# Blog -- Neon Docs The latest product updates from Neon This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Inside LuBot's database-per-tenant architecture DevFeed: [Inside LuBot's database-per-tenant architecture]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-09-14T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Database](), [Software as a service](), [Supabase](), [Provisioning](), [stripe](), [API]() Tags: [api](), [architecture](), [billing](), [case-studies](), [checkout](), [database](), [infrastructure](), [product](), [provisioning](), [saas](), [scale](), [stripe]() ### AI overview This article explains LuBot's database-per-tenant architecture. LuBot uses Supabase for shared administrative data such as organizations, users, and billing state, while each paying customer receives an isolated Postgres database. Stripe checkout triggers a webhook that provisions a Neon branch through the Neon API, with deterministic naming to prevent duplicate tenants. Neon branches provide independent compute and can scale idle tenants to zero. ### Source excerpt It took six days to go from zero to isolated Postgres per tenant, provisioned the moment someone pays. A traditional Postgres setup would have been a full quarter of work. ## Claimable Neon: Provisioned by agents, claimed by humans DevFeed: [Claimable Neon: Provisioned by agents, claimed by humans]() Original publisher: [Read original article]() Author: Andre Landgraf Published: 2026-09-10T12:00:00Z Content type: release Language: en Sources: [Blog -- Neon Docs]() Topics: [AI-assisted coding](), [AI Bots]() Tags: [agents](), [api](), [auth](), [backend](), [community](), [database](), [oauth](), [postgres](), [product](), [sql]() ### AI overview Claimable Neon lets agents anonymously provision a temporary Neon project, continue building with scoped credentials, and generate a link for a human to claim the project later. ### Source excerpt Claimable Neon implements the anonymous registration method in auth.md, the open agent registration protocol authored by WorkOS, to give agents a way to provision a temporary Neon project without creating an account or collecting payment details. ## Improving Lakebase Postgres Compute Cache on Neon, Part 1 DevFeed: [Improving Lakebase Postgres Compute Cache on Neon, Part 1]() Original publisher: [Read original article]() Author: Sunil Kamath Published: 2026-09-09T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Caching](), [Databases](), [Operating system](), [Amazon S3](), [cpu](), [Latency](), [autoscaling]() Tags: [autoscaling](), [cache](), [compute](), [cpu](), [engineering](), [latency](), [memory](), [os](), [postgres](), [s3](), [storage]() ### AI overview Neon describes a Lakebase Postgres compute-cache change that allocates most memory to shared buffers backed by huge pages. The goal is to keep hot pages in DRAM, reducing storage reads, CPU use, and latency; the article reports up to roughly 2x throughput on specified fixed-size computes. ### Source excerpt On large fixed-size Lakebase Postgres computes on Neon, we now put most of the machine's memory into Postgres shared buffers and back that cache with huge pages. Hot pages stay in DRAM instead of falling through to a local disk cache, so the same working set is served faster and with less CPU. ## Comparing token economics across 42 AI models DevFeed: [Comparing token economics across 42 AI models]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-09-04T12:00:00Z Content type: comparison Language: en Sources: [Blog -- Neon Docs]() Topics: [AI Chat](), [AI-assisted coding]() Tags: [ai-gateway](), [ai-models](), [benchmark](), [community](), [cost](), [llms]() ### AI overview A comparison of 42 AI models on a synthetic workload of 100 support tickets, evaluating cost, usable responses, and completion time. It highlights differences in token use, accuracy, speed, and total workload cost. ### Source excerpt If headcount is often a company's largest cost, tokens will be a close second soon as agents take on more work. Imagine how much $$ a software company will be able to save just by pre-selecting models that are efficient to run, especially in large companies with many engineers using coding agents. ## Give your agent Neon tools DevFeed: [Give your agent Neon tools]() Original publisher: [Read original article]() Author: Andre Landgraf Published: 2026-09-03T12:00:00Z Content type: release Language: en Sources: [Blog -- Neon Docs]() Topics: [MSP MCP]() Tags: [agent](), [api](), [mcp](), [mcp-server](), [openapi](), [product](), [sdks](), [tools](), [workflows]() ### AI overview Neon announces @neon/tools, which adapts its SDK client into typed tools for agents and expands the hosted Neon MCP Server. The article explains progressive tool discovery and why agent-oriented tools should bundle workflows rather than mirror raw REST operations. ### Source excerpt Today