What it notices.
A metric drifting, a customer message, a competitor announcement, a deadline approaching. The exocorp catches things — most of them won't earn action, and that's the point. Noticing isn't the same as reacting.
Attention
The exocorp has to decide what’s worth doing before it does anything.
Most AI products give you a way to start a task. An exocorp also handles what comes before that — figuring out which tasks are worth starting. Inside, the exocorp tracks four things at any given moment. Tasks come from these, not the other way around.
What it believes.
The exocorp's working picture of the world — dated, sourced, marked as belief rather than fact. When evidence stops supporting a belief, the belief is demoted, replaced, or retired. The exocorp knows what it used to think.
What it's trying.
Bets the exocorp is actively running. Each one has a cost, a way it could be proven wrong, and a finite life. The point of a bet isn't to be right — it's to find out cheaply, on purpose.
What it owes.
Every promise the exocorp has made — to a customer, a partner, another department — with a way it can close: fulfillment, renegotiation, repair, retirement. Owed work doesn't disappear when nobody's looking at it.
Tasks are downstream. Once the exocorp has decided something is real, a team owns it — with a clear trail back to the signal that started it.
Six stages
The same news means different things at different stages.
An exocorp moves between six stages of work. A competitor launch is something worth noticing in discovery, an experiment to design in validation, a promise to renew in operations, evidence to pivot in reshaping. Where the exocorp is determines what the news means — which evidence weighs, how long the decision gets, which parts of the exocorp wake up.
Formation
Setting the exocorp up. What it's for. What it owes. Who's responsible for what. The shape it starts with.
Discovery
Watching the world. Letting signals come in. Forming a view on what's worth doing before committing to do it.
Validation
Designing the smallest experiment that could prove a bet wrong. The work is being able to lose, cleanly and on purpose.
Execution
Building the thing. Real workflows, real attempts, real evidence of what changed and what didn't. Every task traces back to a reason it exists.
Operations
Keeping promises met and conditions held over time. Customers served. Skills safe. Costs in line. Repetitive manual work is treated as a design problem, not as diligence.
Reshaping
Changing the exocorp itself. Retiring mandates. Folding departments. Updating what the exocorp believes about its own setup. The reshape is in the workflows, not just the slides.
The kind of agent that’s right early on isn’t the kind that’s right later. Stages shape who’s working and how they think.
Work, end to end
How work starts, runs, and closes.
Work doesn’t appear out of thin air. Something concrete starts it — and it leaves a trail you can follow at every step, from arrival to result.
01
Something happens.
A customer replied. A metric crossed a line. A deadline arrived. A prior promise came due. Whatever it is, it's small and concrete — and it carries why the exocorp is paying attention now.
02
A work item gets owned.
The exocorp decides this is real and a team picks it up. The item carries the source that started it, the team that owns it, the result the exocorp is going for, and where it stands today. Repeated signals update the same item — they don't multiply tasks.
03
A team takes a run at it.
A run is one bounded attempt — researching, drafting, reviewing, delivering. It ends honestly: success, failure, waiting on someone, or escalating. Either way, it leaves evidence about what moved.
04
It closes.
The item ends — fulfilled, breached, expired, canceled. Memory updates. Affected promises close. The trail stays, so the next time something similar shows up, the exocorp recognizes it.
Periodic checks keep an eye on conditions that need to stay true. They’re separate from new work arriving.
Changing direction
Pivoting means changing what’s running, not what you’ve announced.
Any company can declare a new direction while its old workflows, mandates, prompts, and metrics keep running things underneath. That isn’t a pivot — that’s an announcement. An exocorp operates on the things actually doing work, so a change in direction shows up where it has to: in the machinery.
What it remembers.
The exocorp's knowledge doesn't sit still. Claims get demoted when their evidence weakens. Contradictions get flagged for review. What the exocorp knew six months ago doesn't have to be what it acts on today.
What each department owns.
A mandate isn't permanent. It can be narrowed when scope was too wide, paused when a bet stalled, replaced, demoted, or retired. The change is real — the workflows actually stop running, not just the description.
What it owes.
Every promise has a way it can close: fulfillment, renegotiation, breach repair, explicit retirement. A breached promise is an event the exocorp has to deal with, not a line that quietly disappears.
What it can do.
New capabilities get tried in labs and promoted into the org once they're proven. Old ones get deprecated when they stop being useful. The skill mix evolves; the org chart isn't fixed.
Reshaping isn’t a phase the exocorp enters once. It’s something the exocorp can do at any time, in any size.
Failure modes
How an exocorp can fail.
The obvious failure is silence — the exocorp stops, you notice, you fix it. The harder failure is the opposite: the exocorp stays busy, ships work, closes tickets, hits its numbers, while solving the wrong problem. The warning signs are counterintuitive.
More work, weaker reasons.
The number of tasks goes up, but the reason each one exists gets thinner. New items trace back to "we should probably do this" instead of to something concrete in the world.
More agreement, less argument.
Departments writing to each other, beliefs not getting tested. Agreement looks like progress, but can mean the exocorp has stopped checking its own assumptions.
More testing, less reality.
Internal evaluations grow; the gap between what they measure and what customers actually care about grows with them. An exocorp can pass its own tests and still lose.
More roles, less variety.
New roles and new agents, all inheriting the same memory and the same playbook. More agents doing the same kind of thinking, not different thinking.
Being busy isn’t the same as being on track. The exocorp is built to notice the difference — and bring it up.
Your role
What only you decide.
What the exocorp is for. Which markets to enter, which to leave. When to keep going and when to fold. Whether a deal is the right deal. Whether the work is on-brand. Whether a particular tradeoff is the one to make for now. Those calls are yours. The exocorp brings them up when they come up, and waits.
Past those, the exocorp runs on its own. Departments take work. Memory updates. Promises move. When the exocorp sees a gap in its own org chart, it proposes a role and waits for your sign-off. When a mandate stops earning its place, the CEO surfaces it. When operations pressure builds, you hear about it before customers do.
Most of what you used to do at a normal company doesn’t apply. The alignment meetings, the OKR check-ins, the inbox triage, the staffing fights. The coordination still happens — inside the exocorp, on its own schedule. The trail is yours to read whenever you want it.
Build
Ready to run one?
Exocorp is in early access. The first ones are running now.