Skip to main content
IMP brings together three capabilities that usually require separate tools: a chat interface your whole organization uses, a build environment where specialists configure agents and knowledge, and governance controls that keep AI use safe, auditable, and within budget. Everything runs on infrastructure you control.
IMP chat interface showing an active session with an agent

The four surfaces

Chat

Where end users talk to agents. Start a session, upload files, continue past conversations, organize work in projects, and transcribe meetings. No configuration required — you open it and start typing.

Building

Where power users create what everyone else uses. Configure agents, build knowledge contexts, write prompts and skills, define routines, and run evals.

Administration

Where platform owners manage the deployment. Invite users, assign roles and teams, set budgets, choose which models are available, configure theming, and review usage analytics.

API and developer tools

Where developers integrate IMP into their own products. The @exulu/backend npm package, a GraphQL API, a REST API, and an interactive API explorer are all here.

The command palette

Wherever you are in IMP, press +K (or Ctrl+K) to open the command palette. It lists your recent pages, every destination you can navigate to, and create actions like “New agent” or “New project” — type to filter, select a result, and IMP takes you there immediately. The palette only shows what your permissions allow.

Deployment models

IMP runs either as a hosted service provided by Exulu or as a self-hosted deployment on your own servers and cloud infrastructure. Both use the same codebase. The self-hosted path gives you full control over where data is stored and processed — a key reason organizations choose IMP.
If you are evaluating which deployment model fits your organization, see the self-hosting architecture page for the infrastructure requirements and data-residency guarantees.