Pilot
Run AI agents like one team.
Give agents roles and budgets, and keep their work tied to company goals. See what they are doing and what it costs at a glance.
Discuss a pilotA person sets the company goal, timeline, and what success means.
Agents propose a plan, a team shape, and a budget.
The owner sets the budget cap, approval rules, and how far agents can act on their own.
Runtimes do the work and report cost and results.
The owner reviews results and handles items that need a decision or went wrong.
Set a goal, and the agents start moving
“Build an AI quality-control company for Korean manufacturers and finish the first field test within 90 days.”
- 01Set the company and goal
Define which goal every piece of work connects to, first.
- 02Agents plan their work
They propose the work needed, based on product direction and priorities.
- 03People set the rules
People decide the goal and budget, how far agents can act alone, and which decisions always need approval.
- 04Build teams and roles
Agents are placed into the roles the work needs — research, product, field operations.
- 05Check progress and results
Track progress, cost, pending approvals, and results by goal and owner.
More agents do not make a team
Running one agent and moving many toward one goal are different jobs. When you cannot see who owns what or what it costs, there is no way to step in.
Pilot links goal, team, work, budget, approval, and result into one line. Leaders read business status and problem items, not every prompt.
See who does what, and what it costs
Who owns what?
See each agent’s role, team, and current work on an org chart.
Why does this matter?
Every piece of work links back to a company goal, so priorities have a reason.
Where does the money go?
Budgets and actual spend by team, goal, and task.
What is waiting on a decision?
Plan changes, budget increases, and other items needing a person are shown separately.
Every piece of work traces to the goal
Producing lots of work is not the result. Pilot connects where work starts, who owns it, and what it produced.
- 01company · objective · constraints
Company goal
Decide what to achieve by when, what success means, and which rules to keep.
- 02strategy · team goal · board decision
Plan and team goals
Agents propose a plan; the owner approves order and resources.
- 03owner · task · budget · approval
Work with owners
Every task carries its goal, owner, budget, and approval conditions.
- 04runtime · heartbeat · usage · result
Runtimes do the work
Runtimes execute and report progress, cost, and results.
- 05artifact · exception · intervention
Review results
Results come first; failures and over-budget items go to people.
You see results, not logs
Executives can read it
See whether work is on track without reading logs.
Results first
What was built and whether it passed review come before hours worked.
Details on demand
Team, task, and execution records are there when needed; the main screen stays clean.
Bounded self-direction
Agents work fast inside the budgets, permissions, and stop conditions you set.
Not publicly available yet
We do not present it as a finished self-serve product. We look first for organizations to try it in real situations.
- Product stage
- Pre-release (before public launch)
※ No public signup or immediate rollout is promised.
- Core elements
- Company, goal, agent, team, work, budget, approval, artifact
- Operating principle
- Every piece of work traces to a goal; costs and pending approvals are never hidden
- Relationship to runtimes
- Coordinates several agent runtimes from above
※ Pilot does not execute the work itself.
- How to adopt
- Built together with a few early organizations
Change the tools; the records stay.
Pilot does not lock you into one runtime. Execution happens in each tool; Pilot keeps the records and accountability.
Single team
Set roles, budgets, and approval procedures around one goal.
- Clear roles and goals
- Task and budget tracking
- Human approval list
Multi-team company
Manage team goals, budgets, and runtimes at the company level.
- Org chart and ownership
- Resources by team and goal
- Cross-team work and issues
Any runtime
See status and results from different runtimes in the same way.
- Runtime adapters
- Shared task and artifact model
- Stop and recover
What Pilot does and does not do
It does not rebuild tools you already use.
Chat is how work starts
Conversation is how you request and check on work. What Pilot actually manages is the goal, ownership, budget, approvals, and results.
It connects to your tools
It does not replace Jira or GitHub. It connects where needed and shows how work ties to company goals.
It does not execute
Execution belongs to each tool. Pilot checks that budget and approval standards hold.
It moves within a defined scope
Set the budgets, permissions, approvals, and stop conditions, and agents move within that range.
Tell us about the AI team you want to run.
Share your first goal, the agent roles you need, and your budget and approval rules—we will build the operating model together.