01 · TASK
What needs to be accomplished?
Describe the goal, current problem, work stages and evidence of outcomes. Distinguish current context, AI plans and role commitments.
OUR METHOD
Evidence for decisions. Questions for uncertainty.
What work does a role require? What has the candidate actually done? Can both sides take the next step together? These questions organize our materials and conversations.
Four questions that bring resumes and JDs back to the work.
02 · CONTRIBUTION
What did the person decide and do?
Distinguish the person, team, agent and existing system. Results, process and validation complement one another; tool familiarity alone does not establish capability.
03 · CONDITIONS
Is there a genuine hard conflict?
Check compensation basis, location, timing, authorization and working practices. Address hard conflicts separately and investigate unknowns; strengths do not cancel hard limits.
04 · INTEREST
Why would both sides continue?
Teams explain resources and deliverable support; candidates explain goals and tradeoffs. Advisers explain their assessment, and both sides confirm participation and next steps.
EVIDENCE PRACTICE
Distinguish facts, judgments and questions.
| Layer | How to express it | Next step |
|---|---|---|
| Personal statement | Record who said what and when it applies. | Review with the person and source key facts. |
| Checkable evidence | Explain artifacts, contribution, validation and limits. | Review permitted materials or work samples as needed. |
| Case judgment | State the reasons, assumptions and uncertainty. | Seek confirmation from the parties or interviews. |
| Important unknown | Explain what is missing and why it matters. | Identify who can answer and what to ask. |
Use feedback for this decision without creating a permanent label.
Interviews can reveal new information and change role conditions or personal preferences. Keep feedback sources and scope; revise the role, pause introductions or close the mandate when appropriate. Silence does not establish a lack of ability.
PUBLIC RESEARCH
AI Work Standards Observatory
The team maintains sourced work-standard drafts and external result snapshots to support discussion of tasks, human–agent work and contribution. The standards remain research, not talent certification or proven optimal allocation.
START WITH ONE CLEAR GOAL
Bring one role. Talk about the work it needs to do.
GIL coordinates initial contact. Scope and case responsibilities are agreed before starting.