Marketplace / meeting-prep-extractor-memory

Meeting Prep — Extractor memory

Surfaces per-attendee relationship memory (recent commitments, follow-ups, notes) from the Knowledge Extractor's ledger into pre-meeting briefs. Read-only.

by Agentum 1 installs v1.0.4 integrations
Open in Agentum
Requires the Agentum app

Meeting Prep — Extractor memory

Surfaces per-attendee relationship memory — recent commitments, follow-ups, and notes — from the Knowledge Extractor into the briefs Meeting Prep generates before each meeting. Read-only: it never modifies your extractor data.

What it does

When Meeting Prep assembles a brief for an upcoming meeting, this add-on looks up each attendee in the Knowledge Extractor’s ledger and pulls back their most recent facts. Those facts become a Relationship memory section in the brief — a short list per attendee covering things you’ve agreed to, asked about, or noted in past interactions.

The goal: when you sit down with someone you’ve met before, you don’t have to remember what was outstanding from last time. The brief reminds you.

This add-on is read-only. It never adds, edits, or removes anything from the extractor — it only reads.

It also degrades quietly:

  • If the Knowledge Extractor isn’t installed, briefs are unchanged.
  • If an attendee has no facts yet, that attendee is skipped.
  • If a single attendee’s lookup fails, the rest of the brief still renders.
  • If no attendee has facts, the Relationship memory section is omitted entirely.

Settings

Setting What it does Default
Facts per attendee The maximum number of recent extractor facts to include per attendee in each brief. Raise it for more context, lower it for a tighter summary. 5

Integrations

  • Knowledge Extractor — required. This add-on reads from the extractor’s ledger. Without it installed, the Relationship memory section will not appear (your briefs are otherwise unaffected).
  • Meeting Prep — required. This add-on plugs into the brief assembly Meeting Prep already runs. Without Meeting Prep, this add-on has nothing to plug into.

Known limits

  • Self-attendees are skipped. The calendar owner’s own entry on the attendee list is filtered out — you won’t see your own facts in the brief.
  • Email is the match key. Attendees are matched to the extractor’s ledger by email address. If the same person shows up under two different addresses, their facts won’t merge.
  • Recency, not relevance. Facts are pulled by recency, capped at the Facts per attendee limit. There’s no per-meeting topical filtering — the add-on doesn’t try to guess which past notes matter most for this specific meeting.
  • No write-back. Nothing in the brief feeds anything back into the extractor. This add-on is one-way memory readout.
  • Limit is a ceiling. If the extractor only has 2 facts for an attendee, you’ll see 2 even if the limit is 5.