PersonaGuard

Idea

A drop-in compliance SDK for AI companion and character-chat apps. PersonaGuard turns California, New York, and Washington's new AI-companion-safety laws into a widget and API a two-person app studio can ship this week: the required AI-disclosure banner, the break-reminder cadence, and a self-harm crisis-response layer, all logged to an immutable audit trail. It replaces hand-rolled compliance UI and law-tracking newsletters with actual runtime code.

Market gap

Three states now regulate AI companion chatbots directly, and more are moving fast. California's SB 243 took effect January 1, 2026, and lets any injured user sue for the greater of actual damages or $1,000 per violation plus attorney fees. New York's AI Companion Models Law has been in effect since November 5, 2025. Washington's HB 2225 / SB 1546 followed in March 2026. Baker Botts counted 78 chatbot-related bills filed across 27 states in the first weeks of 2026 alone, on top of 58 pending lawsuits. Each live law mandates specific, checkable behavior: an AI-not-human disclosure at session start and at least every 3 hours, extra break reminders and content limits for minors, and a self-harm and suicide crisis-referral protocol that cannot be suppressed.

None of this ships by default. The market is genuinely underserved at the builder level. The one product I found that's built specifically for this space, ailawsbystate.com, sells a bill tracker and alert subscription (Pro at $39/mo, Team/Enterprise at $149 to $499/mo) that tells an operator what the law requires. It does not ship a widget or SDK that implements the disclosure, break-reminder, or crisis-routing behavior itself. General AI-governance platforms (Credo AI, Holistic AI) sell risk-documentation workflows to enterprise legal teams via demo-gated sales, not a runtime a solo developer drops into a React Native chat screen. A spot check of current App Store listings for AI girlfriend and companion apps (My AI Girlfriend, AI Girlfriend Chat, AI Girlfriend - Katarina, and others) shows only static "fictional, 18+" disclaimers. None show the break-reminder or crisis-protocol UI the live laws now require.

Timing is now because the two biggest laws are already in effect and enforcement risk is not hypothetical: Google and Character.AI settled a wrongful-death lawsuit in January 2026 over a companion chatbot's role in a teenager's suicide. That is the exact fact pattern the crisis-protocol requirement targets, and it happened at the largest, best-resourced company in the category. Every smaller operator carries the same exposure with none of the legal budget.

Total Addressable Market (TAM)

Bottom-up, from the operator side.

• Base: 337 active, revenue-generating AI companion apps worldwide as of 2026, with 128 released in 2026 alone, a 60 percent year-over-year increase (companionguide.ai market report).
• Assumption: roughly 85 percent (286) have US-reachable users and therefore fall under at least one live state law, since English-language consumer apps on the US App Store and Play Store default to US distribution. Source: same report plus the state law scope descriptions above (CA/NY apply to any operator serving state residents regardless of where the company is based).
• Assumption: the largest 30 percent of that group (the handful of brands with tens of millions of users, like Character.AI and Replika) already have in-house legal and engineering to build this themselves or have already been sued into fixing it, and are not the buyer. That leaves a SAM of roughly 200 operators: small and mid-size companion/character-chat studios.
• Pricing: MAU-tiered plans averaging roughly $3,600/year blended across Starter, Growth, and Scale (see Pricing strategy).
• SAM: 200 operators x $3,600/year = $720,000/year at today's app count.
• SOM (Year 1): capture 6 to 8 percent of SAM through direct outreach into indie AI-app-builder communities and app-store-adjacent SEO = roughly 13 to 16 paying accounts, or about $50,000 in Year 1 ARR.

This is a small but fast-refreshing market: 128 new companion apps launched in 2026 alone, and the number of states with a live law more than doubled in under a year. Both inputs to the TAM (operator count, states requiring compliance) are growing, not static.

Monetization strategy

Self-serve SaaS subscription, billed monthly by tracked MAU, no sales call. The buyer is a small technical team that wants to drop in an API key and a widget component, not sit through a demo. They keep paying because the obligation is ongoing, not a one-time fix: new states add new rules on a rolling basis, and losing coverage means re-exposing the app to $1,000-per-violation liability under a live private right of action.

Pricing strategy

• Starter: $149/mo, up to 5,000 MAU. Covers CA, NY, and WA disclosure, break-reminder, and crisis-routing logic, 90-day audit log retention.
• Growth (anchor tier): $349/mo, up to 50,000 MAU. All current and newly added states, 1-year audit log retention, minor-detection heuristics, priority support.
• Scale: $799/mo, up to 250,000 MAU. Dedicated support channel, localized crisis resources, exportable compliance reports.
• Enterprise: custom, above 250,000 MAU or white-label needs.

