Partnership Proposal — PersonaSpec
Status: Open Date: 2026-07-07 Partner Role: CEO / Co-founder (Sales + Strategy + B2B)
Elevator Pitch
PersonaSpec is a web service that transforms psychometric testing into cognitive profiling infrastructure for LLMs.
A user completes a smart questionnaire (10–30 questions, forced-choice) → the system builds a cognitive profile (6 dimensions: how a person makes decisions, handles uncertainty, communicates, learns, and forms values) → generates a personalized system prompt for ChatGPT / Claude / Copilot.
Not MBTI. Not a template prompt. A personal AI interface that understands how you think.
Market (July 2026)
| Sector | 2026 | 2030 | CAGR |
|---|---|---|---|
| Prompt Engineering | $1.49B | $4.51B | 32% |
| Hyper-Personalization | $35.9B | $144.7B | 22% |
| AI Personalization (total) | $545B | $661B | 5% |
Key trends in 2026:
- AI in recruiting — 27% of HR departments already use AI tools; personality assessments in hiring grew 69% year-over-year
- Skills-based hiring — 81% of employers shift from credential screening to skill validation; cognitive profiles are the new standard
- Prompt engineering is no longer niche — it's becoming a mandatory layer in enterprise AI stacks
- Agentic AI requires user-specific calibration — behavioral policy beats generic prompting
Product
Current state (~70% complete)
Backend (TypeScript, Express):
- 6 cognitive blocks: uncertainty navigation, decision making, constraint handling, communication style, learning patterns, value systems
- 3 scenarios: quick (10 questions, 3-5 min), standard (18 questions, with DOSPERT), deep (30 questions, with DOSPERT + IPIP-NEO)
- Deterministic scoring + cross-validation (forced-choice × classic scales)
- LLM enhancement + LLM-as-judge (external profile validation)
- 8 REST endpoints, JSON file storage
Frontend (React + Vite):
- Live SPA: person.white-shadow.ru
- Scenario selection, one question per screen, result with confidence metrics
- Probe refinement for low-confidence dimensions
- Profile hub (save/view history)
- "Improve with LLM" button (generates a living system prompt)
What's unique:
- We don't type people — we build a behavioral policy (how a person thinks)
- LLM-as-judge: an external LLM validates profile consistency against raw responses
- Deterministic scoring (confidence > 0.6) — not a black box
- Language- and culture-invariant (forced-choice format)
What's missing for revenue
| Component | Status | Estimate |
|---|---|---|
| Payment system (ЮKassa / Telegram Stars) | ⬜ | 1-2 weeks |
| Pricing page + legal offer | ⬜ | 1 week |
| B2B features (permissions, audit, ephemeral) | ⬜ | 3-4 weeks |
| Rate limiting + production security | ⬜ | 1 week |
| Stealth calibration (adaptive questions) | ⬜ | 2-3 weeks |
| Cross-cultural (EN version) | ⬜ | 2-3 weeks |
Target Audiences
1. B2C — Individual Users (Prompt Engineering Market)
Who needs it: AI users who feel ChatGPT "doesn't get" their thinking style. Developers, writers, managers, analysts — anyone who works with LLMs daily.
Pain point: Generic prompts work for "the average." A personalized prompt improves response quality by 30-60% (our A/B tests).
Monetization: Subscription ($5/mo) or one-time report ($10).
2. HR / Recruiting (AI-Powered Hiring Market)
Who needs it: Companies tired of fake CVs and AI-generated resumes. HR departments transitioning to skills-based hiring.
Pain point: 69% growth in personality assessments in hiring (2025-2026), but existing tests (MBTI, DISC) are self-report and trainable. Our forced-choice approach + LLM validation gives objective results.
Monetization: B2B subscription (from $1,500/mo for corporate access).
3. Headhunters / Executive Search
Who needs it: Recruiters searching for "rare profiles" — people with specific cognitive traits for specific roles.
Pain point: You can't find a CTO with high uncertainty tolerance + low communication need + system-2 decision making from a resume. Our profile hub + cognitive dimension search is the first tool of its kind.
Monetization: Profile database access (from $500/mo), one-time search ($100).
4. B2B — Enterprise Segment
Use case 1: Team cognitive mapping — collect team profiles, identify cognitive gaps, find complementary hires.
Use case 2: LLM assistants for employees — each team member gets a prompt tuned to their working style.
Use case 3: Onboarding — a new hire's cognitive profile immediately shows how to communicate with them.
Monetization: Enterprise license (from $5,000/mo).
Partner Role
CEO / Co-founder focused on:
- B2B sales (first 10 corporate clients)
- Go-to-market strategy (priority: HR-tech → B2C → headhunters)
- Partnerships (HR platforms, recruiting agencies)
- Investment raising (seed round H2 2026)
What the partner gets:
- Equity (50/50 on co-founder terms)
- Full access to code, infrastructure, research
- Joint product roadmap ownership
- Immediate sales pipeline — contact base + demo access
Roadmap (H2 2026)
| Month | Milestone |
|---|---|
| July | Payment system launch, pricing page, B2B features |
| August | EN version, stealth calibration, first 10 B2B clients |
| September | Profile-as-a-Service (JSON Schema, LLM output formats) |
| October | Profile hub + cognitive dimension search |
| November | API for HR-tech integrations (ATS, HRIS) |
| December | Cross-cultural validation, seed round |
Why Now
- Market is exploding — prompt engineering grows at 32% CAGR, AI in HR at 69%
- Blue ocean — no competitor does cognitive profiling for LLM. Character.AI (entertainment), Replika (relationships), MBTI tests (typing). Nobody builds behavioral policy for AI interaction
- Product is 70% ready — not an idea, not an MVP, but a working service with validated scales and external audit (LLM-as-judge)
- First-mover advantage — forced-choice + cognitive domains + LLM integration = proprietary approach that's hard to replicate
Contact
Tech stack: TypeScript, React, Express, Vite, OpenRouter API Live demo: https://person.white-shadow.ru/ Proposal domain: https://partner.exo.white-shadow.ru/
PersonaSpec — cognitive profiling infrastructure. We don't type people — we tune AI to the person.