Your AI Reads Entire Files. It Only Needs 6%. Here's How We Fixed It.
Coding agents waste context when they read entire files. Every AI coding agent I've used does the same thing: it reads your entire file, every single time. That is the entry point for "Your AI Reads Entire Files. It Only Needs 6%. Here's How We Fixed It.", where Rajkumar Sakthivel shows how AST-aware retrieval and local vector search can give agents only the code they need in practice.
Session details
Every AI coding agent I've used does the same thing: it reads your entire file, every single time. On a mid-size Python project we work on, that's 45,000 tokens per query. The agent only needed about 4,900. We were burning through input tokens and didn't even realise how much until the bill landed. So we built something to fix it. An open-source tool that indexes your codebase locally using tree-sitter for AST-aware chunking and sqlite-vec for vector search. It plugs into your editor through the Model Context Protocol, so the agent gets exactly the code it needs instead of everything. The result: 94% fewer tokens, measured across 20 real-world coding queries. What made this tricky was getting retrieval quality right. Pure vector search missed too much; keyword search returned noise. We ended up combining both with Reciprocal Rank Fusion and adding confidence scoring so the agent knows when it should fall back to reading the full file. In this talk, I'll walk through the architecture decisions we made and the ones we got wrong. Why we picked sqlite-vec over LanceDB. Why we chose truncation over LLM-based summarisation. How we handle secret redaction before anything hits the embedding model. And I'll show live benchmarks comparing retrieval with and without the index. You'll walk away knowing how to build your own local RAG for code, how to measure token savings properly, and practical ways to cut your AI tooling costs without changing how you work.
Primary action
Share the public session page with the tracked partner link, then use the copy variants below for LinkedIn, short social posts, and newsletter or email mentions.
Agenda facts
- Format
- Talk / Presentation
- Kind
- Conference Session
- Topic
- Artificial Intelligence and Machine Learning
- Level
- Intermediate
- Time
- Oct 30, 2026, 7:15 AM
- Room
- Room C
- Duration
- 60 min
Speakers
Link partners to speaker kits when available, or to the public speaker profile on mitechcon.net.
Rajkumar Sakthivel
Copy to paste
LinkedIn, short social, and newsletter or email copy variants for this session.
LinkedIn
Session LinkedIn Post
Coding agents waste context when they read entire files. Every AI coding agent I've used does the same thing: it reads your entire file, every single time. That is the entry point for "Your AI Reads Entire Files. It Only Needs 6%. Here's How We Fixed It.", where Rajkumar Sakthivel shows how AST-aware retrieval and local vector search can give agents only the code they need in practice. 📅 October 28–30, 2026 📍 Oakland University, Rochester, MI 🔗 https://www.mitechcon.net/sessions/your-ai-reads-entire-files-it-only-needs-6-here-s-how-we-fixed-it/?utm_source=partner-kit&utm_medium=partner&utm_campaign=mitechcon-2026&utm_content=session-your-ai-reads-entire-files-it-only-needs-6-here-s-how-we-fixed-it #MITechCon #CodingAgents #ContextEngineering #AIEngineering
Short social
Short Session Post
Coding agents waste context when they read entire files. Learn how AST-aware retrieval and local vector search can give agents only the code they need. https://www.mitechcon.net/sessions/your-ai-reads-entire-files-it-only-needs-6-here-s-how-we-fixed-it/?utm_source=partner-kit&utm_medium=partner&utm_campaign=mitechcon-2026&utm_content=session-your-ai-reads-entire-files-it-only-needs-6-here-s-how-we-fixed-it #MITechCon #CodingAgents
Newsletter / email
Newsletter Or Email Blurb
Feature this MITechCon 2026 session in your newsletter or email: "Your AI Reads Entire Files. It Only Needs 6%. Here's How We Fixed It.". Coding agents waste context when they read entire files. Every AI coding agent I've used does the same thing: it reads your entire file, every single time. That is the entry point for "Your AI Reads Entire Files. It Only Needs 6%. Here's How We Fixed It.", where Rajkumar Sakthivel shows how AST-aware retrieval and local vector search can give agents only the code they need in practice. Format: Conference Session. Topic: Artificial Intelligence and Machine Learning. Level: Intermediate. Scheduled for Oct 30, 2026, 7:15 AM in Room C. Learn more and share the session: https://www.mitechcon.net/sessions/your-ai-reads-entire-files-it-only-needs-6-here-s-how-we-fixed-it/?utm_source=partner-kit&utm_medium=partner&utm_campaign=mitechcon-2026&utm_content=session-your-ai-reads-entire-files-it-only-needs-6-here-s-how-we-fixed-it
Share image
Generated session card with alt text, dimensions, file size, and download link.
Your AI Reads Entire Files. It Only Needs 6%. Here's How We Fixed It. share image
Generated session social card for partner promotion.
- Alt text
- Social share image for Your AI Reads Entire Files. It Only Needs 6%. Here's How We Fixed It. at MITechCon 2026
- Dimensions
- 1024 x 1024px
- Size
- 576 KB
- Updated
- 2026-07-08