--- title: Are AI Coding Agents Spending Too Many Tokens Just... description: A discussion raises the question of whether AI coding agents waste too many tokens just understanding a codebase before doing useful work. It references... canonical: https://daily.dev/posts/are-ai-coding-agents-spending-too-many-tokens-just-understanding-the-codebase--znyegczsx twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Are AI Coding Agents Spending Too Many Tokens Just Understanding the Codebase? | daily.dev og:description: A discussion raises the question of whether AI coding agents waste too many tokens just understanding a codebase before doing useful work. It references... og:url: https://daily.dev/posts/are-ai-coding-agents-spending-too-many-tokens-just-understanding-the-codebase--znyegczsx og:image: https://api.daily.dev/og/posts/ZnYegCZSX.png og:image:alt: Are AI Coding Agents Spending Too Many Tokens Just Understanding the Codebase? og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Are AI Coding Agents Spending Too Many Tokens Just Understanding the Codebase? **[DevOps Daily](https://daily.dev/sources/devopsdaily)** · [@divyansh30](https://daily.dev/divyansh30) · 1 min read · 1 upvotes · 1 comments ## Summary A discussion raises the question of whether AI coding agents waste too many tokens just understanding a codebase before doing useful work. It references Graphify, an open-source project that converts a codebase into a queryable knowledge graph, citing a claimed 71.5x reduction in tokens versus having an agent parse raw files in a production test. The poster asks fellow engineers whether token costs are a real concern, whether structured context speeds up agents, and whether a knowledge-graph approach would be worth trying on large production repos. ## Content I've been thinking about how much work an AI coding agent has to do before it can actually answer a question. In a large repository, it may need to search through files, follow dependencies, understand the project structure, and figure out which code is relevant. All of that uses tokens before the actual task even starts. I came across Graphify, an open-source project that takes a different approach by turning a codebase into a queryable knowledge graph. The interesting part is the reported result: **71.5× fewer tokens in the production test** compared with having the agent work through the raw files. [https://www.kubeblogs.com/graphify-claims-71-5x-fewer-tokens-we-tested-it-on-real-production-work/](https://www.kubeblogs.com/graphify-claims-71-5x-fewer-tokens-we-tested-it-on-real-production-work/) I'm curious what other DevOps engineers think: - Are token costs becoming a concern when using AI coding agents? - Do you think structured codebase context can make agents faster? - Would you try a knowledge-graph approach on a large production repository? Has anyone tried something similar? ## Similar posts on daily.dev - [How Semantic Code Navigation Cuts Agent Token Costs by up to 36%](https://daily.dev/posts/how-semantic-code-navigation-cuts-agent-token-costs-by-up-to-36--tlu1vjjqp) · Daily Dose of Data Science \| Avi Chawla \| Substack · 0 upvotes · 0 comments - [Stop Burning Your AI Limits: A Token Diet for Long Coding Days](https://daily.dev/posts/stop-burning-your-ai-limits-a-token-diet-for-long-coding-days-atlnqpbej) · ITNEXT · 1 upvotes · 0 comments - [The Harness Tax: The Dead Weight Inside Your Coding Agent](https://daily.dev/posts/the-harness-tax-the-dead-weight-inside-your-coding-agent-iiu0gkchd) · portkey · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#context-engineering](https://daily.dev/tags/context-engineering) [View this post on daily.dev](https://daily.dev/posts/are-ai-coding-agents-spending-too-many-tokens-just-understanding-the-codebase--znyegczsx) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/are-ai-coding-agents-spending-too-many-tokens-just-understanding-the-codebase--znyegczsx","headline":"Are AI Coding Agents Spending Too Many Tokens Just Understanding the Codebase?","text":"A discussion raises the question of whether AI coding agents waste too many tokens just understanding a codebase before doing useful work. It references Graphify, an open-source project that converts a codebase into a queryable knowledge graph, citing a claimed 71.5x reduction in tokens versus having an agent parse raw files in a production test. The poster asks fellow engineers whether token costs are a real concern, whether structured context speeds up agents, and whether a knowledge-graph approach would be worth trying on large production repos.","url":"https://daily.dev/posts/are-ai-coding-agents-spending-too-many-tokens-just-understanding-the-codebase--znyegczsx","datePublished":"2026-08-17T06:19:28.159Z","dateModified":"2026-09-13T21:33:08.854Z","author":{"@type":"Person","name":"Divyansh","url":"https://daily.dev/divyansh30","image":"https://media.daily.dev/image/upload/s--f9lbkpFv--/f_auto/v1744873189/avatars/avatar_JMQ4qarKTe7zPurSPRrzM","description":"Aspiring engineer with an interest in AI, tech and DevOps :)","interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"EndorseAction"},"userInteractionCount":1070}},"image":"https://media.daily.dev/image/upload/s--E17EQqCb--/f_auto/v1786941836/posts/4t2AcmS1f?_a=BAMAMicg0","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"isPartOf":{"@type":"WebPage","url":"https://daily.dev/squads/devopsdaily","name":"DevOps Daily"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"DevOps Daily","item":"https://daily.dev/squads/devopsdaily"},{"@type":"ListItem","position":3,"name":"Are AI Coding Agents Spending Too Many Tokens Just Understanding the Codebase?"}]} ```