ProductNatural Language Rule Engine
Rules your team approves. Agents that can't quietly ignore them.
The rules are sentences, not code. Trail proposes them from your source material in the same plain English your policy is written in; your team reads, edits and approves them; every connected agent then applies the approved version with the clause it came from attached. Nothing becomes available to an agent because a model inferred it.
What happens to a rule between the document and the agent.
Four stages, and a human decides at the second one. That gate is the product, not a setting — an unreviewed rule is never reachable by an agent.
- 01
Proposed
Trail reads the source material and drafts the rule, together with an explanation and the exact text it was drawn from. Upload an SOP and it also outlines the agent's role, splits the process into phases and attaches the tools each step needs.
- 02
Reviewed
Your team edits, approves or rejects. Where a proposed rule conflicts with one you already approved, both are shown side by side with their sources, and you decide which governs.
- 03
Scoped
An approved rule carries what it governs — entity, account, plan, jurisdiction — and who may read it. A finance rule can be invisible to every agent outside finance.
- 04
Served & audited
Agents read rules through MCP, a retriever, or the API. Every read, approval and version is logged, so what an agent applied on a given day can be reconstructed.
What a rule carries
A rule is not a prompt — it's a governed object.
Prompts drift, get copied between tools and lose their origin. Each of these five properties travels with every approved rule, which is what makes an agent's answer reviewable months later.
Provenance
Source citation · Where did this come from?
The document, page and exact text it was drawn from
Scope
Applicability · What does it govern?
Entity, account, plan, payer or jurisdiction
Approval
Governance · Who signed it off?
A named approver, with a timestamp
Version
History · What did it say in March?
Prior versions retained, so past decisions replay
Access
Control · Who can read it?
Role-scoped, for people and for agents alike
From the clause to the rule.
Trail reads the document, proposes the rule, and shows the sentence it rests on. Your team approves it before any agent can use it.
…discounts above fifteen percent require approval by the regional director, and may not be combined with promotional pricing within the same order…
If a discount exceeds 15% or is combined with promotional pricing, require regional director approval.
- Written in plain English
- Human approval before use
- Conflict detection against approved rules
- Versioned against its source
- Role-based access
- Full read and approval audit log
- MCP, REST & GraphQL access
Consumed by the agent your team already runs.
Trail isn't a destination app. Point Claude, ChatGPT, Copilot or your own stack at one Trail brain through a native Model Context Protocol server, native retrievers, or plain REST and GraphQL.
Security built in, not bolted on.
Your data and your company brain never leave your boundary. Independently audited every year, with compliance controls enforced by the platform — not promised on a page.
SSO and SCIM, role-based access, and an audit log on every read — deployable single-tenant or inside your own VPC with residency pinned.
SOC 2 Type IIOngoing security controls
GDPREU data protection
ISO 27001Global ISMS standard
HIPAABAA on enterprise plansRule engine questions.
Can an agent create or change a rule?
No. Agents read approved rules. Proposing comes from Trail, deciding comes from your team, and an unreviewed proposal is not reachable by any agent.
What happens when two rules conflict?
The conflict is surfaced during review, with both rules and both sources side by side. Most resolve through scope — the two rules govern different accounts or jurisdictions — and the rest are a decision your team records.
Can we edit a proposed rule before approving it?
Yes. Edits are part of review, and an edited rule is versioned like any other change, keeping the link to the text it came from.
How do our agents actually read the rules?
Through the native Model Context Protocol server, retrievers for LangChain and LlamaIndex, or plain REST and GraphQL — whichever your stack already speaks.
See it built on your own context,
with your agents.
Bring one policy document and one real decision. We'll show you what Trail parses out of it, and where each piece came from.
Looking for your team's use case? See all solutions →