Artificial Intelligence
Preferred on Google
Google News

AI Coding Token Reduction: Rajkumar Sakthivel on Local Code Index

Rajkumar Sakthivel from Tesco discusses how a local code index reduced AI coding tokens by 94%, optimizing costs and performance by focusing on context over model improvements.

9 min read
Rajkumar Sakthivel presenting on AI coding token reduction with a local code index.
AI Engineer

In a presentation at the AI Engineer World's Fair, Rajkumar Sakthivel, representing Tesco, shared a compelling strategy for optimizing AI coding tools. The core revelation is that by implementing a local code index, they were able to achieve a remarkable 94% reduction in AI coding tokens, significantly cutting costs and improving performance.

AI Coding Token Reduction: Rajkumar Sakthivel on Local Code Index - AI Engineer
AI Coding Token Reduction: Rajkumar Sakthivel on Local Code Index — from AI Engineer

Visual TL;DR. Excessive AI Context leads to High Costs & Latency. High Costs & Latency led to Local Code Index. Local Code Index leads to Combined Retrieval Methods. Combined Retrieval Methods leads to Input Optimization. Local Code Index resulted in 94% Token Reduction. 94% Token Reduction enabled Optimized Costs & Performance. Input Optimization challenges Knowing When Retrieval Wrong.

Related startups

  1. Excessive AI Context: sending ~45,000 tokens, only ~5,000 useful
  2. High Costs & Latency: inefficiency led to increased costs and latency
  3. Local Code Index: focusing on context optimization, not model improvements
  4. Combined Retrieval Methods: leveraging multiple techniques for better context selection
  5. Input Optimization: crucial for efficient and effective AI interactions
  6. 94% Token Reduction: achieved remarkable reduction in AI coding tokens
  7. Optimized Costs & Performance: significantly cutting costs and improving performance
  8. Knowing When Retrieval Wrong: the hard part of identifying incorrect context
Visual TL;DR
Visual TL;DR, startuphub.ai Excessive AI Context leads to High Costs & Latency. High Costs & Latency led to Local Code Index. Local Code Index resulted in 94% Token Reduction. 94% Token Reduction enabled Optimized Costs & Performance led to resulted in enabled Excessive AI Context High Costs & Latency Local Code Index 94% Token Reduction Optimized Costs & Performance From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Excessive AI Context leads to High Costs & Latency. High Costs & Latency led to Local Code Index. Local Code Index resulted in 94% Token Reduction. 94% Token Reduction enabled Optimized Costs & Performance led to resulted in enabled Excessive AIContext High Costs &Latency Local Code Index 94% TokenReduction Optimized Costs &Performance From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Excessive AI Context leads to High Costs & Latency. High Costs & Latency led to Local Code Index. Local Code Index resulted in 94% Token Reduction. 94% Token Reduction enabled Optimized Costs & Performance led to resulted in enabled Excessive AI Context sending ~45,000 tokens, only ~5,000 useful High Costs & Latency inefficiency led to increased costs andlatency Local Code Index focusing on context optimization, notmodel improvements 94% Token Reduction achieved remarkable reduction in AI codingtokens Optimized Costs & Performance significantly cutting costs and improvingperformance From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Excessive AI Context leads to High Costs & Latency. High Costs & Latency led to Local Code Index. Local Code Index resulted in 94% Token Reduction. 94% Token Reduction enabled Optimized Costs & Performance led to resulted in enabled Excessive AIContext sending ~45,000tokens, only ~5,000useful High Costs &Latency inefficiency led toincreased costs andlatency Local Code Index focusing on contextoptimization, notmodel improvements 94% TokenReduction achieved remarkablereduction in AIcoding tokens Optimized Costs &Performance significantlycutting costs andimproving… From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Excessive AI Context leads to High Costs & Latency. High Costs & Latency led to Local Code Index. Local Code Index leads to Combined Retrieval Methods. Combined Retrieval Methods leads to Input Optimization. Local Code Index resulted in 94% Token Reduction. 94% Token Reduction enabled Optimized Costs & Performance. Input Optimization challenges Knowing When Retrieval Wrong led to resulted in enabled challenges Excessive AI Context sending ~45,000 tokens, only ~5,000 useful High Costs & Latency inefficiency led to increased costs andlatency Local Code Index focusing on context optimization, notmodel improvements Combined Retrieval Methods leveraging multiple techniques for bettercontext selection Input Optimization crucial for efficient and effective AIinteractions 94% Token Reduction achieved remarkable reduction in AI codingtokens Optimized Costs & Performance significantly cutting costs and improvingperformance Knowing When Retrieval Wrong the hard part of identifying incorrectcontext From startuphub.ai · The publishers behind this format
Visual TL;DR, startuphub.ai Excessive AI Context leads to High Costs & Latency. High Costs & Latency led to Local Code Index. Local Code Index leads to Combined Retrieval Methods. Combined Retrieval Methods leads to Input Optimization. Local Code Index resulted in 94% Token Reduction. 94% Token Reduction enabled Optimized Costs & Performance. Input Optimization challenges Knowing When Retrieval Wrong led to resulted in enabled challenges Excessive AIContext sending ~45,000tokens, only ~5,000useful High Costs &Latency inefficiency led toincreased costs andlatency Local Code Index focusing on contextoptimization, notmodel improvements CombinedRetrieval Methods leveraging multipletechniques forbetter context… InputOptimization crucial forefficient andeffective AI… 94% TokenReduction achieved remarkablereduction in AIcoding tokens Optimized Costs &Performance significantlycutting costs andimproving… Knowing WhenRetrieval Wrong the hard part ofidentifyingincorrect context From startuphub.ai · The publishers behind this format