we're launching @neon/tools, a package that turns the @neon/sdk ergonomic client into typed agent tools with adapters for MCP, Mastra, and Eve. We're also using it to expand the hosted Neon MCP Server. ## Building patterns unlocked by scale to zero DevFeed: [Building patterns unlocked by scale to zero]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-09-01T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Databases](), [autoscaling](), [ci]() Tags: [autoscaling](), [building](), [ci](), [compute](), [database](), [postgres](), [product](), [scale]() ### AI overview The article explains how autoscaling and scale to zero reduce idle Postgres compute costs. It presents ephemeral database branches for previews, tests, and individual development work as patterns enabled by suspending inactive compute. ### Source excerpt Most hosted Postgres pricing works like this: you pick an instance size, the vendor assumes the instance will run for about 730 hours a month, and you pay for those hours. Lakebase Postgres addresses that waste with autoscaling and scale to zero. ## Neon is now available in Grok Bot DevFeed: [Neon is now available in Grok Bot]() Original publisher: [Read original article]() Author: Dominik Koch Published: 2026-09-01T12:00:00Z Content type: news Language: en Sources: [Blog -- Neon Docs]() Topics: [Bot](), [Databases](), [Model Context Protocol]() Tags: [agent](), [agent-skills](), [bots](), [database](), [mcp](), [mcp-server](), [migration](), [product](), [sql](), [testing](), [tools](), [workflows]() ### AI overview Neon is available as a Grok Bot plugin, combining Neon agent skills with the Neon MCP Server to let authorized bots manage Neon database workflows. ### Source excerpt Neon is now available as a plugin in Grok Bot. You just have to connect your Neon account once and you'll be able to ask a Grok bot to build on Neon. ## Autoscaling Lakebase Postgres DevFeed: [Autoscaling Lakebase Postgres]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-31T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Database](), [cpu]() Tags: [architecture](), [autoscaling](), [cache](), [compute](), [cpu](), [database](), [engineering](), [memory](), [postgres](), [storage]() ### AI overview The article explains how Lakebase Postgres autoscales database compute without stopping Postgres. Its architecture separates compute from durable storage, while the autoscaler derives a target size from CPU load, memory use, and compute-cache working-set size. ### Source excerpt Choosing a database instance size before you know the workload is an old building pattern. Lakebase Postgres omits the sizing experience altogether thanks to autoscaling. ## Introducing Neon Labs DevFeed: [Introducing Neon Labs]() Original publisher: [Read original article]() Author: Savannah Longoria Published: 2026-08-26T12:00:00Z Content type: release Language: en Sources: [Blog -- Neon Docs]() Topics: [migration](), [Databases](), [Replication](), [Transactions](), [Availability]() Tags: [agent](), [availability](), [database](), [migration](), [playground](), [postgres](), [product](), [replication](), [tools](), [transactions]() ### AI overview Neon Labs is a new home for experimental tools built around Lakebase Postgres. Its first release includes a Postgres Upgrade Assessment that identifies version-upgrade blockers and a Migration Assistant that recommends a migration method. The tools are intended for development or testing while they remain experimental. ### Source excerpt Today we're launching Neon Labs, a home for experimental tools built around Lakebase Postgres. We want Neon Labs to be a playground for ideas that could help the broader Postgres community. ## Just landed in the Neon CLI DevFeed: [Just landed in the Neon CLI]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-25T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Command-line interface](), [Databases](), [MCP](), [Agent Skills](), [API keys]() Tags: [agent-skills](), [agents](), [ci](), [cli](), [coding](), [database](), [diagnostics](), [mcp](), [product](), [tool]() ### AI overview An update on recent Neon CLI changes, including the shorter npm package name, agent-focused setup for MCP and skills, project and branch linking, environment-variable management, database diagnostics, dashboard access, and project-scoped keys. ### Source excerpt The Neon CLI moved fast over the last couple of months. It is a key tool for agents, who do all the work without ever opening the Console - let's take a look at some of recent changes. ## WAL + S3: Lakebase storage for the era of agents DevFeed: [WAL + S3: Lakebase storage for the era of agents]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-24T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Database](), [Amazon S3](), [Transactions](), [data](), [migration]() Tags: [agents](), [data](), [database](), [engineering](), [migration](), [product](), [s3](), [storage](), [transactions]() ### AI overview The article argues that PostgreSQL workloads associated with agents need storage built around transaction history rather than only the current state. It presents the write-ahead log as an existing timeline of database modifications and examines placing S3 beneath PostgreSQL to support isolated copies, point-in-time restoration, historical queries, and many disposable environments more efficiently. ### Source excerpt Agents treat code as lightweight: branch it, deploy it, throw it away. The infrastructure under that code should work the same way. Working with an OLTP database is traditionally heavy and full of friction, but very little of that is a Postgres problem. ## Open-weight models are fast on Neon AI Gateway. Here's why DevFeed: [Open-weight models are fast on Neon AI Gateway. Here's why]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-20T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [databricks](), [Inference Performance](), [model-deployment](), [Low-Latency Inference](), [Caching](), [Mixture of Experts (MoE)](), [gpt-oss](), [Large Language Model]() Tags: [caching](), [databricks](), [gpt-oss](), [inference](), [mixture-of-experts-moe](), [optimization](), [performance](), [product](), [speed]() ### AI overview The article explains why open-weight models can deliver low latency on Neon AI Gateway when paired with Databricks Foundation Model APIs and an optimized serving stack. It highlights continuous batching, KV-cache paging, custom kernels, quantization, multi-GPU layouts, prompt caching, and Mixture-of-Experts architectures. ### Source excerpt Open-weight models are built for performance, but how they're served determines whether end users actually feel that speed. Neon AI Gateway calls Databricks Foundation Model APIs for Databricks-hosted open-weight models, a stack that has years of inference engineering behind it. ## Query Neon backend logs DevFeed: [Query Neon backend logs]() Original publisher: [Read original article]() Author: Andre Landgraf Published: 2026-08-18T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Back end](), [observability](), [log management](), [Command-line interface](), [MCP Server](), [Serverless](), [Server](), [Node.js](), [Amazon S3](), [SDKs]() Tags: [api](), [backend](), [cli](), [logs](), [mcp-server](), [node](), [observability](), [product](), [s3](), [tool]() ### AI overview Neon adds branch-scoped access to backend logs from Neon Functions and Object Storage outside the Console. The logs can be queried through the CLI, MCP server, Logs API, SDK, or Loki endpoints, with filters and raw LogQL support. ### Source excerpt We keep expanding our backend observability - the most recent addition: you can now query backend logs outside the Console! neon logs reads what Neon Functions and Object Storage emit on a branch, with filters for source, severity, and message text. ## LLMs belong in your backend DevFeed: [LLMs belong in your backend]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-17T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Large Language Model](), [Back end](), [API]() Tags: [ai-gateway](), [api](), [api-keys](), [backend](), [llms](), [product]() ### AI overview Neon presents its AI Gateway as part of a backend stack for calling LLMs through a unified credential, endpoint, billing flow, and model catalog. ### Source excerpt Neon AI Gateway is how we put LLM calls into the Neon backend the same way we put files, functions, and auth there. It allows you to call models right from Neon, instead of collecting lab accounts. ## Neon skills landed in the Vercel CLI DevFeed: [Neon skills landed in the Vercel CLI]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-14T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [AI-assisted coding]() Tags: [agent](), [agent-skills](), [backend](), [cli](), [coding](), [database](), [deployment](), [integration](), [postgres](), [product](), [skills](), [vercel]() ### AI overview The article explains that installing a Vercel Marketplace integration through the CLI can also install provider agent skills. It uses Neon as an example: one command provisions a database, connects it to a Vercel project, makes connection variables available locally, and gives a coding agent instructions for using the database. ### Source excerpt Vercel just gave their CLI a new trick: when you install a Marketplace integration from the terminal, it also installs that provider's agent skills. A single command now provisions a Neon database on your Vercel project and hands your coding agent the instructions it needs to actually use it. ## Neon for Herdr: A Neon pane in your Herdr workspace DevFeed: [Neon for Herdr: A Neon pane in your Herdr workspace]() Original publisher: [Read original article]() Author: Dominik Koch Published: 2026-08-13T12:00:00Z Content