Mergeable by default: Building the context engine to save time and tokens — Peter Werry, Unblocked

2026-05-03 · Source: AI Engineer · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Robotics & Autonomous Systems · Depth: Intermediate, extended

Summary

This session introduces "context engines" for AI agents, defining them as systems that supply optimized, relevant context to agents, enabling streamlined task execution aligned with organizational best practices. The speaker, Peter, and his colleague Brandon, discuss the evolution of context management from human-driven prompting to advanced background agents. They highlight the human bottleneck in managing parallel agents and emphasize that access to data does not equate to understanding. The presentation debunks three myths: naive RAG over documents is a context engine, connecting multiple MCPs is sufficient, and larger context windows solve all problems. A key concept introduced is the "satisfaction of search" problem, where agents stop searching prematurely. The session outlines why a context engine is crucial, detailing its ability to understand organizational context, resolve conflicts, manage access controls, and deliver timely, relevant information. It also covers key requirements, performance impacts, hard lessons learned, and various use cases, including planning, code review, ticket enrichment, triage, incident management, and customer support. The session concludes with a workshop component to build a social engineering graph.

Key takeaway

For AI Engineers and MLOps teams building or deploying AI agents, understanding and implementing a robust context engine is critical. Relying solely on naive RAG or large context windows will lead to "doom loops" and inefficient task execution. You should prioritize systems that can distill organizational knowledge, resolve data conflicts, and manage access controls to ensure agents operate effectively and align with your organization's unique practices, significantly reducing token usage and task completion times.

Key insights

Context engines optimize AI agent performance by providing highly relevant, curated organizational context, moving beyond simple data access.

Principles

• Access to data does not equal understanding.
• Context engines must resolve data conflicts and manage access controls.
• Personalization is crucial for effective retrieval in large organizations.

Method

A context engine integrates diverse data sources (code, docs, conversations) to build relationships, distill best practices, and identify experts. It uses procedural knowledge graphs, vectorization, and LLM-heavy distillation for conflict resolution and targeted retrieval.

In practice

• Integrate context engines in planning and review phases for maximum impact.
• Use context engines for ticket enrichment, triage, and incident management.
• Build skills to curate context and automate workflows.

Topics

• Context Engine
• AI Agent Orchestration
• Organizational Knowledge Management
• Retrieval-Augmented Generation
• Token Optimization

Best for: AI Engineer, Machine Learning Engineer, MLOps Engineer

Related on AIssential

• Stop babysitting your agents... — Brandon Waselnuk, Unblocked — AI agents require dynamic, comprehensive organizational context, not just intelligence or access, to operate effectively and produce mergeable outputs.
• AI Dev 26 x SF | Brandon Waselnuk: Building the Context Engine AI Agents Need — AI agents require a dynamic context engine to achieve organizational intent and avoid costly errors.
• Why Enterprise AI Fails Without a Context Engine - with Eran Yahav of Tabnine — An enterprise context engine provides AI agents with crucial organizational knowledge, significantly boosting success rates and efficiency.
• How RAG, GraphRAG, and Context Engineering Improve AI Performance — Effective AI performance hinges on context engineering, not just raw model intelligence.
• Context Engineering 2.0: MCP, Agentic RAG & Memory // Simba Khadder — Unifying RAG, memory, and structured data into a single "context engine" is crucial for effective AI agents.
• AI agents are learning on the job — just not for your whole team — Enterprise AI agents lack shared memory, hindering team productivity; shared context is vital for multi-user workflows.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI Engineer.