max · short video generation The joint MiniMax/fal model generates short shots, generally 5–15 seconds, that can be assembled into a larger piece. We budget the full 768p list rate of $0.08 per generated second. Three takes for 30 usable seconds therefore cost $7.20 before other services. Temporary promotions are not built into the business case. MiniMax published model prices
Muse Voice Transcribe · speech to text The closest verified match to “Muse voice translate” is Meta Muse Voice Transcribe. It transcribes speech; Astra or another translation model must translate the transcript. Rate: $0.003 per input audio minute. Muse turn-level timing is not word-level caption alignment. Meta speech-to-text documentation
Muse pricing The published transcription rate is $0.18 per hour. Ten input minutes cost $0.03 before translation, narration, storage or human review. Meta pricing and rate limits
Speech output · a separate stage MiniMax Speech 2.8 Turbo is budgeted at $60 per million characters. Use an authorized stock voice for the first product. At an assumed 900 characters per spoken minute, ten output minutes cost $0.54. This site’s play button uses browser speech, not MiniMax. MiniMax speech pricing
Production access still needs a real pilot Public documentation establishes the model offering, not your account entitlement, rate limits, latency or output quality. Test one language pair and one deliverable with API billing and usage caps before taking production orders. H3 shots may need review and rerendering; accurate property footage and procedural screens should remain source footage. H3 Max endpoint and schema