Institutional memory for agencies & consultancies
Per-client memory that never leaks. Firm-wide skills that keep compounding.
Without memory
A firm juggling many client engagements carries context in people’s heads. When staff rotate, engagement context walks out the door — and mixing client A’s details into client B’s work isn’t just embarrassing, it’s a breach.
With Rekall
One hive per client keeps every engagement’s context, decisions, and history isolated with organization-scoped, membership-verified boundaries — while the firm’s own cross-client skill library keeps compounding. Switch clients, switch memory. Nothing leaks.
How it works
One hive per engagement
Each client’s memory is org-scoped across every store — vector, graph, and relational — with membership checks on every recall.
Staff and agents join per hive
Consultants and their AI agents get the client hive’s full context on day one of the engagement, and lose access cleanly when they roll off.
Skills stay with the firm
Codified skills — how you run a discovery, a migration, an audit — live at the firm level and improve with every engagement.
See it in action
import Rekall from '@rekall/agent-sdk'const rekall = new Rekall({ apiKey: 'rk_...' })const acme = await rekall.hives.create({ name: 'client-acme' })const globex = await rekall.hives.create({ name: 'client-globex' })// Recall is hive-scoped — an agent working the Acme// engagement can never surface Globex contextconst context = await rekall.memories.search({query: 'agreed auth approach',hiveId: acme.id,})