The Problem with Excessive Context

Sakthivel highlighted a common assumption in AI coding tools: the belief that sending as much context as possible to the model leads to better results. However, their experience revealed that out of approximately 45,000 tokens sent per query, only around 5,000 were actually useful. This inefficiency led to increased costs and latency, prompting a search for a more optimized approach.

The Solution: Local Code Indexing

The team focused on optimizing the context rather than the AI model itself. They explored several avenues, including better prompts, adjusted model settings, and output compression. However, the most impactful solution identified was the introduction of a retrieval layer between the codebase and the AI agent. This layer, running locally, indexes code and retrieves only the relevant chunks, drastically reducing the token count.

The architecture involves several steps:

  • Tree-sitter Chunking: Breaking down code into meaningful AST-aware splits across 10 languages.
  • Hybrid Retrieval: Employing both vector search (for conceptual similarity) and BM25 (for exact keyword matching).
  • Chunk Compression: Compressing the retrieved code chunks, achieving an 89% reduction.
  • Code Graph: Utilizing a code graph to understand relationships and dependencies within the code.
  • Confidence Scoring: Implementing a system to filter results based on confidence levels, ensuring only relevant information is passed to the AI.

Crucially, this entire process runs locally, eliminating cloud dependencies, API calls, and ensuring data privacy. The system stores data in three SQLite files, making it efficient and accessible.

The Power of Combined Retrieval Methods

Sakthivel also discussed the limitations of relying on a single retrieval method. Vector search, while effective for finding conceptually related code even with different naming conventions, can miss exact matches. Conversely, BM25 excels at precise keyword matching but struggles with semantic similarity. By employing Reciprocal Rank Fusion (RRF), they combined the strengths of both approaches, achieving a recall of 0.90 and covering each method's blind spots.

The Importance of Input Optimization

A key takeaway presented was the breakdown of where AI coding tokens are actually used. The data showed that 90% of tokens are typically input (file reads, search, context), while only 10% are output (agent replies, code). This starkly illustrates why optimizing the input side is critical for cost savings. Input retrieval, in particular, was shown to save approximately 61% of the total bill, whereas output compression saved around 8%.

The Hard Part: Knowing When Retrieval is Wrong

Sakthivel emphasized that the most challenging aspect wasn't simply retrieving relevant code, but determining when the retrieval process itself was flawed. They experimented with LLM-based scoring, which was accurate but added latency and cost. Fixed thresholds also proved problematic, as they struggled with short queries and long queries alike. Ultimately, a simpler heuristic approach, using a weighted average of similarity, keywords, and recency, proved most effective, delivering results with low latency and no API calls.

Measurable Impact and Open Source

The presentation included benchmark data from a real project using the FastAPI framework. This demonstrated a reduction from 83,681 tokens per query in a full-file baseline to just 4,927 tokens after retrieval, and further down to 523 tokens after compression, achieving a 94% saving. The recall remained high at 0.90. The project is open-source, with a QR code and GitHub link provided for users to try it themselves and run the benchmarks.

Multi-Agent and Shared Memory

The system is designed for a multi-agent environment, allowing various AI coding tools like Claude Code, Cursor, VS Code Copilot, Codex CLI, Gemini CLI, Tabnine, and OpenCode to utilize a single, shared index. This shared index is per-project, not per-agent, and decisions persist across sessions and tools, creating a unified and efficient coding experience.

© 2026 StartupHub.ai. All rights reserved. Do not enter, scrape, copy, reproduce, or republish this article in whole or in part. Use as input to AI training, fine-tuning, retrieval-augmented generation, or any machine-learning system is prohibited without written license. Substantially-similar derivative works will be pursued to the fullest extent of applicable copyright, database, and computer-misuse laws. See our terms.
#Rajkumar Sakthivel#Tesco#AI Engineering#Large Language Models#Code Context Engine#Vector Search#BM25#RRF Fusion#Open Source