AI Knowledge & Meeting Intelligence
Sits over Google Workspace + Slack + your meeting stack
Headline outcome
Every meeting searchable; action items auto-routed
What the operation looked like before.
A 40-person professional services firm running 30+ client meetings a week across Zoom, Meet, and in-person. Notes lived in five Notion workspaces ("Acme - 2023", "Acme - working", "Acme - billing", and two by individual associate). Action items got dropped between Monday syncs and Friday status. Onboarding a new account lead meant two weeks of context-rebuilding from Slack scrollback plus Google Doc archaeology.
The bottleneck
A managing partner asked "what did we promise the COO at Acme on the Q2 expansion?" Three associates spent 40 minutes pulling threads from Notion, Slack DMs, Zoom recordings, and an email forwarded six months ago. The answer was in a Notion page nobody could remember the name of. By the time it surfaced, the client meeting was over.
How the system is wired.
Calendar webhooks pick up meetings as they end. Transcripts get chunked, embedded into Qdrant, and tagged by participant, account, and topic. A retrieval layer over Claude answers natural-language questions with citations to the exact meeting and timestamp. Slack reminders nudge action-item owners. Google OAuth gates everything by org.
Transcript Intelligence
AI-powered meeting transcription with speaker identification, sentiment analysis, and automatic topic segmentation.
Semantic Search
Qdrant vector database enables natural language search across all meeting history. Find decisions, not just keywords.
Action Items
Automatic extraction and assignment of action items with deadline tracking and Slack reminders.
Team Knowledge Base
Every meeting contributes to a living knowledge base. RAG ensures context is never lost across team changes.
Where humans approve, review, decide.
Every consequential decision routes through a person on your team. Speed without abandoning judgement.
Sample approval surface
Approves auto-extracted action items before they're assigned
Flagged- Decision frequency: Per meeting
- Routes to the operator daily review queue
Account lead
| Role | Decision | Frequency |
|---|---|---|
| Account lead | Approves auto-extracted action items before they're assigned | Per meeting |
| Account lead | Confirms decisions before they're indexed as canonical | Per meeting |
| Ops admin | Curates which meetings stay private vs go to the team index | Weekly |
What changed for the operation.
3
Outcomes the operating team validated. Numbers below are anonymized to client policy.
Meeting context lookup
seconds
down from ~1 hr
Action-item drop-off
<5%
from ~30%
New-hire ramp on an account
2–3 days
from 2 weeks
We share the underlying breakdown in a discovery call under NDA.
What runs underneath.
Build with us
Have an operation shaped like this one?
We'll start with a 2-week Discovery against your real systems and operating people. If we're not a fit we'll say so on the call.