Policy answers for the person asking
Most policy answers depend on who is asking: their region, tenure and contract. The Policy Answerer applies the actual policy to the actual person, and knows what a bot should never answer.
"How much parental leave?"
Policy AnswererPolicy question. Answering from the source, not from memory.
Searching the policy library3sChecking their tenure & region2sThe answer depends on region, and theirs qualifies for the enhanced policy, not the headline number.
Replying with the cited answer1s
How it works
Question in #ask-hrEmail to HR inboxDM to the agentPolicy updated
Cited answer in-threadForms & next steps linkedSensitive cases routed privatelyHandbook gaps loggedTriggered by a question in Slack
Questions asked in the HR channel, by DM, or by email start a run instantly, at any hour.
Reads the policy, checks the person
The agent finds the governing sections, applies the asker’s region, tenure and contract, reconciles policies that interact, and recognizes the questions that need a human instead.
Replies in-thread with sources
Answers land where the question was asked, with the policy cited and the relevant form linked. Repeat questions are logged for the handbook.
Without Relevance
With Relevance
The question sits in the HR queue for two days, for something the handbook technically covers.
Answered in the Slack thread within a minute, with the policy document cited.
The generic handbook answer goes out, wrong for anyone outside the default region or tenure band.
The agent checks the asker’s region, tenure and contract first, and answers with the variant that applies to them.
Answers come from whoever replies, from memory, and hallway folklore hardens into policy.
Every answer links the exact section it came from. If the source does not say it, the agent does not say it.
The same five questions consume the team’s week, every week, forever.
Routine questions are handled instantly, and the team spends its time on the cases that need a human.
A grievance typed into a chatbot gets an automated answer, which is exactly the wrong experience.
The agent recognizes what it should not handle and routes it privately to an HR partner, without a canned reply.
Nobody tracks what gets asked, so the confusing sections stay confusing.
Repeated questions are logged as a signal, so the third parental-leave question this month becomes a handbook fix.
Held to a quality bar, on every run
Every answer is checked against eval test cases written for this exact job before it posts. If a run fails the bar, it never ships.
Policy AnswererPass rate · last 14 days
Sampling 2% of live runs · 95% passing this week
Train your agent like an employee
Onboard it with your playbooks and correct it in plain English. It learns the lesson for good, and nothing changes until you approve it.
Built for enterprise teams
Run agents on your real data with the access controls, audit trails, and residency guarantees enterprises require all built in.
Monitoring
Real-time visibility into every agent’s activity, performance, and cost.
RBAC
Control who can use, build, edit, and deploy agents.
Data residency
Multi-region deployment keeps your data within your required geography.
Version control
Full version history on every agent. Roll back to any previous state.
Audit logs
Every action, every decision, every tool call logged and exportable.
Human-in-the-loop
Set approval gates on any action or allow agents to ask questions.
SSO / SAML
Enterprise single sign-on with SAML 2.0. Centralize identity & access.
PII masking
Detect and redact personally identifiable information.
"The ability to be vendor agnostic and the ability to scale across a breadth of functions is a really key feature."
Levi Watters
Partner, KPMG Australia
"The key for us was how we can modularize industry knowledge and the best playbooks, and apply it."
Allen Roh
Senior Marketing Manager, Autodesk
"We're looking for every place where AI can allow sellers and customer success reps to be more engaged with customers."
Rob Giglio
Chief Customer Officer, Canva
"The ability to be vendor agnostic and the ability to scale across a breadth of functions is a really key feature."
Levi Watters
Partner, KPMG Australia
Team it up with more human resources agents
Every agent you add shares one stack
These agents all run on one platform: one gateway, one router, one eval suite, one audit trail. Every agent after the first ships faster.
- Triggered by key events or signalslike Zapier
- Given access to contextlike Zep
- Connected to your appslike Composio
- Access to all LLMslike OpenRouter
- Performance evaluatedlike Braintrust
- No-code agent builderlike Dust.tt
- Coordinated into teamslike CrewAI
- Chained into multi-step processeslike n8n
- Kept alive through long-running worklike Temporal
- Traced at every step of every runlike Langfuse
Put the Policy Answerer to work this quarter
We partner with enterprise teams to take this work off their plate. An agent built with you, proven against your quality bar, and live in weeks.