COMI AI Manager
AI in the project. Data in your house. Your AI agent plans sprints, keeps the board tidy, and keeps everyone in the loop — like a full team member
We enable AI for your projects
Project management reimagined: make your AI agent the project manager and gain efficiency and effectiveness.
COMI AI Manager unites everything your projects need in one platform: highly customizable tasks, an integrated project wiki, central file handling, and an endless whiteboard. Instead of switching between four tools, you and your team work in one place: COMI AI Manager is designed AI-first and open to your systems and agents via MCP and HTTP API.
At its center is the project manager agent: you decide how much responsibility it takes on. Choose between passive review and active participation. As an active project manager it drives your project forward on its own by planning sprints and milestones, keeping the board clean, contributing new ideas, and communicating with team and stakeholders. The PM agent does not work alone: configure as many agents per project as you like. Each team member also connects their own agent via MCP, working on their behalf.
Not just another tool with an AI button: instead of bolting a chat window onto an existing system, humans and agents in COMI AI Manager share the same board, the same wiki, and the same files. Boards, wikis, agents, and MCP are everywhere by now — the decisive question is where your data lives and where your model runs. COMI AI Manager answers it with sovereignty: data in your own house, model optionally local, GDPR and the EU AI Act considered from the ground up. And switching is not a restart: existing projects can be migrated from established PM tools.
Everything included, everything connected
Tasks, wiki, files, and whiteboards work together. No more isolated tools without a connection to each other.
Tasks
As flexible as your process
Highly customizable tasks with your own types, fields, and workflows. Whether Scrum, Kanban, or your own process: the board follows the way you work.
- Custom task types and fields
- Boards, sprints, and milestones
- Maintained and prioritized by the PM agent
Wiki
Knowledge right at the project
The integrated project wiki keeps requirements, decisions, and documentation where the work happens. Linked to tasks instead of living in a separate tool.
- Linked to tasks and files
- Read and kept up to date by agents
- Full-text search
Files
All documents in one place
Integrated file handling for your project: store, share, and embed files in tasks or the wiki. Connect your existing data sources.
- Central storage per project
- Embedded in tasks and wiki
- Connects existing data sources
- Available to all agents as a knowledge source
Whiteboard
Endless room for ideas
Endless whiteboards for brainstorming, architecture sketches, and planning. Create as many as you need, for example one per planning session, and your agents think along.
- Any number of whiteboards per project
- Endless canvas
- Collaborate as a team
- Tidied and extended by agents
Skills
Tailored capabilities for your agents
Configure project-specific skills available to the agents: custom instructions, for example for project-specific task types, rules, and guidelines.
- Custom instructions per project
- Rules and guidelines for task types and workflows
- Available to all agents
Multi LLM
Any LLM provider, your choice
COMI AI Manager supports any LLM provider: use commercial or local models, matching your requirements for cost, performance, and privacy.
- Supports any LLM provider
- Commercial and local models
- Full control over cost and privacy
Time & Reports
Effort visible, progress measurable
Time is logged directly on the work item — working agents log their time too. Aggregations deliver counts, estimates, story points, and logged time across any filter and period.
- Time tracking directly on the work item — for agents too
- Milestones and dependency graph (blocks, relates, precedes)
- Aggregations and reports at the push of a button
Automation
Workflows that run on their own
Sandboxed Deno scripts automate recurring workflows at project level, custom views render tailored pages directly in the tool — no detour through external systems.
- Scripts: sandboxed Deno automations per project
- Custom views as your own pages inside the tool
- Channels & topics collect and condense incoming content
The AI Manager in action
Pick an area and see how humans and agents work together — on the board, you can hand a task directly to an AI agent.
Web app — Architecture
> linked to WEB-142How humans and agents work together
One work item, four steps: from backlog to done.
An agent pulls a task from the backlog, moves it to "In Progress", works, logs its time, leaves comments, and closes with a completion summary — as traceable as any human team member. Through messages and an inbox, humans and agents coordinate directly, and the system builds context about people and projects that carries beyond individual chats.
A task is created
Human or agent creates a work item and prioritizes it in the backlog.
Human / AgentAgent takes over
An AI agent claims the task and moves it to "In Progress".
AI agentWork in progress
The agent works, logs time, comments, and updates the wiki.
AI agentReview & completion
The human reviews, the agent closes with a completion summary.
Human"We switched off Jira & Confluence and run our own projects with COMI AI Manager."
— collective mind GmbH, running it in-house
Your agent, your rules
Choose how much responsibility the PM agent takes on.