type: release Language: en Sources: [Blog -- Neon Docs]() Topics: [SDKs](), [TypeScript](), [Terminal](), [dashboards](), [OAuth](), [Databases](), [autoscaling](), [Command-line interface]() Tags: [autoscaling](), [browser](), [community](), [databases](), [linux](), [macos](), [oauth](), [typescript]() ### AI overview Neon for Herdr is a terminal workspace plugin that adds a Neon project-management pane built with Ink, Effect, and the Neon TypeScript SDK. It supports project and branch operations, compute controls, connection-string retrieval, fuzzy search, and OAuth authentication with PKCE. ### Source excerpt I just released Neon for Herdr, a Herdr plugin that puts your Neon projects in a pane next to whatever you are already doing, so creating a branch, grabbing its connection string, or suspending a compute is only a keypress away. ## Neon Functions: backend logic next to your data DevFeed: [Neon Functions: backend logic next to your data]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-12T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Back end](), [Node.js](), [Database](), [Serverless](), [WebSocket](), [Cloud]() Tags: [backend](), [database](), [hosting](), [network](), [node](), [node-js](), [product](), [serverless](), [vpc]() ### AI overview Neon Functions provide long-running Node.js 24 backend compute deployed beside a Neon Postgres branch. They automatically receive database and related Neon service credentials, reducing network hops and supporting streaming, WebSockets, SSE, and persistent connection pools. ### Source excerpt Neon Functions are Node.js 24 compute you deploy onto a Neon branch, in the same region as your Lakebase Postgres database, with DATABASE_URL injected automatically. They're long-running enough that agents can stream for minutes and WebSockets or SSE can stay open while data flows. ## Electric is joining team Neon at Databricks DevFeed: [Electric is joining team Neon at Databricks]() Original publisher: [Read original article]() Author: Nikita Shamgunov Published: 2026-08-11T13:00:00Z Content type: news Language: en Sources: [Blog -- Neon Docs]() Topics: [databricks](), [Database](), [real-time](), [wasm](), [AI-assisted coding]() Tags: [agentic](), [company](), [postgres](), [real-time](), [sync](), [wasm]() ### AI overview Electric is joining Neon within Databricks, bringing PGlite and its real-time Postgres synchronization technology. The article presents these as primitives for lightweight applications and agent workloads. ### Source excerpt Today Databricks announced that Electric, the team behind PGlite and the Electric sync engine, is joining Databricks. We have our own reason to celebrate: Electric is joining the Neon team within Databricks, bringing lightweight WASM Postgres + real-time sync to Neon. ## Building Neon Object Storage DevFeed: [Building Neon Object Storage]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-08-06T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Amazon S3](), [Back end](), [Databases](), [cloud-infrastructure]() Tags: [architecture](), [backend](), [data](), [product](), [s3](), [storage]() ### AI overview Neon Object Storage is a beta, S3-compatible object store integrated into the Neon backend. Its defining feature is branch-aware storage: database branches inherit buckets and objects, while copy-on-write keeps branches isolated and avoids upfront duplication. ### Source excerpt We just launched Neon Object Storage, a branch-aware, S3-compatible object store built into the Neon backend. If you've used Neon branches before, you know the workflow: branch main into a child and get an isolated copy of your database in about a second. Object Storage adds files to that picture. ## Training 100x Cheaper Retrieval models Neon and Castform DevFeed: [Training 100x Cheaper Retrieval models Neon and Castform]() Original publisher: [Read original article]() Author: Pranav Aurora Published: 2026-08-05T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Retrieval Augmented Generation (RAG)](), [Training AI Models](), [Machine learning](), [Open Source Models & Datasets](), [Large Language Model](), [Open Source](), [Latency](), [Database]() Tags: [agentic](), [agents](), [cost](), [embedding](), [infra](), [llms](), [machine-learning](), [model](), [models](), [open-source](), [performance](), [post-training](), [product](), [rag](), [retrieval](), [rl](), [scale](), [search](), [systems](), [tools](), [training]() ### AI overview The article explains how Castform uses reinforcement-learning post-training to improve open-weight models for agentic retrieval. It contrasts multi-step retrieval with one-shot embedding search, emphasizing the cost and latency of repeated frontier-model calls and the potential for smaller open models to perform specific search tasks more cheaply. ### Source excerpt "Most teams' best training data is just sitting in