Relevant
Hold only the picture that can change the decision—not everything that can be collected.
Pre-launch · private development
AI can do the task.
It still loses the situation.
Today’s AI can answer a request and still miss the changed fact, conflict, or decision that makes its answer wrong. Henosis is developing systems designed to hold the relevant picture together before action, recognize when that picture changes, and stop where a person still needs to decide.
DifferenceSituation integrity
The difference is not a smarter answer. It is whether the system catches when the answer no longer fits.
A meaningful change should change—or stop—what AI does next. Henosis calls that target situation integrity: a current, coherent, authorized picture of what changed, what conflicts, and who still decides. The pieces exist. Making them work together as one dependable condition before action remains an open problem.
Scenario 01A familiar failure
Every task was completed.
The situation was lost.
A launch moves. The calendar changes while the announcement, checklist, and approvals do not. Every task can look complete while the plan becomes wrong. This is a conceptual target, not a product result.
The launch moved
A decision changes the launch from Tuesday to Friday.
The tools kept going
The calendar updates, while the announcement and checklist still follow the old plan.
The conflict surfaced
Legal has not cleared the change, and no one approved the new plan.
The next action should stop
The change should reach affected work and return the decision to a person.
StandardWhat must remain true
The relevant picture, under human control.
Authorized
Use only the context a person deliberately connects, within limits they can see and change.
Accountable
Surface what changed, what remains uncertain, and which decisions still belong to a person.
JoinAt the beginning
The right relationship starts with a specific reason to talk.
Henosis is meeting builders, prospective evaluators, researchers, and experienced operators who may fit a future defined opportunity. An introduction is not a request for unpaid work.
01
Future build roles
For builders and product thinkers who want to be considered when a defined opportunity opens.
Explore fit02Future evaluation interest
For operators with a recurring, high-context workflow who may fit a later private evaluation.
Explore fit03Research exchange
For people working on memory, evaluation, human agency, cognition, or AI governance.
Explore fit04Strategic introductions
For experienced operators, advisors, and ecosystem partners who could strengthen the initial proof.
Explore fitFounderMakaia Childress
Turning a hard AI problem into something that can be tested.
Makaia Childress is building Henosis around one practical question: can AI recognize when the situation has changed before it acts? The work spans research, product, software, and systems, with an initial proof being developed privately.
UpdatesRegister interest
Register your interest in updates.
Leave an email if you want Henosis to ask you to confirm when occasional research, product, and public progress updates are ready. No update is sent now.
Henosis ResearchThe governing principle