AI is commoditizing professional knowledge. CBRP builds the infrastructure to develop and capture what remains: human judgment.
$1–$1.25M Seed · cbrp.ai
"Entire classes of jobs will go away."
Sam Altman · CEO, OpenAI
"Being a lawyer or an accountant or a project manager — most of those tasks will be fully automated by AI within the next 12 to 18 months."
Mustafa Suleyman · CEO, Microsoft AI · Co-founder, DeepMind · Feb 2026
"AI automation is taking away the fundamental learning experiences that built legal judgment. This could lead to a generation of lawyers with diminished independent judgment."
New York State Bar Association · 2025
"Within four years, AI will be smart enough to do most cognitive work — any job done with a laptop."
TIME Magazine · AI Leaders Survey
"AI will eliminate 50% of entry-level white-collar jobs within five years — including law, finance, and consulting."
Dario Amodei · CEO, Anthropic · May 2025
The builders say it's inevitable. The profession says it's not prepared.
The Answer Isn't Less Human
The professionals who thrive won't be those who just become more efficient by using AI.
They'll be the ones AI made sharper.
Personalized legal intelligence that keeps in-house lawyers informed before it matters.
AI is automating the reps and tasks that built judgment. Acuity gives them back — so the 50% who remain stay ahead of the curve, not behind it.
DirectorBrief delivers the intelligence a director needs to satisfy their fiduciary duty when AI-driven information surplus makes it harder, not easier, to know what matters.
"Legal advice depends on professional judgment and interpersonal trust. Complex balancing acts requiring human judgment, ethical consideration, and organizational understanding are beyond AI's capabilities." — NYSBA
Human-first. AI-powered. Built to sharpen, not replace.
Jayden Daniels broke the NFL rookie completion record using VR to simulate thousands of defensive reads — because the old model of learning through physical reps couldn't keep up. Legal is facing the same problem in reverse: AI is automating the document review, contract markup, and research that built pattern recognition. The apprenticeship reps on many tasks are gone or going. Associates need a flight simulator.
The NYSBA warns AI is creating an "AI Crutch" — removing the grunt work that built foundational legal skills.
AI handles more of the process. But the volume of regulations, enforcement actions, and compliance obligations isn't slowing down — 97% of GCs report it's accelerating. Every one of those developments requires someone to decide: does this matter to us? AI can draft the memo. It can't make the call. The judgment bottleneck is getting worse, not better.
FTI/Relativity General Counsel Report, Feb 2026
The legal AI ecosystem is building execution tools (Harvey, Ironclad, CoCounsel). Without intentional development of the judgment layer, AI-driven efficiency sacrifices the fundamental skill AI cannot replace: human judgment. The company that captures this preference data first is positioned to become the integration layer.
Once execution tools commoditize, the judgment layer becomes the bottleneck. First mover owns the data.
| Harvey | Focused on contract execution and AI-assisted drafting |
| GC AI | Focused on document interaction and retrieval |
| CoCounsel | Focused on legal research workflows |
| Ironclad | Focused on contract lifecycle management |
| Diligent | Focused on board materials and governance process |
Their focus is execution. They make work faster. They don't help you think better.
| Proactive | Surfaces what matters before you ask |
| Role-aware | Same regulation, different framing for different roles |
| Contextual | Knows your companies, committees, jurisdiction |
| Networked | Learns from collective attention patterns |
| Training | Actively develops judgment via micro-learning tools |
This layer doesn't exist yet. CBRP is building it.
Information informs. Engagement develops judgment. Every CBRP product is designed to develop, not just deliver.
Personalized legal intelligence for in-house counsel. Every alert is a practical application built from real developments, not a headline. Company-specific context turns generic news into specific briefings. Every scored interaction builds a portable Judgment Profile: revealed preferences that compound over time. We don't compete with Harvey. We feed it. 100K+ US GCs.
The flight simulator for legal judgment. Same engine, different persona. Two deployment models: direct SaaS subscriptions + sandboxed law firm deployment (their API key, their data, nothing touches our servers). AI is eliminating the reps that built pattern recognition. Acuity replaces them — so the associates who remain become the next generation of human judgment. CLE credit potential.
The NYSBA is calling for exactly this: exercises where lawyers deconstruct AI output, verify reasoning, and build independent judgment.
Board intelligence for individual directors. Nobody serves the individual director: Diligent sells to the corporate secretary, AlphaSense costs $10K+/seat, NACD is education only. Committee-contextualized alerts. D&O claim severity up 27% to $56M avg settlement. 250K+ US board seats.
Same intelligence engine, calibrated for a non-legal audience. 33M+ US small businesses, 77% relying on Google for regulatory guidance. Translates regulatory complexity into clear, actionable guidance business owners can act on without a lawyer. Low lift, low price point, largest TAM. The volume play on the same engine.
One engine. Four market gaps. Each one deepens the advantage.
CounselBrief $499/yr individual + enterprise
Source: ACC, CLOC membership data
CB Acuity SaaS + sandboxed enterprise
Source: NALP, AmLaw data
DirectorBrief at $149/mo
Source: BoardEx, NACD, PitchBook data
RegulatorPulse at $29–49/mo
Source: SBA, U.S. Chamber of Commerce
Each market has different buyers, different price points, and different pain. The same engine — with a different persona — serves all four. Every vertical teaches the engine something the next one inherits.
What we don't collect: AI conversation content. Ever. The AI layer is private by design. This is architecture, not policy.
