Lean Canvas¶

Phase: 4 — Strategy (Fast Track: Lean Canvas only) Project: likeness Date: 2026-05-09 Confidence: Medium-High where research-grounded; Low on customer-side numbers (no primary discovery)

1. Problem (top 3)¶

For adult creators (primary customer):

#	Problem	Existing alternatives
1	Unauthorized AI use of creator likeness with no consent, no revenue, no recourse	DMCA takedowns (slow, expensive, Tier 3 effective); platform reports (band-aid); doing nothing
2	Revenue ceiling on real content; no easy way to monetize an additional AI-likeness layer	Bootstrapped AI tools (technical lift); not doing it (most common)
3	Platform/payment risk: rule changes (Mastercard/Visa/Fansly AI ban) or account bans can wipe out earnings	Cross-posting to multiple platforms; gray-market processors (high risk)

For adult creator subscribers (secondary customer):

#	Problem	Existing alternatives
1	Synthetic AI girlfriend platforms (Candy.ai, SoulGen, etc.) feel hollow vs. real-creator engagement	Pay synthetic platforms anyway; settle for real-content-only on creator platforms
2	Want more content from a specific creator than the creator is producing manually	PPV unlocks; custom requests (creator time-bottlenecked)

2. Customer Segments¶

Primary: Verified mid-tier adult creators with existing audiences¶

US-based at MVP
Top 1-10% of OnlyFans creators (monthly gross $5K-$80K range)
Active fan audience 1K-50K
Comfortable with multi-platform stack but skeptical of new platforms
Early adopters: creators who have been deepfaked without consent and are actively frustrated by the lack of recourse — for them, Likeness solves a current pain, not a hypothetical opportunity

Secondary: Engaged adult creator subscribers¶

Existing OnlyFans / Fansly subscribers paying $20-$200/month spread across multiple creators
Some willingness to pay for "extras" (PPV, customs, AI)
Parasocial engagement with specific creators is a real driver
Early adopters: existing top-spend fans of concierge-cohort creators

3. Unique Value Proposition¶

For verified adult creators who are tired of being deepfaked without consent or compensation: Likeness is the consent-first AI likeness platform where the model never leaves our servers, every output ties to a license you control, and you can revoke at any time.

Short form (creator-facing):

Your likeness. Your rules. Revocable.

Short form (investor-facing):

Consent infrastructure for AI-generated adult content, structurally aligned with where regulation is going.

4. Solution (top 3 features)¶

#	Feature	Maps to problem
1	License-gated AI generation — every prompt parses against the creator's structured license object before any model loads; outputs carry watermark + signed metadata + license ID	Problems 1 (consent) + 2 (revenue line) for creators; Problem 1 (real-creator-engagement) for fans
2	No model export, ever — architectural commitment, not just policy. Per-creator LoRA adapters are encrypted, isolated, never leave the platform	Problem 1 (consent + control) for creators
3	Revocable license + creator approval queue — creator can revoke entirely or per-fan; submission queue lets creator approve or reject fan generations and earn submission fees	Problems 2 (additive revenue) + 3 (platform risk = leaves with you) for creators; Problem 2 (more from your favorite creator) for fans

Plus, supporting infrastructure that's load-bearing: - Multi-processor payment redundancy (Civitai precedent) - Identity verification + 2257 records system - C2PA-compliant provenance from day 1 - Takedown / detection component bundled into concierge phase (per Variation B absorption from 00-intake/brainstorm.md)

5. Channels¶

Inbound¶

Press / earned media off the consent-first regulatory-tailwind angle. Story for press: "the legitimate platform regulators are catching up to."
Industry trade press (XBIZ, AVN online) for creator-side awareness.
C2PA membership announcement as positioning + credibility multiplier (low cost, high signal).

Outbound (primary)¶

Founder + Creator Ops cofounder 1:1 outreach via insider network introductions. The primary creator-acquisition channel for MVP.
Adult industry events (XBIZ, AVN, EXXXOTICA) for relationship-building.
Existing Discord / Telegram creator communities (with the right Creator Ops profile to access).

Year 2+ channels¶

Talent agency partnerships
Press / external advocacy from creator-rights organizations
Eventually, fan-side SEO (long-tail "[creator name] AI" once concierge cohort exists)

6. Revenue Streams¶

Per founder brief — confirmed by research that 59% of comparable creator revenue comes from one-off transactions:

Stream	Mechanism	Likeness take	Notes
Subscription tiers ($15-$200/mo illustrative)	Standard creator subscription	~20% (industry standard)	Floor revenue; not the largest line
AI generation compute credits	Fan pays compute cost + creator-set markup	Platform compute cost-pass-through + small platform margin	The new revenue line for creators; growth driver
PPV unlocks of approved generations	One-off; creator sets price	~20%	Aligns with 59% one-off behavior in OnlyFans
Submission fees (approval queue)	Fan pays for review by creator	~20%	Creator-controlled scarcity
Custom requests / messaging	Standard creator practice	~20%	Carry-over from existing creator economy
Premium creator tools (analytics, takedown monitoring)	Creator-paid add-ons or % of revenue share	TBD	Year 2+

LTV: Highly variable by creator size. [Estimate] $2,000-$20,000 per creator-year platform revenue at maturity, weighted heavily by top creators.

