--- title: How to use AI agents better than 99% of people | daily.dev description: When multiple AI agents share the same project, each maintains its own isolated memory, causing context fragmentation. Graph-shaped memory solves this by... canonical: https://daily.dev/posts/how-to-use-ai-agents-better-than-99-of-people-5tzhgrsk9 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How to use AI agents better than 99% of people | daily.dev og:description: When multiple AI agents share the same project, each maintains its own isolated memory, causing context fragmentation. Graph-shaped memory solves this by... og:url: https://daily.dev/posts/how-to-use-ai-agents-better-than-99-of-people-5tzhgrsk9 og:image: https://api.daily.dev/og/posts/5TZHgrsK9.png og:image:alt: How to use AI agents better than 99% of people 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. # How to use AI agents better than 99% of people **[System Design Newsletter](https://daily.dev/sources/systemdesignnews)** · 28 min read · 1 upvotes · 0 comments ## Summary When multiple AI agents share the same project, each maintains its own isolated memory, causing context fragmentation. Graph-shaped memory solves this by storing facts as nodes and relationships as edges in a typed, shared graph. Omnigraph is an open-source engine that implements this pattern on top of Amazon S3, using atomic writes via a manifest-pointer approach (similar to Apache Iceberg), Git-style branching so agents propose changes before they reach shared truth, compare-and-swap for conflict prevention, and a hybrid retrieval system combining graph traversal, BM25 keyword search, and vector k-NN with Reciprocal Rank Fusion. The post also covers when graph memory is the wrong tool, such as high-throughput transactional workloads. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How does Omnigraph achieve atomic writes on object storage like Amazon S3 that doesn't natively support transactions? Omnigraph splits the graph into separate versioned stores per node and edge type, writes new versions without touching a small manifest file, then updates the manifest in a single atomic operation that moves all pointers at once. Before commit, agents see the old versions; after, they see the new ones; crashes leave orphaned but invisible files, ensuring consistency. _Developers weighing how to add transactional guarantees to object storage can track patterns like this through daily.dev._ ### How does Omnigraph prevent AI agents from overwriting each other's changes when writing to a shared graph? Omnigraph uses compare-and-swap (CAS) before updating the manifest, checking whether each store is still on the version the writer originally read. If another writer already published a newer version, the commit is rejected and the writer retries from the latest state, preventing silent overwrites of concurrent updates. _Teams designing concurrency control for multi-agent systems can follow this kind of architecture discussion on daily.dev._ ### How does Omnigraph combine graph traversal, keyword search, and vector search into a single query for AI agent context retrieval? Omnigraph builds three indexes (graph relationships, full-text via BM25, and vector via k-nearest neighbors with cosine similarity) from one copy of the data in object storage, keeping them in sync. Each search returns its own ranked list, and Reciprocal Rank Fusion merges the rankings so results appearing near the top across multiple methods rise to the final top. _Anyone comparing retrieval strategies for agent context can find deep dives like this on daily.dev._ ## Similar posts on daily.dev - [Top AI Labs Share an Agent Memory Trick Most Miss](https://daily.dev/posts/top-ai-labs-share-an-agent-memory-trick-most-miss-f9zsfcwik) · Daily Dose of Data Science \| Avi Chawla \| Substack · 0 upvotes · 0 comments - [Context Graphs and AI Memory Across the Globe](https://daily.dev/posts/context-graphs-and-ai-memory-across-the-globe-etrhxkqsg) · neo4j · 1 upvotes · 0 comments - [Making AI Agents work with Knowledge Graphs](https://daily.dev/posts/making-ai-agents-work-with-knowledge-graphs-qbkhvarrq) · Last9 · 2 upvotes · 0 comments - [Agent Memory: Why Your AI Has Amnesia and How to Fix It](https://daily.dev/posts/agent-memory-why-your-ai-has-amnesia-and-how-to-fix-it-nkpwfuhgz) · Tech Lead Digest · 1 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#architecture](https://daily.dev/tags/architecture), [#rag](https://daily.dev/tags/rag) [View this post on daily.dev](https://daily.dev/posts/how-to-use-ai-agents-better-than-99-of-people-5tzhgrsk9) ```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":"TechArticle","headline":"How to use AI agents better than 99% of people","url":"https://daily.dev/posts/how-to-use-ai-agents-better-than-99-of-people-5tzhgrsk9","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-to-use-ai-agents-better-than-99-of-people-5tzhgrsk9"},"datePublished":"2026-07-07T11:50:44.647Z","dateModified":"2026-09-13T21:04:19.297Z","description":"When multiple AI agents share the same project, each maintains its own isolated memory, causing context fragmentation. Graph-shaped memory solves this by...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2058e2d34cde6230cb72285dad2d1fe3?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/2058e2d34cde6230cb72285dad2d1fe3?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"System Design Newsletter","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"System Design Newsletter","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/eec821422ebe4672a738b8fbfa2c0dbd","url":"https://daily.dev/sources/systemdesignnews"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-to-use-ai-agents-better-than-99-of-people-5tzhgrsk9","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,architecture,rag","timeRequired":"PT28M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"System Design Newsletter","item":"https://daily.dev/sources/systemdesignnews"},{"@type":"ListItem","position":3,"name":"How to use AI agents better than 99% of people"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-to-use-ai-agents-better-than-99-of-people-5tzhgrsk9#faq","mainEntity":[{"@type":"Question","name":"How does Omnigraph achieve atomic writes on object storage like Amazon S3 that doesn't natively support transactions?","acceptedAnswer":{"@type":"Answer","text":"Omnigraph splits the graph into separate versioned stores per node and edge type, writes new versions without touching a small manifest file, then updates the manifest in a single atomic operation that moves all pointers at once. Before commit, agents see the old versions; after, they see the new ones; crashes leave orphaned but invisible files, ensuring consistency. Developers weighing how to add transactional guarantees to object storage can track patterns like this through daily.dev."}},{"@type":"Question","name":"How does Omnigraph prevent AI agents from overwriting each other's changes when writing to a shared graph?","acceptedAnswer":{"@type":"Answer","text":"Omnigraph uses compare-and-swap (CAS) before updating the manifest, checking whether each store is still on the version the writer originally read. If another writer already published a newer version, the commit is rejected and the writer retries from the latest state, preventing silent overwrites of concurrent updates. Teams designing concurrency control for multi-agent systems can follow this kind of architecture discussion on daily.dev."}},{"@type":"Question","name":"How does Omnigraph combine graph traversal, keyword search, and vector search into a single query for AI agent context retrieval?","acceptedAnswer":{"@type":"Answer","text":"Omnigraph builds three indexes (graph relationships, full-text via BM25, and vector via k-nearest neighbors with cosine similarity) from one copy of the data in object storage, keeping them in sync. Each search returns its own ranked list, and Reciprocal Rank Fusion merges the rankings so results appearing near the top across multiple methods rise to the final top. Anyone comparing retrieval strategies for agent context can find deep dives like this on daily.dev."}}]} ```