image_The-workplace

The workplace

Aipokit is the workplace a company — or a team inside one — actually runs. Files you own. People and agents on the same work. Underneath, four pillars — place, work, doors, and policy. A week in the firm →

Workspace and folders
Place

Workspaces and folders you own

Knowledge is organised as folders with types — courses, sites, media, processes, apps, plain files. Agents and people work on the same paths. Multi-app folders share a data plane so management and audit views, for example, read the same CSVs without private copies.

1 shared file substrate
N folder types & apps
Authoring in a Studio
Editors

Author over your files — no lock-in.

Site, course, presentation, diagrams, forms, and text. Focused editors on the same workspace files. People write there; the operations underneath are the same ones Ask and agents you bring use.

Doors: Ask (in-product AI) and MCP for external agents
Doors

Ask in product. MCP outside. Same policy.

Ask is the AI assistant built into Aipokit — chat and act on any folder type when you are in that place, and manage processes (list, start, complete, open). MCP lets Claude, Hermes, OpenClaw, or a custom loop use the same tools with risk tiers and step-up approvals. External agents do not receive a private back door.

Core · Ask (in-product AI) + Workspace MCP
Process flow metaphor
Workforce · process engine

Model it in BPMN, then run it

Draw a process in BPMN Studio, deploy it, and run supervised workflows where agents and people share tasks. Live instances, human approvals for consequential writes, and a run history you can audit. Process results can land on the folder data plane for multi-app views.

Knowledge graph
Knowledge & memory

A memory graph over your workspace

Aipokit indexes the workspace into a connected graph — search, relationships, and context that people and agents share. Prefer knowledge links over bulk static assets; the graph is the memory layer on infrastructure you control.

Published output
Publishing

From source to live output

Sites, courses, and presentations publish as static, SEO-safe output. Federate across a private authoring node and public delivery nodes. This site is built by the Aipokit site generator from workspace files.

Policy and governance
Policy & self-hosting

Your infrastructure, your boundary

Per-account isolation, access control, and residency choices. Aipokit supports AI Act and GDPR obligations with human oversight, logging, and exportable files — obligations remain with the deploying organisation. Self-host or host with us. Export is part of the product, not a promise after the fact.

DB per-account isolation
1+N authoring + delivery
Architecture

Lean core, rich surfaces.

01
Truthful core
One action = one service function, exposed to UI, REST, Ask (in-product AI), and MCP. No fake success, no private back doors.
02
Studios on top
Human authoring in Studios that consume the same contracts agents use — clear shell over a truthful core.
03
Verticals on the platform
Courses, sites, and ops planning are product lines and verticals — they do not redefine the platform.

See the platform on your work.

A pilot partnership walkthrough — place, process, Ask, and policy on your work. Seminars are one way in; the product is operations.

↗ selective pilot partnerships