Intel / video

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

Source: youtube.com
Author: AI Engineer
Date: Published Sep 9, 2026

Context

A coding can find a relevant document and still follow an obsolete decision. Unblocked, a company building tools for software teams, describes addressing that problem by comparing code, documents and development conversations before returning information to the agent.

Its company account says the system weighs recency and expertise inferred from authorship and review activity. When sources disagree and the conflict cannot be resolved, it can show both accounts with citations instead of silently discarding one. That supplies the mechanism behind the talk's argument: connecting an agent to more information is only part of the problem. It also needs help identifying which information reflects the team's current decisions.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.

Chapters

0:00Why AI code should feel like a teammate wrote it
1:50How bad context compounds as agents scale
4:20Two dead ends: curated context and the MCP plateau
5:52What a context engine actually has to do
7:26Inside the engine and its six characteristics
9:16The same prompt with and without context
10:11Three open source tools to try
12:25Beyond code generation, and closing

Read the source www.youtube.com