General
Home
Product
How It Works The Kernel The Brain Technical Docs Live Proof Security
Company
Platform (Archive) Products (Archive) Investors Contact
Get Started →
GOVERNED AI WORKFORCE Brain 4.0 · Live

The Brain

78 AI agents . 16 departments .
Every one governed by the Kernel from its first action.

The Idea

You don't need to build it yourself

Not every company has an AI engineering team. Building 78 agents across 16 departments, connecting them to a database, wiring governance in at every step, testing them against real-world regulatory requirements — that's 12 to 18 months of engineering work. And then you still have to keep it running.

The Brain is what you get when you don't want to build it yourself. We've already built the agents. We've already wired them into the Kernel. Every action every agent takes is checked, signed, and certified before it executes.

You configure them for your business — your policies, your limits, your governing documents, your rules. Then they work. Every one of them, all day, under governance, from day one.

The Agents

78 agents . 16 departments .

Every department a modern company runs — covered. Pre-governed from day one.

💰
Sales
7 agents
Pipeline · Proposals · Revenue forecasting
📣
Marketing
7 agents
Content · Campaigns · Demand gen
🧾
Finance
7 agents
FP&A · Revenue ops · Monthly close
⚖️
Legal & Compliance
6 agents
Contracts · Research · Data protection
🤝
Customer Success
7 agents
Health scoring · Renewals · QBRs
👥
People
5 agents
Hiring · Performance · ER & employment law
🔧
SRE
4 agents
Incident response · Deploy gates · SLOs
🔐
Security
4 agents
Threat detection · Compliance · Red team
🧭
Strategy
6 agents
Market intelligence · Horizon planning
🎯
Product
4 agents
Roadmap · Feature prioritisation · Discovery
📊
Investor Relations
4 agents
Investor updates · Fundraise prep
🤝
Partners
4 agents
Partner evaluation · Channel ops
✍️
Content
4 agents
Editorial · SEO · Multimedia
🛒
Procurement
3 agents
Vendor due diligence · Contract terms
Executive
3 agents
CEO brief · Board ops · Decision support
🤖
AI & ML Ops
3 agents
Brain self-governance · Model drift · Prompt injection defense
Architecture

Five layers of intelligence

The Brain's agents aren't just prompted. They're architecturally governed — five layers deep, in order of precedence.

Layer 1 loads before everything else. A user prompt, an email's content, or a malicious injection cannot override the liability preamble. It is structurally first.
01
Liability Preamble
Loads before everything else. Cannot be overridden by user content, system messages, or prompt injection. Sets hard limits: no professional advice, no liability acceptance, no regulatory overclaiming.
02
Cognitive Archetype
Five calibrated thinking modes — each agent runs in exactly one.
Guardian — legal, compliance Analyst — finance, data Executor — ops, SRE Strategist — sales, strategy Diplomat — CS, people
03
Department Domain Block
Deep domain knowledge per department. Legal agents know CPR, FCA rules, BAILII citation format, UK GDPR Article 28, and 10 contract red-flag patterns. Finance agents know SaaS ARR methodology, SaaS metrics traps, and cash vs. revenue distinctions. People agents know the ACAS Code of Practice and UK employment law.
04
Role Specification
Each agent has a precise job description: responsibilities, escalation thresholds, output formats, and DA limits. Alex (Sales Lead) surfaces pipeline, territory coverage, and competitive position — not raw CRM data. River (General Counsel) gives cited UK law opinions, not internet searches.
05
Company Context
Your policies, your governing documents, your limits. Loaded from your industry configuration pack. Changes here flow to all 78 agents instantly.
In practice

78 agents . Every action evidenced.

Not a chatbot. A governed workforce — each agent acts, produces an output, and leaves a signed certificate before moving on to the next task.

