Today, AI systems infer your organisation from whatever they can find. Kodiac Agent gives them something better: a governed Brand Agent that answers directly — accurate, current and traceable — over MCP and trusted APIs. When AI can ask you, it stops guessing about you.
From £1,499/mo · Built on the foundations of Phase 2
Soon the entity evaluating your organisation won't be a person reading a page. It will be their agent, querying systems it trusts.
Phase 3 is about being one of those systems.
The Brand Agent is a queryable interface to the structured knowledge you built in Phase 2. AI assistants, customer agents and enterprise agents discover it over the Model Context Protocol or call it as an API — and get governed answers, drawn live from approved, versioned records.
Because the answers come from your knowledge layer, they are always current: pricing, availability, specifications and policies reflect real-time data, not a snapshot from the last crawl.
No new content is written for this phase. The structure you built in Phase 2 is the agent's entire knowledge.
AI systems & agents ask
AI assistants · Customer agents · Enterprise agents · Partner systems
Your Brand Agent
MCP + trusted APIs · guardrails · verification · interaction log
Governed knowledge answers
Accurate · Current · Traceable to an approved, versioned source
Participation without control would be a risk, not an advantage. Every capability here exists so that what your agent says is exactly what you approved.
Configured, not coded. Define scope, tone and boundaries; the agent answers only from governed knowledge, version-controlled with rollback.
Exposed over the Model Context Protocol, so AI assistants and agent frameworks can discover and query your organisation as a first-class source.
REST endpoints for systems that integrate directly — partners, marketplaces, internal platforms — all reading the same governed source.
Structured answers designed for machine consumers: typed fields, source attribution and confidence — not prose to be re-interpreted.
Dynamic responses drawn live from your knowledge layer and connected systems — pricing, availability and specifications as they are now.
Hard boundaries for pricing, legal and regulated topics. Keyword, topic and intent rules, written in plain language, enforced on every response.
Signed responses and token authentication, so querying systems can verify they are talking to your official agent — not an imitation.
Every interaction recorded and traceable. Intent classification surfaces high-value conversations and routes them to your CRM.
Test before going live: pre-built scenarios, simulated AI systems, adversarial prompts and per-response source attribution.
An agent is only as trustworthy as the knowledge behind it. Kodiac Agent works because Phase 2 already did the hard part: one structured, governed, versioned representation of your organisation. The agent adds an interface, guardrails and identity — it never invents knowledge of its own.
We don't deploy Brand Agents on ungoverned knowledge — an agent that answers from fragmented sources would repeat the very problem Kodiac exists to solve. The path is deliberate: understand → improve → participate.
AI stops assembling your organisation from fragments. It queries you — and relies on what it gets back.
From observed, to understood, to relied upon.
Three phases of organisational maturity in an AI-mediated world. Agent is the destination — and each phase builds directly on the last.
Starter, Growth and Enterprise tiers, sized by interactions and endpoints. Guardrails, verification and the interaction log included in every tier.
Start with the free audit to see how AI understands you today — then we'll map the path from understood to relied upon.