Gross margin: Adult-friendly processor fees of 5-15% compress margin vs. mainstream SaaS. Realistic platform-level gross margin: 45-65% depending on inference volume and processor mix.

7. Cost Structure¶

Per founder budget at docs/budget.md — confirmed by research as plausible:

Personnel: $940K of $1.5M (~63%) — six-seat founding team at cofounder-modest cash
Legal & professional services: $130K (~9%) — including AI/likeness counsel, processor-friendly counsel, insurance
Infrastructure & tools: $160K (~11%) — compute, cloud, identity verification, dev tooling
Compliance & T&S infrastructure: $35K (~2%) — 2257 system, classifier tooling, watermarking, audit
Creator acquisition: $40K (~3%) — events, travel, signing incentives, no paid media
Office / remote: $25K (~2%) — quarterly retreats
Reserve: $170K (~11%) — for processor / legal surprises

CAC at MVP: Effectively the time-cost of founder + Creator Ops cofounder 1:1 outreach. No paid acquisition. Per-creator CAC at concierge phase: [Estimate] mostly time, with $1,500 (avg) signing incentives + travel.

Variable costs: Platform-side inference compute is passed through to fans as credits with markup. Identity verification fees per onboarding are real.

8. Key Metrics (AARRR)¶

Stage	Metric	Target (Year 1 / concierge)
Acquisition	Concierge creator signups	5-10 invite-only creators
Activation	Creator's first AI generation purchase by a fan	≥80% of concierge cohort
Retention	Concierge creator continuing participation at month 3, month 6	70%+ at month 3, 50%+ at month 6
Revenue	Per-creator monthly platform revenue	$400-$2,000/mo (concierge avg)
Referral	Concierge creator → introduces another creator within 6 months	≥40%

Plus — load-bearing operational metrics: - Processor uptime / chargeback rate — chargebacks <1% (Visa VIRP), processor-redundancy operational - License-rule violation attempts blocked vs. allowed — high block rate signals working enforcement; high allow rate signals over-permissive license framework - Take-down request response time — under 48 hours always (TAKE IT DOWN Act floor)

9. Unfair Advantage¶

This is a real question. The architecture is buildable by any sufficiently funded competitor. The honest list:

Architectural commitments treated as non-negotiable. No model export, license-gated inference, per-creator isolation are easy to compromise on under commercial pressure. A team committed to NOT compromising — even when competitors are — is itself a moat. Real but soft.
Compliance posture as positioning. Likeness's compliance design IS the consent infrastructure that processors and regulators are converging toward. Competitors retrofitting this ship slower. Real but compresses over time.
First-mover credibility in the consent-first + explicit-allowed cell. No funded direct competitor today. Real but finite.
Concierge-phase creator relationships. If 5-10 creators participate and feel respected, peer referrals become the primary acquisition channel. Trust networks in the adult industry are tight; positive reputation compounds. Real, accumulating, hard to copy directly.

What's NOT an unfair advantage (despite often being claimed)¶

Better AI quality — commoditizing fast. Don't lean on this.
Better UX — important hygiene; not a moat.
Capital efficiency — at $1.5M pre-seed, Likeness has less, not more, capital than mainstream consent-infrastructure competitors. Capital is not the moat.

What WOULD become an unfair advantage if executed¶

A deep partnership with a creator-rights advocacy organization that endorses (or doesn't oppose) Likeness's worker-controlled framing.
An ML Lead with materially better watermark adversarial robustness than community standard.
A multi-processor agreement that's robust to any single processor pulling out — Civitai didn't have this; Likeness should.

Strategic Connections¶

The "Architecture commitments as positioning" frame in this canvas connects directly to 01-discovery/competitor-landscape.md's differentiation analysis.
Revenue streams reflect the 59%-one-off-vs-41%-subscription split surfaced in 01-discovery/raw/customers-demand.md.
Cost structure matches the $1.5M raise sizing in docs/budget.md — research confirmed this is plausible but tight.
Key metrics reference both standard AARRR and the operational floor metrics that are load-bearing for the regulated category.

Flags¶

Red Flags: - Customer Segments and Channels both depend on hires that haven't happened. Creator Ops cofounder is the channel; CEO is the processor BD. Risk concentration is high.

Yellow Flags: - LTV / per-creator revenue figures are estimates, not validated. Concierge-phase data will replace them. - "Unfair advantage" list is honest about what's actually defensible vs. what looks defensible. The four real items are real but each has a finite half-life.

Sources¶

01-discovery/market-analysis.md — TAM / SAM / unit economics
01-discovery/competitor-landscape.md — competitive positioning
01-discovery/target-audience.md — segments and pains
01-discovery/industry-trends.md — regulatory / market trajectory
docs/founder-brief.md and docs/budget.md — original product and capital plan