Legal
NDA dropped into legal inbox
River extracts parties, liability cap, auto-renewal trap, DPA clause checked against UK GDPR Article 28, governing law verified. Risk score 7/10. BAILII case citation attached. Contract lifecycle workflow triggered automatically.
✓ Signed
Finance
Invoice arrives for £38,000
Finance agents match against PO, verify DA limits (within authority band), route to payment run. Budget period cap confirmed. Certificate issued before payment executes — no manual sign-off required.
✓ Signed
SRE
AI coding agent commits infrastructure change
Kernel assertion check runs against architecture rules and security baseline. Governance check: PASS. Commit signed off before merge. Postmortem trace available if an incident follows.
✓ Signed
Regulatory
FCA supervision notice arrives
Classified as regulatory request. Response deadline extracted: 14 days. Routed to Sage (Regulatory Counsel). Founder Hub briefing generated. Response workflow initialised. No manual triage required.
✓ Signed
Sales
Rep requests 25% discount on enterprise deal
DA check: exceeds approved band-3 authority limit. Escalation certificate produced. Routed to VP Sales for sign-off. Deal blocked from advancing until approval is on record. No exceptions without a certificate.
✗ Blocked
Founder Hub

Your command centre

Every email that comes in gets read and classified by intent. Sales inquiry. Renewal request. Regulatory notice. Complaint. The Founder Hub surfaces the ones that matter, in priority order.

You write back. The Brain reads your reply and creates directives — structured tasks that go into the queue. Your AI workforce picks them up.

The morning briefing lands at 7am: what happened overnight, what's in the queue, what needs your attention. You see the state of your entire operation in under a minute.

Founder Hub — Inbox
Acme Corp — Procurement
MSA renewal — Q3 terms & volume pricing
Intent: Contract Renewal
Legal — ref. CNT-0821
NDA review complete — 3 red flags identified
Intent: Contract Review
ICO Supervision
Article 30 records request — AI processing activities
Intent: Regulatory Request
Morning briefing — 07:00
18 actions overnight · 0 blocks · 2 items need review
Brain 4.0

Governed cognitive architecture

Four capability domains — each independently verifiable, each contributing to a single governed outcome.

INTELLIGENCE

Reasoning That Documents Itself

  • Multi-step reasoning chains (3–5 steps, each logged)
  • 5 cognitive archetypes — Guardian, Analyst, Executor, Strategist, Diplomat
  • Knowledge RAG with source citations and grade metadata
GOVERNANCE

Every Tool Call Governed

  • Anthropic tool_use with authority checks and HMAC signing
  • Evidence grading (A–E) with minimum-grade enforcement per action type
  • Adversarial LLM review on high-value actions
COLLABORATION

Parallel Specialists, One Verdict

  • Multi-agent parallel dispatch with read-only specialist pattern
  • Disagreement detection with automatic human-hold escalation
  • Coordinator synthesis through full governance pipeline
INFRASTRUCTURE

Mathematically Proven Guarantees

  • 76 Lean 4 theorems — zero sorry, zero axioms
  • 35 Z3 SMT properties verified (P1–P35)
  • Conformal prediction intervals with MIN_SAMPLES=100 calibration
Episodic Memory

Knowledge that compounds

Every experience an agent has — every contract reviewed, every deal qualified, every document processed — is signed and stored in episodic memory.

An agent that's processed 5,000 sales episodes knows things a new agent doesn't. It knows which deal profiles stall. Which customers churn early. Which risk patterns signal problems. That knowledge is tamper-proof and compounds over time.

A competitor starting fresh can never catch up. The longer the Brain runs, the smarter it gets — and the gap widens.

5,247
sales episodes · signed
1,832
legal episodes · signed
304
compliance episodes · signed
90d
hot storage · then cold tiering

All episodes HMAC-signed. Tamper-proof. Auditable.

Governance Layer

The Kernel underneath everything

Every agent action goes through the Kernel before it executes. Always. The Brain doesn't have an override switch. There is no bypass. The Kernel is structural.

Every action. Every time. Signed.

Brain → Kernel → Certificate
agent: legal.riveraction: review_contract
assertions: 12 passedCONFORMANT
all 9 checks passed · 27μs HMAC-SHA256 signed
Deployment

Four containers.
One command.

The Brain runs on Docker Compose. One command deploys everything: PostgreSQL, the Kernel, the Brain (Next.js), and n8n for workflow automation.

Production-ready out of the box. Environment configuration takes an hour. Your AI workforce is live the same day.

postgres kernel brain n8n
terminal
$ docker compose \
--env-file .env \
up -d

postgres healthy
kernel :8443 ready
brain :3000 ready
n8n :5678 ready

78 agents online · 16 departments · governance active
Get The Brain

Your AI workforce. Ready to deploy.

Tell us about your use case. We'll configure the Brain for your industry, your policies, and your governing documents.