Level 1: Direct Match
GC scored IP assignment as "Matters" 12 times. Any preference form could capture this. Low barrier. Necessary but not sufficient.
Level 2: Adjacent Inference
GC never scored derivative works, but their IP assignment, trade secret, and open source scores triangulate around it. The system infers a priority the GC hasn't consciously articulated. No preference form captures this.
Level 3: Pattern Inference
GC scores regulatory enforcement as "Matters" across four unrelated practice areas. The system recognizes a cross-cutting organizational priority. Requires sustained longitudinal data across 15 practice areas.
Why This Data Matters
"Ant colonies work through a network of very simple local interactions.
Individual ants use their recent experience to decide what to do.
The whole colony operates through a network of these very brief interactions."
— Deborah Gordon, Stanford University
Relational. Discrete. Explicit.
Judgment scores across 15 areas
Smart Dismiss reasons
Company intelligence profiles
Vote signals & corrections
Acuity drill performance
What users tell us.
No single user creates this.
▸ What the profession cares about right now
▸ Where judgment is weakening across roles
▸ Judgment profile archetypes by industry
▸ Market-level regulatory attention signals
▸ Predictive patterns no survey captures
What the colony knows.
Vector. Continuous. Implicit.
Dwell time & engagement depth
Navigation path signatures
Topic combination patterns
Cross-user attention convergence
Temporal priority shifts
What users reveal without knowing it.
Through isolated interactions, each ant produces macro intelligence.
Similarly, each user's interaction with our tools produces valuable macro data that compounds with every user who joins.
Each user's company becomes an intelligence node. At signup, we build a structured profile: what they do, where they operate, what regulations apply. Every alert is evaluated against this model. "This EPA amendment impacts your polymer coatings product line," not "this may affect manufacturing." Cold start drops from months to weeks.
Across all users, attention patterns reveal what's emerging before it's news. When 30 fintech GCs shift focus to state money transmission licensing, the 31st sees it proactively. Cross-role correlation multiplies: GC + director at same company engaging the same development = confirmation signal neither produces alone.
Both models train from passive signals — dwell time, clicks, scroll depth. No chat logs. No surveys. Users engage with their briefing and the system learns.
Harvey handles process.
CBRP develops judgment.
Phase 1: Now
Third-Party LLM API
Third-party AI writes the practical applications. CBRP owns the enrichment prompts, the company models, and the scoring data. Product works today. Revenue starts here.
Advantage: Content architecture + scoring data
Phase 2: ~50 DAU
Fine-Tuned Open Model
Judgment scoring data trains a fine-tuned model on cloud GPU. The model learns what matters to GCs — not from documents, but from thousands of revealed preference signals. Cost per enrichment drops. Quality rises.
Advantage: Training data unique to CBRP
Phase 3: Scale
Self-Hosted Proprietary Model
Self-hosted on owned hardware. Zero API dependency. The model IS the product — trained on judgment data that exists nowhere else. A competitor can copy the UX. They cannot copy what the model knows.
Advantage: Proprietary model trained on proprietary data
The Judgment Profile API makes CBRP embedded infrastructure. Harvey ranks its results based on our risk profile data — surfacing the 4 issues this GC cares about most, not all 15. That's our profile working. Harvey knows what companies like yours care about. We know what you care about — because you told us, 20 different ways, over 12 months. We don't compete with execution tools. We make them smarter.
Every month of accumulated scoring data increases the value of the platform — to users, to integration partners, and to CBRP. The compounding starts the moment the first GC taps "Matters."
$1–$1.25M deployed into user acquisition and team. Targeted: GCs, directors, regulated businesses. Quality audience from day one.
Subscribers generate behavioral data. Behavioral data makes personalization better. Better personalization reduces churn. Lower churn makes acquisition economics better.
Collective attention signals detect emerging issues before they're news. The product improves as users engage. The data advantage deepens with every subscriber.
Subscriptions fund the data. Enterprise contracts (law firm Acuity deployments, API licensing) add higher-ACV revenue. Each new vertical extends the platform while costs stay fixed. Revenue grows across products on the same engine.
CEO & Founder
28 years in BigLaw. Saw firsthand how judgment is built — and how the apprenticeship model is breaking. Built all four CBRP products.
Full-stack/backend
Month 1 · $12–15K/mo
Frontend/infra
Month 2–3 · $10–12K/mo
Editorial & compliance
Month 3 · $6–8K/mo
The founder built the product because he lived the problem for 28 years.
Key signal: Product-market pull without paid acquisition.
| Year 1 | Year 2 | Year 3 | |
|---|---|---|---|
| Subscription ARR | $780K | $2.6M | $5.2M |
| Total Revenue | $825K | $3.0M | $6.0M |
| Paid Subscribers | ~950 | ~3,200 | ~6,000 |
This raise builds the team and funds the path to profitability. A Series A becomes a choice, not a necessity.
CBRP Co. | cbrp.ai | The Judgment Layer.
12-month
500 paying CB individual subscribers + first enterprise contracts. $500K+ ARR. Unit economics proven.
18-month
DirectorBrief at 150+ paying directors. CFOBrief launched. Total ARR $1.2M+.
24-month
4 verticals live. Attention network data self-reinforcing. Profitable — or raising Series A from a position of strength.
40% — Team
$500–625K
30% — User Acquisition
$375–470K
20% — Content Engine + Model
$250–312K
10% — Operations
$125–156K
"The subscription funds the data. The data trains the model. The model sharpens the product."