Hound
Hound studies the evidence an environment actually makes available. It recognizes what it can and preserves what remains uncertain.
Hound examines authentic events, records and relationships to recognize work within an authorized, bounded environment.
Discover electronic work. Change how it gets done.

We believe people, AI agents, software and robotics will increasingly perform connected work together—even through quiet hours. Before those contributors can work together responsibly, we must identify electronic work, then understand it.
Hound studies the evidence an environment actually makes available. It recognizes what it can and preserves what remains uncertain.
Integrated Fractal Work aims to coordinate qualified Human and non-Human contributions, with authority, judgment and verifiable outcomes.
Our conviction: the future of work is shared. Understanding electronic work is a foundation for responsible orchestration—not an optional extra.
Hound has evolved through successive research generations. Earlier bounded experiments identified electronic-work occurrences under the criteria active at the time. Its scheduled hourly research now examines authentic external evidence and investigates what it cannot yet understand.
Earlier bounded tests identified electronic-work occurrences under their active recognition criteria. That does not establish universally accurate recognition or completed business outcomes.
Hound encounters real evidence; HOME investigates uncertainties; proposed refinements are tested before release.
Authentic source · Identifiable occurrence · Traceability · Defensible meaning. These are research hypotheses, not a settled universal standard.
Read-only discovery in an authorized, bounded environment. The result reflects available electronic evidence and recognition capability—not a promise to find every activity.
Policies, documentation and interviews describe work. Electronic events may reveal only part of that activity or relationships that are difficult to connect.
Recognized electronic work, uncertain observations, source references and limits of visibility—without silently inventing missing work.
GREEN is defined before the colour categories: an observed work occurrence meets the recognition criteria active for that test; it does not itself prove completion or business outcome.
Recognized against active criteria
Possible work needing clarification
Observed; work meaning unknown
GREY supported by independent reference comparison
Hound natively reports GREEN, GREY and BLACK. VIOLET is a separate Product A interpretation, not verified GREEN. YELLOW is not a Product A classification.
• What electronic evidence the environment preserves
• What Hound can currently recognize
A future operating vision—not a measured claim of workforce savings.
A Human receives a request, checks information, routes it, follows up and intervenes repeatedly.
Qualified Humans retain judgment and accountability while suitable electronic work is assigned, executed and verified through governed capabilities—including during quiet hours.
Standards increasingly represent events, objects, agent tasks and execution traces. They can improve the evidence available for discovery, but none alone establishes that business work occurred.
Object-Centric Event Log: events, objects and their relationships.
Agent2Agent: communication and tasks across participating agents.
Observability conventions for agent and tool operations.
Standards support portions of the problem. Our recognition criteria remain open to independent challenge.
Hound is internal research; customer discovery occurs only within an authorized boundary. Fractal Work does not claim complete coverage or guaranteed ROI.