For a venue that has its submissions but no reviewers yet. The planner reads the submissions and proposes whom to invite from the expert database, as reviewers and as area chairs, so that every paper is covered within everyone's load.

1. 1 Reviewers: how many per paper, how many papers each at most, and the share of invitations you expect to be accepted. With 3 per paper and 60% acceptance the planner aims at 5 invitations per paper.
2. 2 Area chairs are planned the same way after the reviewers, and never overlap with them. Put 0 per paper to skip them.
3. 3 The bar of each role is shown under its numbers; change opens the Recruitment bar.
4. 4 Plan the pool. It takes under a minute. Tick Plan from scratch to ignore the people already in the pool; otherwise they count toward the coverage and the plan only completes the pool. Leave Name the topics from LLM keywords ticked to have the topics below named by an LLM, which sends the submissions' titles and abstracts to it; untick it to keep everything on this server.
5. 5 The result: the number of invitations, how many are expected to accept, the invitations a paper gets, how many of the places could be filled, and the papers the plan could not cover.
6. 6 Balance across topics. The submissions are grouped into topics of similar papers; every topic should have its places planned (100%). A topic is named by its research area (in small capitals) and the keywords that set it apart, extracted from its papers by the LLM in the way the expertise database was built.

1. 7 Everyone is ticked to begin with; untick the people you do not want. in pool and listed mark those you already have.
2. 8 The papers they would take show why each person is on the list. Fit is a percentile against the whole database: 99 is the top 1%.
3. 9 Shortlist the ticked people to follow the invitations on the Recruits page: mark them invited, then accepted (they join the pool) or declined.
4. 10 Add to the pool directly when you do not need to track invitations.

In the demonstration the plan invited 390 reviewers and 36 area chairs. To show what happens when fewer accept than planned for, half of the reviewers joined (the acceptances are simulated; nobody was contacted).