POLIS
Menu

How it works

Your team keeps its tools. The AI employee works in them, under its own name.

Every external action goes through one gateway that enforces your rules, and a named person decides wherever your policy asks.

The AI Developer

From a Jira ticket to a draft pull request your team reviews.

At L1 · Supervised, a person approves the plan before any code is written.

  1. 01 · Jira

    Ticket assigned

    To the AI employee’s own account.

    Each AI employee has its own named account, roles and limits.

  2. 02 · Slack · L1

    A person approves the plan

    It posts a plan. A named person approves it before any code is written.

    The decision is logged.

  3. 03 · Sandbox

    Code and tests, isolated

    No integration credentials or provider keys, and outbound traffic only to an allow-list.

    The sandbox runs under gVisor on servers. Its only token is a short-lived one for its own model calls.

  4. 04 · GitHub

    Draft PR on polis/*

    With a test summary. Polis never merges.

    Tokens are short-lived, issued per call and limited to one repository.

  5. 05 · Audit

    Your developers review

    As usual. Every step lands in the audit log.

    The audit log is append-only.

The Atlassian access manager

An access request, as a list of changes a person approves.

Groups, product access and Confluence space permissions in Jira and Confluence. Identity is confirmed by verified email, never guessed.

  1. 01

    Proposed as data

    Each change shows the person, the target, and the access level before and after.

  2. 02

    A person approves the set

    In the panel, or in Slack for sets of up to 10 changes. Slack approval is limited to approvers an admin names.

  3. 03

    Applied and verified

    Test mode by default. Each change is read back to verify it.

  4. 04

    Undo in one click

    An applied change can be undone in one click.

  5. 05

    Daily summary

    Owners get the access changes made, and any safety rule switched off.

The access manager in depth

Autonomy per project

Autonomy is a dial, set per project.

Start where your team is comfortable and move up when the results earn it.

  1. Level 00

    L0 · Advisory

    Analyses the work and advises. Changes no code.

  2. Level 01

    L1 · Supervised

    A named person approves the plan before any code is written.

  3. Level 02

    L2 · Autonomous

    Implements on its own. A person still reviews the pull request.

No AI employee runs above its project’s level, each AI employee can be capped lower, and approvals can be required for specific actions across the organisation.

Approvals, audit and the switch

A named person decides. Every decision is recorded. One switch stops it all.

  • One gateway

    Every external action goes through one gateway: policy, approval, credential, log.

  • Policy and approvals

    Policies per project and action. Where a policy asks for approval, a named person decides and the decision is logged.

  • Plans and approvals in Slack

    The Slack conversation layer: plans are posted and approved in Slack.

  • Append-only audit

    Every decision, approval, tool call and model call is recorded in an append-only audit log.

  • Pause all AI work

    One switch pauses all AI work for everyone, and a banner shows it.

Governance and safety rules

Book a demo

See it on your own workflow.

A 30-minute call: about 15 minutes of live product, including the pause-all switch, then your questions. Bring your CISO if you like.

What we show

  • Roles that can’t overreach, with step-up for sensitive actions
  • A model provider locked to a data class
  • Approvals by a named person, in Slack or the panel
  • The pause-all switch

Email polis@polis.guru