Talk to your projects
Ace is the AI assistant inside WORKSPACE.PM: it knows your real project data, answers questions in seconds and gets things done - always within your own permissions and only after your confirmation.
Ace
knows your project data
The knowledge is there - just not where you are right now
Between scattered projects, status questions and busywork, there's too little time for what matters.
Answers are buried deep
Who works on what, which risk escalated, how a project stands - the information exists, but finding it costs clicks and time.
Setting up projects is busywork
Every brief and every tender turns into a project profile, structure, risks and wiki by hand - every time, from scratch.
AI tools are often a black box
A foreign cloud, unclear data paths, no control over what the model does - a dealbreaker for regulated organizations.
Costs run unchecked
Usage-based AI without a cap and without transparency quickly becomes a budget risk no one sees coming.
How Ace works
Natural language in, a traceable action out - without losing control.
You ask
In plain language - right from the module you're working in, or in the full-page chat.
Ace understands your data
The assistant reads your real project data - strictly within your own permissions.
Ace proposes
Answers with clickable references, and for changes an action card with every detail - transparent, not a black box.
You confirm
Only your click triggers a write action. Nothing happens behind your back.
Stay in control
The assistant may act - but never without your yes
Every write action Ace proposes appears as a card with the exact parameters. You execute or discard. A server-side approval ledger makes sure no unconfirmed action ever slips through.
- Create, change, delete - always only after your explicit confirmation
- Every action runs with your permissions, never with more
- A server-side approval ledger prevents tampered executions
- The model's reasoning is available on demand - traceability instead of a black box
Action proposed
Move due date
Read project status, open tasks and budget - within your permissions.
Briefing_TownHall.pdf
PDF read
AI project setup
AI project setup
From a briefing PDF to a ready-to-start project
Upload a tender or brief - Ace distills a project profile and, on request, proposes structure, risks, opportunities, wiki and an announcement. All a draft to review, nothing mandatory.
- Project profile including goals, benefits and client extracted automatically
- Work breakdown structure and milestones with suggested dates
- Up to 10 risks and 10 opportunities each as a solid first draft
- Project wiki (up to 10 pages) and an announcement post for the team
- Every field re-generated individually - with your own hint to the AI
For PMO & administration
Full control over usage, cost and permissions
See tenant-wide who consumes how much, cap budgets per person and precisely control who may use the AI at all - up to a complete off switch.
- Token and credit audit per user, including estimated cost
- Monthly token limit per person - enforced server-side, not just displayed
- From a read-only assistant to an acting assistant - controllable per role
- «Generative AI» permission level: none, view or edit
- Module kill switch: disable AI tenant-wide with a single toggle
12
Active users
4.8 M
Tokens (month)
≈ €62
Estimated cost
Consumption per user
Monthly limit
On plan. 2 tasks are overdue, the budget is at 68 %.
Open directly
Read project status, open tasks and budget - within your permissions.
Suggestions
Ask, find, link
An assistant that knows your projects
Ace answers status questions, finds records and links straight to them. Context suggestions lower the barrier to start, and the history stays private and searchable.
- Context-aware suggestions per module - instead of an empty input field
- Clickable deep links: from the chat straight to the affected record
- Automatic project summary on the overview - at a glance
- Private history: rename, tidy up, condense with /compact
- Your own token statistics any time via /tokens
Your choice, your data
Your own key or managed - you decide
Connect your own AI access (bring-your-own-key) and keep full control over cost, privacy and model - down to locally hosted models. Or start without your own contract in managed mode and pay via credit.
- Claude, OpenAI, Gemini, DeepSeek, Mistral or local models
- The connection is verified before saving - no guessing model names
- Managed mode: get started instantly, billed via credit
- The model picker loads the actually available models live
- Switch, disable or disconnect the provider any time
Your own key
Full data sovereignty, your provider
Managed
Ready instantly, pay via credit
Local models possible too
MCP endpoint
https://your-tenant.workspace.pm/mcpConnected clients
Role gate: ViewClaude Desktop
B. Bolzmann
Claude
M. Lang
Claude Desktop
S. Kern
Open standard
Open WORKSPACE.PM to external AI assistants
Through an MCP endpoint, MCP-capable AI clients such as Claude connect - via OAuth as an individual person, with exactly their permissions. A coarse role gate additionally limits external access.
- Streamable-HTTP MCP endpoint for any MCP-capable client
- OAuth sign-in per person - permissions are never diluted
- Usage overview: who connects, how often, with which client
- A coarse role gate (none, view, edit) specifically for external access
AI that doesn't become a burden
The three most common worries - and how WORKSPACE.PM removes them.
No loss of control
Ace never does anything unprompted. Every change is an approval, and it's switchable off any time via the kill switch.
Your data stays yours
Your own key, local models possible, clear data paths - privacy is built in, not optional.
Costs under control
A budget cap per person, a credit gate and a consumption audit make sure nothing spirals unnoticed.
On request - or all by itself
Ace helps when you ask. Whatever repeats is handled by no-code workflows in the background - same platform, same control.
Frequently asked questions
What PMOs and project leads want to know before they start.
Connect your own access to Claude, OpenAI, Gemini, DeepSeek or Mistral - or run a local model. Without a contract of your own, use managed mode and pay via credit.
You keep control: with your own key, requests run through your provider, and local models stay entirely in house. Ace also only accesses data you're authorized for.
No. Every write action is shown as a card with all parameters and only executed after your confirmation - secured by a server-side approval ledger.
Per person you can set a monthly token limit that's enforced server-side. An audit shows consumption per user including estimated cost; in managed mode a credit gate applies on top.
Yes. The «Generative AI» permission level controls access per role (none, view, edit), and a module kill switch disables all AI features tenant-wide with a single toggle.
Try the assistant on your own projects
30 days free, no risk - with full control over permissions, data and cost.