their databases. The problem is that turning raw data into something usable is hard, and letting agents read, search, and mutate data cheaply at scale requires advanced infra. Pointing Castform at Neon skips both." -- Ying Hang Seah, cofounder, Castform ## Neon now has per-project permissions DevFeed: [Neon now has per-project permissions]() Original publisher: [Read original article]() Author: Russ Dias Published: 2026-08-03T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Authorization](), [Databases]() Tags: [agents](), [contributors](), [external](), [product](), [safety]() ### AI overview Neon introduces per-project permissions alongside four organization-level roles: Admin, Editor, Viewer, and Collaborator. The two permission layers combine to provide scoped access for users, agents, and external contributors, though project-level permissions currently only expand access and cannot restrict an organization-wide role. ### Source excerpt Neon now supports a new permission structure, with four org-level roles (Admin, Editor, Viewer, and Collaborator) and per-project permissions that let you control exactly which projects your agents and collaborators can access. ## CommSync Runs Text, Vector, and Hybrid Search on Postgres with Lakebase Search DevFeed: [CommSync Runs Text, Vector, and Hybrid Search on Postgres with Lakebase Search]() Original publisher: [Read original article]() Author: Carlota Soto Published: 2026-07-30T12:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Latency](), [Embeddings](), [Artificial Intelligence](), [App](), [Extension]() Tags: [ai](), [app](), [case-studies](), [embeddings](), [feature](), [features](), [latency](), [postgres](), [product](), [search](), [vector]() ### AI overview CommSync describes using Neon's Lakebase Search extensions on Postgres for lexical, semantic, and hybrid inbox search. The article reports substantially lower query latency than the previous GIN-based approach and explains a hybrid implementation combining BM25, embedding-based cosine similarity, and Reciprocal Rank Fusion. ### Source excerpt "We're running lakebase_text and lakebase_vector across three different parts of our product. Query latency improved 1,000x and we don't need a separate search stack sitting next to our Postgres." ## My agent cloned GitHub (sort of) using the Neon backend DevFeed: [My agent cloned GitHub (sort of) using the Neon backend]() Original publisher: [Read original article]() Author: Roman Zaynetdinov Published: 2026-07-23T09:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Git](), [GitHub](), [Hono](), [JavaScript](), [Database](), [Amazon S3](), [Serverless](), [web applications](), [Authorization](), [Back end]() Tags: [app](), [auth](), [backend](), [cli](), [code](), [database](), [git](), [github](), [javascript](), [object-storage](), [product](), [s3](), [server](), [serverless](), [web]() ### AI overview An experiment uses an AI agent and Neon's beta backend suite to build a working Git host without a filesystem. Git objects are stored in Postgres, large blobs are offloaded to Neon Object Storage, and a serverless Hono application serves Git over HTTPS. The implementation supports cloning, pushing, pulling, commit and file browsing, authentication, and testing against edge cases such as binary, large, empty, and oddly named files. ### Source excerpt GitHub has been down a lot lately. During one of those episodes, I had the idea of asking my agent to use our new backend suite (still in beta) to build a working Git host. This is genuinely outside of my comfort zone as I had no prior knowledge of Git server internals, so I didn't have much hopes, but it went surprisingly well... ## Debug your Postgres from the terminal: a tour of \`neon inspect db\` DevFeed: [Debug your Postgres from the terminal: a tour of \`neon inspect db\`]() Original publisher: [Read original article]() Author: Savannah Longoria Published: 2026-07-22T18:00:00Z Content type: article Language: en Sources: [Blog -- Neon Docs]() Topics: [Database](), [telemetry](), [SDKs]() Tags: [cli](), [debug](), [diagnostics](), [json](), [observability](), [performance](), [postgres](), [product](), [scripting](), [sql](), [statistics](), [tooling]() ### AI overview Neon introduces `neon inspect db`, a Neon CLI command suite for read-only Postgres diagnostics. Its subcommands run bounded, versioned queries against Postgres statistics and catalog views, returning tables or JSON/YAML without requiring users to assemble connection strings or write catalog queries. ### Source excerpt The Neon CLI now ships `neon inspect db`: read-only, high-signal Postgres diagnostics that answer "what is slow?" without leaving the terminal or writing a single catalog query. Each subcommand runs one known-good query against Postgres' own statistics and catalog views and prints a clean table (or JSON/YAML for scripting). [Next page]()