Core Concepts
Tribal Knowledge: The Hidden Intelligence in Your Organization
Tribal knowledge is the exception logic, unwritten rules, and institutional memory that lives in people's heads. It's what makes experienced employees invaluable—and what AI agents desperately need to operate autonomously.
What is Tribal Knowledge?
Tribal knowledge refers to the informal, undocumented information that employees accumulate through experience working in an organization. It includes:
- ●Unwritten rules — "We always give Tier 1 customers an extra week on invoices"
- ●Exception patterns — "When the CFO approves, we can skip the legal review"
- ●Historical context — "We tried that approach in 2022 and it didn't work because..."
- ●Relationship knowledge — "Always CC Sarah on deals over $100k"
- ●Workarounds — "The system says 30 days but operations can actually do 45"
The Tribal Knowledge Problem
Tribal knowledge represents both immense value and significant risk:
The Value
- • Enables handling of edge cases
- • Powers efficient decision-making
- • Contains decades of accumulated wisdom
- • Differentiates experienced from new employees
The Risk
- • Walks out the door when employees leave
- • Impossible to transfer to AI systems
- • Creates key-person dependencies
- • Blocks automation and scaling
“The missing layer that actually runs enterprises is decision traces—the exceptions, overrides, precedents, and cross-system context that currently live in Slack threads, deal desk conversations, escalation calls, and people's heads.”
Where Tribal Knowledge Hides
Tribal knowledge isn't completely invisible—it leaves traces in various places:
Slack/Teams Threads
"Hey, can we do X?" "Yes, but only if Y because of Z" — exception logic buried in chat.
Email Chains
Approval decisions, escalation patterns, and negotiation history lost in inboxes.
Meeting Notes
Key decisions made verbally, documented inconsistently if at all.
Ticket Comments
Workarounds and special handling instructions added as afterthoughts.
People's Heads
The most valuable and most vulnerable repository of all.
Capturing Tribal Knowledge with Context Graphs
Context graphs provide a systematic way to capture and preserve tribal knowledge as decision traces. The approach involves:
- 1
Instrument Decision Points
Identify where tribal knowledge is applied—approvals, exceptions, escalations, overrides.
- 2
Capture the Context
Record not just what was decided, but why—the reasoning, precedents, and factors considered.
- 3
Connect to Entities
Link decisions to business objects—customers, products, policies, people involved.
- 4
Enable Precedent Search
Make tribal knowledge searchable so AI agents (and humans) can find similar past decisions.
Why AI Agents Need Tribal Knowledge
AI agents that lack access to tribal knowledge are severely limited:
“Without context graphs, an AI agent is like an extremely smart intern on day one—it can follow written rules but gets tripped up by every unwritten exception.”
By capturing tribal knowledge as decision traces in a context graph, organizations can give AI agents the institutional memory they need to handle real-world complexity. The agent doesn't need to know every rule—it needs to find similar past situations and apply the same reasoning.
References
This article is based on insights from the following sources:
- •Eyal Toledano — “Context Graphs Can't Organize Knowledge That Was Never Captured” on X/Twitter
- •Arvind Jain (CEO, Glean) — “Context is the next data platform” on X/Twitter