Independent demo, not an official EU site. AI summaries can be wrong; data from 5 Oct 2026. Disclaimer

Run trace Institute for European Environmental Policy

brief stage, up to the human go / no-go started 5 Oct 2026, 23:33. The run then paused for a person to say go or no-go per consultation; what happened after that stays in the local app.

What each agent did

The matches below are the ones this run saw when it ran. The organisation page recomputes them live, so the two lists can differ after new consultations open or old ones close.

#AgentToolInputOutputCallsTokens in / outCacheTime
1 supervisor route escalations profiler
2 profiler register+memory 934329423960-72 Institute for European Environmental Policy (Think tanks and research institutions), 0 past submissions, 0 reviewer edits 13 ms
3 supervisor route profile, escalations, next matcher
4 matcher pgvector 0 corrections applied; rerank: 15 cached, 0 calls 5 matches: 25794:0.63, 26097:0.52, 26171:0.55, 25801:0.57, 24198:0.59 hit 25 ms
5 supervisor route profile, matches, escalations, next landscape
6 landscape feedback_analyses cache 5 publications 25794: 5 analysed (3 wants changes); 26097: 1 analysed (1 wants changes); 26171: 0 analysed; 25801: 3 analysed; 24198: 41 analysed (1 opposes, 2 supports, 29 wants changes) hit 31 ms
7 supervisor route profile, matches, landscape, escalations, next lobby
8 lobby SQL+publication_meetings 5 publications 25794: 30 meetings; 26097: 30 meetings; 26171: 30 meetings; 25801: 30 meetings; 24198: 30 meetings hit 34 ms
9 supervisor route profile, matches, landscape, lobby, escalations, next reader
10 reader publication_summaries+chunks 5 publications, profile vector 5/5 summaries from cache; 25794: 8 chunks; 26097: 8 chunks; 26171: 8 chunks; 25801: 0 chunks; 24198: 8 chunks hit 37 ms
11 supervisor route profile, matches, landscape, lobby, reading, escalations, next go_no_go
12 go_no_go interrupt waiting for the reviewer run paused in the checkpointer 16 ms
12 node executions00 / 04 hits140 ms

The supervisor routes on what the run already holds; it calls no model. Summaries and match judgments came from caches shared by every organisation, which is why most steps take milliseconds.

Matches