Govern

Monitor all tool calls and optimise token usage.

Because the gateway is in the path of the call rather than reading about it afterwards, the record is complete, the rules are applied before anything happens, and the part of the bill nobody reads never gets paid for.

Audit log

Every call and every activity, recorded

One record covering two kinds of event: the tool calls made through the gateway, and the activity coming back from the provider APIs. Who, what, when, and the reason attached.

What it does
  • Every tool call, allowed or refused, with why
  • Every file, chat and shared project the provider APIs report
  • Filter by person, team, capability, tool or window
  • Export as evidence, or push into the case system already in use
Rules engine

Deterministic rules, with judgement on top

Your own watch list and policies applied to everything, not a sample. Deterministic rules give the same answer every time; a model reads for the sensitivity a rule could not anticipate.

What it does
  • Watch list terms, keywords, regexes, file names, user groups, MCP servers
  • Same input, same answer, every time — the part you show a regulator
  • AI classification for business context: a codename nobody registered, a client spelled three ways
  • A risk score on every event
Exceptions

Raised with the reason attached

What the rules engine surfaces becomes a case somebody can act on: the event, the rule it matched, the person, and enough context to decide without going and asking.

What it does
  • Raise against your own rules, not a generic template
  • The matched term and supporting event travel with the exception
  • Assign, resolve and record the decision
  • Push into ServiceNow, Archer, Actimize or whatever is already reviewed
Token optimisation

Stop paying for what nobody reads

A tool returns everything the record has and the model reads a fraction of it. Distill trims the response in the path of the call, before it reaches the context window.

8,801 → 1,853tokens on one apollo_organizations_enrich call
79%fewer, on that call and every one after it
212,028,063tokens saved across 3,500 calls last month
US$636off the bill, at $3 per million
What it does
  • Trim bloated tool responses before the model sees them
  • Applied at the gateway, so every capability benefits at once
  • The outcome does not change; the bill does
  • Reported per capability, so you can see what each one costs to run

See govern against your own team.