Read the intelligence ring
The center shows the overall team quality estimate. The surrounding segments show five distinct dimensions, each with its own percentage and color. Confidence is displayed separately because a precise-looking score should not hide limited evidence.
The five dimensions
| Dimension | What it asks | Useful action |
|---|---|---|
| Member fit | Do individual members bring relevant, evidenced capability? | Inspect role and specialist fit |
| Coverage | Does the team cover the capabilities represented in its composition? | Add or replace a missing capability |
| Availability | Can the team realistically take on the work? | Confirm availability or rebalance workload |
| Collaboration | Is there evidence that this group can work together effectively? | Review prior collaboration and Team Pulse |
| Resilience | Can work continue if one person becomes unavailable? | Reduce single-person dependencies |
Composition quality and project fit are different analyses
| Mode | Question | Typical recommendation |
|---|---|---|
| Team composition | Is this group balanced, available, compatible and resilient as a team? | Clarify overlapping ownership, confirm availability or reduce dependency on one person |
| Project fit | Does this team cover the roles and evidence required by a specific project brief? | Add or replace a project capability that is not covered strongly enough |
| Member detail | What evidence, workload and collaboration signals support this person's place in the current team? | Review availability, role fit or a contextual replacement |
Open a specialist to understand the score
- Member fit summarizes evidenced role capability, not profile completeness alone.
- Availability combines current status with accepted and pending workload where that evidence exists.
- Collaboration uses real shared-work and private collaboration signals with confidence and recency.
- Relationship lines explain capability or working links without changing the saved team.
- A recommendation is contextual; it does not become a permanent public judgment about the person.
Use Replacement Bench only when evidence supports it
- 1Choose the target specialist
The target role and current composition define what a replacement must preserve or improve.
- 2Review role-specific alternatives
AI-assembled teams use grounded alternatives prepared for that exact role and request.
- 3Check project context
A project replacement also uses the project title, role description and available brief context.
- 4Prefer no result to an unrelated result
Strict replacement search requires semantic evidence. If no reliable option meets the threshold, the Bench remains empty.
- 5Confirm participation
Selecting a replacement updates the project workflow; it does not bypass the required participant or owner confirmation state.
Examples of useful interpretation
| What you see | What it may mean | Useful next step |
|---|---|---|
| High score, low confidence | The visible composition looks plausible but historical evidence is sparse | Confirm availability and review profiles rather than trusting the percentage alone |
| Strong expertise, weak resilience | Important capability is concentrated in one person | Add a backup owner or document the handoff |
| Good member fit, weak availability | The right people may not have realistic capacity now | Adjust the roster or the project start |
| Strong availability, weak coverage | The team has time but not all required capability | Use a project-specific replacement or add one complementary role |
| Weak collaboration with low evidence | The group may simply be new | Agree communication and decision rules, then collect real project evidence |