# Your agents lack context: Here's how to fix "You're absolutely right!" — Brandon Waselnuk, Unblocked

## Executive summary

AI agents require more than just access to information; they need a 'context engine' to achieve true understanding of a company's operations, business logic, and historical decisions. The talk details how agents fail when they lack deep, organizational context, leading to wasted tokens and time. The solution involves building a context engine that ingests data from disparate sources (e.g., ticketing systems, code repos) and provides structured, conflict-resolved context to the model in a token-optimized manner.

## Key takeaways

- The Context Gap: AI-generated code must feel like it was written by a long-term team member, which requires deep operational context that current agents lack. This gap compounds as agents scale from simple tab completion to background processes.
- Failure Modes of Context: Common approaches like the 'curated context trap' (where documentation rots) and the 'MCP plateau' (where agents suffer from satisfaction of search bias) fail because access to information is not the same as understanding it.
- The Role of the Context Engine: A proper context engine must perform conflict resolution (e.g., resolving conflicting architecture diagrams vs. Slack messages), respect permissions (OAuth/SSO), and deliver context in a token-optimized way.
- Efficiency Gains: Implementing a context engine can lead to significant efficiency gains, demonstrated by a reduction from 21 million to 10.8 million tokens and a two-hour reduction in wall-clock time for the same task.

## Technical details

- Context Engineering Principles: The engine must provide six key characteristics: unified system context, conflict resolution, personalized relevance, permission enforcement, token optimization, and the ability to map and discover unknown unknowns.
- Agent Scaling Issues: As agents move into parallel or background roles, they face 'doom loops,' resulting in wasted search tokens and a 'review tax' because they lack understanding of business logic and deployment procedures.
- Contextual Retrieval: The engine ingests real-time data from multiple sources (e.g., instant management tool chains) and outputs context to the exact workflow, going beyond simple Retrieval Augmented Generation (RAG) to handle relational data queries.
- Tools and Techniques: The speaker demonstrated tools like a social network graph that maps team members and commits, and the 'repo rules agent,' which indexes rules files to help deduplicate context and improve retrieval.

## Practical implications

- Implement a centralized context layer to feed AI agents, moving beyond simple document retrieval (RAG).
- Focus on building mechanisms for conflict resolution (e.g., prioritizing recent Slack messages over old architecture diagrams).
- Ensure agents respect organizational permissions (OAuth/SSO) to prevent data leakage.
- Measure token and time savings to justify the investment in a robust context engine.
- Use tools to map internal knowledge graphs (e.g., linking commits to team members and rules files).

## Topics

Context Engineering, AI Agents, LLMs, Build Engineering, DevOps, Token Optimization, RAG, Unblocked, relational context engine workshop, readiness.unblocked.com

Source: https://www.youtube.com/watch?v=KcVkq5L-0f0