Agent

Policy & benefits desk

Most questions the People team answers have already been answered in writing somewhere. This agent replies in the channel where people ask, quoting the policy and linking to the page it came from. Anything sensitive, and anything the handbook does not cover, goes to a named person instead of being answered.

How it runs

Step by step

01

Index the policy set

Handbook pages, benefits documents and process notes are indexed from the sources the People team connects, keeping their access rules.

02

Take the question in Slack

An employee asks in plain language, in the channel or direct message they already use, without filling in a form first.

03

Answer with the source attached

The reply quotes the relevant policy and links to the page and version it came from, so the employee can check it themselves.

04

Escalate anything sensitive to People

Pay, performance, grievance and exit questions are handed to a named People contact rather than answered by the agent.

05

Say when it does not know

Where the policy set has no answer, the agent says so and passes the question on. It does not infer a rule from adjacent pages.

06

People Ops reviews the edge cases

Escalated questions and anything outside the published handbook are answered and approved by People Ops before they reach the employee.

The harness

Exactly what this agent can see, touch and change

The same five controls sit behind every Askollo agent. These are this one's settings — visible before you build it, not buried in an admin screen afterwards.

Context

What reaches the model

People teams whose inbox fills with leave, benefits and process questions, especially across several countries with different policies.

Escalation

When it asks a person

People Ops approves any answer that goes beyond the published handbook

Applications & Rights

Which tools it uses, and what it may do in each
NotionRead only
SharePointRead only
Google DriveRead only
SlackReadWrites newCreates new records or documents. Never edits, moves or deletes anything that was already there.

Verification

How you know it’s right

Every claim links to the document it came from. A statement the agent cannot cite does not make it into the output — which is what makes the result reviewable in minutes rather than re-read end to end.

Who it’s for

People teams whose inbox fills with leave, benefits and process questions, especially across several countries with different policies.

What you’ll need

  • A current handbook in a connected source
  • An agreed list of topics that must escalate
  • Channel access for the agent
  • A named People contact on rota

What you get

  • Cited answers in the channel
  • A queue of escalated questions
  • A log of questions with no policy answer

What it doesn’t do

It quotes and links to the policy you have written. It does not interpret a policy for an individual case, give employment law advice, or handle anything about a specific person's pay or performance.

How you get it

We build the first one with you

Not a template you configure alone. We sit with your team, build it on real data, and hand over the controls.

01

Scope

One session with the people who actually do the work. We agree what the agent reads, what it may write, and who approves.

02

Co-build

Built on your own data, not a sandbox. You watch it being made, so you know why it behaves the way it does.

03

Handover

You own the controls. Change the context, tighten the rights, move the approval gate — without coming back to us.

What it replaces

Parts of the work currently spread across the categories below. It does not replace any of those products outright.

Repeat handbook questions in the People inboxWiki searches that return nothing usefulPolicy answers given from memory
Estimated savingNot setNo figure ships until someone at ollo owns it and the method behind it.

Frequently asked

What stops it inventing a policy?

Answers come from the connected policy set and every one carries the page it came from. Where nothing in the set answers the question, it says so and routes the question to a person. That is a configured boundary, not a preference.

Which questions never get an automated answer?

You choose the list during the co-build. Pay, performance, grievance, disciplinary and exit questions are the usual set, and the agent hands those to a named contact without attempting an answer.

Does it work across different country policies?

Yes, where each country's policy is connected and the agent can tell which one applies to the person asking. If it cannot tell, it asks rather than picking the policy that happens to be nearest.

How do we keep the answers current?

It reads the live source, so an updated handbook page changes the next answer. The log of unanswered questions also shows People Ops where the handbook is thin, which is often more useful than the answers themselves.

Let's Build

AI is a capability you build. Let's build it together.

30 minutes with our team and you'll leave with a real plan — not a sales pitch.