Finance
In development
An agent that maintains financial context and helps turn records into net-worth views, cash-flow tracking, and reviewable reports.
Private alpha, running real work
Your agents need a home base.
Kapelle is a home for the agents you use every day. Give each one a lasting role, keep its work close, and stay in control of what happens next.
Persistent agents Durable outputs Human control
Daily desk
Good morning, Chris.
4TodayThe work you intend to move
3Review nextFresh work from your agents
2Needs a responseOnly the decisions that need you
1Follow-throughActions waiting on an outcome
Illustrative workspaceSample data
Websites. Apps. Agents.
The next application has an agent inside.
First we visited websites. Then we opened apps. Now we can work with agents that remember their responsibilities and keep useful work moving. Kapelle brings those agents into a workspace centered on the person using them.
Fantasy commissioner
In development
An agent that understands a league’s rules, reconciles weekly changes, and prepares an announcement for the commissioner to review.
A private lab workspace
Proposed pilot
An agent that works with an approved document collection and produces source-linked answers, drafts, and recurring reports.
The coordination problem
Chat is where you ask. Kapelle is where the work lives.
A useful agent becomes something you return to: its records, its work, and the decisions waiting for you. As you add agents, you need one place to see what changed and decide what happens next.
Kapelle turns scattered agent activity into one legible loop.
- 01
Capture
Tasks, messages, files, and recurring work enter.
- 02
Dispatch
The right agent receives context and ownership.
- 03
Review
Useful output arrives in a durable place.
- 04
Follow through
Your response becomes the next action and receipt.
One operating layer
The value lives above the model.
Models will keep changing. Kapelle keeps the ownership, context, state, review, routing, and receipts that make them useful together over time.
Agents with a lasting role
Build an ongoing relationship with an agent that knows its job, keeps its records, and produces work you can return to.
Work that stays put
Reports, plans, drafts, and decisions land as durable artifacts instead of disappearing into a chat transcript.
A human review loop
See what matters, respond with context, and route the next action without reconstructing the whole conversation.
Recurring operations
Run reports, monitoring, research, and maintenance on a schedule without writing the same prompt again.
Choose where work runs
Run the workspace on infrastructure you control. Fully private workflows also need local models, tools, storage, and a verified data boundary.
Model flexible
Use the right model for each job. Kapelle is the durable operating structure around the models, not another model lock-in.
Proof through use
Built inside a real, mixed, messy life.
Kapelle is being shaped by lived use across community work, nonprofits, household operations, software, and research. It is not a hypothetical workflow.
Community
Cleveland Park
Research, newsletters, events, nonprofit formation, and board follow-through.
Nonprofit
Trinity
Board work, documents, fundraising, and affordable-housing planning.
Household
Finance
Recurring analysis, reconciliation, reporting, and durable financial context.
Product
Kapelle itself
Strategy, research, roadmaps, builds, verification, and operational history.
Current direction
Useful agents. Work you can return to.
We are building around lasting agent applications: a useful daily workspace, finance records, league operations, and a careful path to private team workflows.
Exploring private AI
Your workspace.
Your model.
Your data boundary.
Some work needs an AI system dedicated to one person, team, or lab. We are exploring how Kapelle can bring the same agent workspace to a model running on your own machine, inside your organization, or on dedicated hosted hardware.
The starting point is one useful workflow: approved documents, a private model, and an output someone can inspect and use.
This is a pilot direction. A local workspace alone does not keep every connected model or tool private; each deployment needs its full data path verified.
Discuss a private AI pilotPrivate alpha
Bring real work. Give your agents somewhere to do it.
Kapelle is for people who already feel AI expanding what they can do and want a dependable operating surface for the agents around them.
Tell us what you are running today, where coordination breaks, or which agent you wish already existed.