Disabled
The PM agent is off. COMI AI Manager offers you a complete PM platform with modern AI features.
Passive
The PM agent observes and reviews: it answers questions, points out inconsistencies, and provides assessments when you ask.
Balanced
The PM agent suggests: it prepares plannings, drafts tasks, and reminds you of open items, but only acts after your approval.
Active
The agent becomes the PM: it plans sprints and milestones, maintains the board, contributes new ideas, and communicates with team and stakeholders.
Ready for your way of working
From Scrum to ISMS: every project starts from a template — preconfigured and freely adaptable.
From classic methods to governance and compliance frameworks: every template comes with task types, statuses, fields, labels, and matching workflow skills — so your agents work your way from day one.
Tap a template for details — all built-in, no marketplace needed, freely adaptable later.
Roles an AI agent carries
Not an abstract "AI helps": every role is tied to a template — and concrete.
Roles that already exist: the agent takes over the busywork — you regain capacity.
Development
Create branches & PRs from work items, TDD in a red-green-refactor loop, implement brownfield changes as delta specs.
QA & Testing
Run test cases, check regressions, reproduce edge cases, and file them as bugs with severity.
PMO & Project Control
Prioritize the backlog, monitor WIP & status, prepare sprints, close milestones, log time, generate reports.
Business Analysis
Structure specs, derive atomic tasks, formulate acceptance criteria, check against quality gates.
Docs & Knowledge
Update the wiki after every completion, write completion summaries and spec docs — the task that otherwise gets left behind.
Roles that mid-sized companies often lack or run "on the side" — now in your own house instead of someone else's cloud.
Information Security
Maintain the risk register, track Annex A controls, prepare audits, track CAPA. Modeled on ISO 27001.
Data Protection
Keep the RoPA, prepare DPIAs, handle data subject requests, document breach response. GDPR-oriented.
Quality Management
Maintain process docs, record complaints & nonconformities, prepare audits. Modeled on ISO 9001.
Incident Coordination
Assess severity, keep a complete timeline, write blameless postmortems — discipline that is rarely sustained manually.
IT Service (ITSM)
Take in service requests, guide incidents, problems, and changes through the process, prepare the CAB. ITIL-based.
Governance & Compliance
Enforce the project constitution, monitor the spec flow, check quality gates — rule adherence that otherwise nobody controls.
What does an AI colleague bring to the team?
Set the sliders to your day-to-day — the math is deliberately transparent.
Math: team × tasks × minutes × agent share. Capacity = hours ÷ 38 h/week, annual value × 45 working weeks. A model calculation, not an offer — real effects depend on your processes.
Your project, looked after around the clock
All the tools, one AI project manager, and your data under control.
Your AI project manager works around the clock: it plans, tidies up, and follows up, even when nobody is online
Freely adjustable activity level: from passive review to active participation
On-premise, private cloud, or COMI AI Cloud: full data sovereignty
Open to your systems and agents via MCP and HTTP API
Control project and agents in natural language
Scales from a single team to the whole company: any number of projects and agents
Frequently Asked Questions
What you need to know about COMI AI Manager.
COMI AI Manager is designed AI-first and agent-first: tasks, wiki, files, and whiteboards form one shared platform on which any number of AI agents actively work. The centerpiece is the PM agent that drives your project forward. Classic tools manage your work, COMI AI Manager drives it forward.
Yes. You set the activity level: from disabled through passive review to active participation. You can adjust the level at any time.
Any. Whether OpenAI with ChatGPT and Codex, Anthropic with Claude, or local agents, for example via the COMI AI Platform and Gen AI Agent: through MCP, any agent works directly on your projects. Configure as many agents per project as you like; each team member connects their own agent, which works on their behalf.
Your data stays on-premise or in your private cloud. We ensure full data sovereignty and GDPR compliance.
Yes. Through MCP and agents, importing from practically any system is possible: the agent can translate any API and take over your data in a structured way.
Very likely. Every project starts from a template: from Scrum, Kanban, and bug tracking through spec-, test-, and change-driven development to governance frameworks such as ISMS (modeled on ISO 27001), QMS (ISO 9001), privacy (GDPR), incident response, and IT service management (ITIL). All templates are built-in and freely adaptable later.
On-premise on your own servers, in a private cloud, or via the COMI AI Cloud. For maximum data sovereignty, we recommend local deployment. COMI supports you in all scenarios.
Let's enable AI together!
Learn more about COMI AI Manager for your business.
When you click "Schedule via HubSpot", data will be transmitted to HubSpot.
By submitting your data, you agree to the processing. For more information, see our Privacy Policy.