The Growth tier is the anchor because most small and mid-size companion apps in the 5,000 to 50,000 MAU range live there. Starter is priced well under a single hour of outside counsel's time and under ailawsbystate.com's own Enterprise tier, despite doing more (implementation, not just alerts).

Lead magnet

A free, no-signup Companion Chatbot Compliance Scanner. An operator pastes their system prompt or a screenshot of their onboarding and chat screen, and the tool returns a state-by-state gap list: which of the CA, NY, and WA required elements (AI disclosure, break-reminder cadence, crisis protocol, minor safeguards) are present or missing, with links to the exact statutory text. It proves the specific gap before asking for a signup.

Social proof that the problem exists

• Google and Character.AI settled a wrongful-death lawsuit in January 2026 tied to a companion chatbot's role in a teenager's suicide, the exact scenario the crisis-protocol requirement exists to prevent: https://www.cnbc.com/2026/01/07/google-characterai-to-settle-suits-involving-suicides-ai-chatbots.html
• Developer community reaction on Hacker News to California becoming the first state to regulate AI companion chatbots, the audience this product would sell to directly: https://news.ycombinator.com/item?id=45573028
• Baker Botts tracked 78 chatbot-related bills filed across 27 states in the first weeks of 2026 alone, alongside 58 pending lawsuits, showing the obligation is expanding fast and not confined to California: https://ourtake.bakerbotts.com/post/102mipe/ai-chatbot-regulation-78-state-bills-58-lawsuits
• ailawsbystate.com already charges $39 to $499/mo for a law tracker and alert service alone, direct evidence operators are already paying to manage this risk even without an implementation product: https://www.ailawsbystate.com/blog/companion-chatbot-laws-state-compliance-guide-2026
• Market sizing showing 337 active revenue-generating companion apps and 128 new ones released in 2026, a 60 percent year-over-year increase in the exact customer base this product sells to: https://companionguide.ai/news/ai-companion-market-120m-revenue

Competitors

• ailawsbystate.com: a companion-chatbot law tracker and alert subscription with a bill-comparison tool and cost estimator. Informational only. No SDK or widget that renders the required disclosure or crisis-response behavior.
• Limina AI: a de-identification platform that strips PII, PHI, and payment data from chatbot transcripts for HIPAA and CPRA exposure. Adjacent to companion chatbot operators but solves a privacy problem, not the disclosure/break-reminder/crisis-protocol requirement.
• Credo AI / Holistic AI: enterprise AI-governance platforms selling risk-assessment frameworks and documentation workflows to Fortune 500 AI and legal teams via sales-assisted deals. Not priced or built for a two-person app studio.
• In-house, ad hoc: what most of the 337 active companion apps do today, per the App Store spot check above: a static 18+ or fictional-character disclaimer and nothing else.

What competitors offer now

ailawsbystate.com ships a searchable law database, a compliance deadline calendar, per-bill status alerts, and a "Compliance Widget" tool that, per its own site content, is not documented as an actual embeddable disclosure or break-reminder component. It tells an operator what the law requires. Limina focuses entirely on scrubbing personal data out of transcripts and has nothing to do with the disclosure UX or crisis-response requirement. Credo AI and Holistic AI sell risk-assessment frameworks for enterprise model governance committees, reached through a demo request, not a self-serve API key. None of the individual companion or AI-girlfriend apps checked show a break-reminder timer or an inline crisis-resource response, only a static age and fiction disclaimer set once at install.

What can be done differently to attract customers

Ship the runtime, not the research report. A JS/React Native widget plus REST API renders the exact, state-correct disclosure copy and break-reminder cadence based on the user's detected state, runs a lightweight classifier on outbound bot-context messages for self-harm and suicide signals, injects the required crisis-resource response inline when triggered, and writes every disclosure and crisis event to an immutable, exportable audit log sized for a world where a single missed disclosure can cost $1,000 in statutory damages. Enter through the free Compliance Scanner instead of a law-firm-style content funnel, sell self-serve with a card and an API key, price by MAU instead of enterprise seats, and go find these builders where they actually are: indie AI-app-building Discords and communities, IndieHackers, and the same app store category pages their apps already live on.