PGlite實時同步進Neon

2026年8月18日 00:00
站內 AI 整理稿

Back to Blog/CompanyElectric is joining team Neon at DatabricksTogether we're building the complete backend for agentic apps.Nikita Shamgunov–Co-founder of NeonAug 11, 2026Subscribe to our changelogReceive only our latest updates.No spam, guaranteed.

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.

The two teams are already at work on new backend capabilities, and we can't wait to see the apps and agents our customers build with them.Why Electric Electric built two things in particular that have become key primitives in the agentic era.

The first is PGlite, a full Postgres database small enough to run inside the application itself.PGlite runs in sandboxes, in embedded apps, and in serverless environments and it has grown from 1M to 13M weekly downloads in the last year.The second is a sync engine.

It keeps data continuously synchronized between a central Postgres database and connected clients like browser tabs, mobile apps, or agents running in a sandbox.

Sync engines power collaborative products like Google Docs, Figma, and Notion, and the same technology is critical to keeping state synchronized across fleets of agents.Both are Postgres to the core, which is why they fit in so well at Neon.

PGlite: Postgres everywhere More software will be built over the next few years than in all of history.As coding agents drive the cost of creation to zero, the number of applications explodes, and most of them are small.

You can't have an age of abundance if every app requires a fixed minimum of compute.In our very first post announcing Neon, we called this out: You have to make the cost of running the service an important design consideration on par with manageability, reliability, and performance.

This has proven to be more true than we could have ever imagined.PGlite pushes in the same direction.A database that runs inside agent sandboxes and browser tabs, synchronized in real time with Lakebase Postgres in the cloud, lowers the entry cost for cloud apps and agents.

Software abundance needs innovations like PGlite: millions of small, fast applications, each with real Postgres underneath.Sync: Realtime that agents can use Complex problems like conflict resolution, partial replication, and reconnection logic make real-time sync difficult to build from scratch.

Now agents are writing most of the code, and the problem persists.Ask a coding agent to build a CRUD app and it succeeds.Ask it to implement real-time sync and the failure rate goes way up.

But luckily for us, Auth and ORMs have shown the way: give agents a primitive so good they never have to improvise.Better Auth did this for auth.Drizzle did it for ORM.Real-time sync needs the same treatment, a primitive that agents reach for by default and get right on the first try.

The Electric team has spent years building exactly that, and it slots into Neon next to the primitives already on the platform: Lakebase Postgres, Auth, Storage, Functions, and the AI Gateway.Shared DNA There's also another reason this is a good fit.

PGlite began as a proof of concept by Neon co-founder Stas Kelvich, exploring Postgres compiled to WASM.Electric took that experiment and turned it into a database that millions of projects build on every week.Bringing the teams together closes a loop that started years ago.

More than the shared history, though, we share an ethos.Electric is a team of Postgres hackers.They take a 30-year-old database we all love and pull it into the future, connecting it to how the world builds apps today.

That's the same instinct that produced branching, scale-to-zero, and instant provisioning at Neon.We're glad to have them.Read the Databricks announcement and hear from Electric's founders on why they joined.

Share:More from NeonProductAug 17, 2026LLMs belong in your backendProductAug 14, 2026Neon skills landed in the Vercel CLICommunityAug 13, 2026Neon for Herdr: A Neon pane in your Herdr workspace

Related

相關文章

IT之家AI Agent

消息稱 NVIDIA 考慮向 Perplexity AI 投資“數十億美元”

作者:溯波(實習) 責編:溯波 評論: 8 月 24 日消息,外媒 The Information 稍早前報道稱,NVIDIA(英偉達)考慮在 Perplexity AI 的最新融資輪中向這家人工智能初創企業投資“數十億美元”。雙方正就該交易展開磋商,還可能達成技術授權協議。

剛剛
AIBaseAI Agent

Guidelight評估五大AI實驗室,OpenAI遏制能力排名第一

該評估覆蓋Anthropic、Google、OpenAI、Meta和xAI,僅依據公開信息考察其是否建立監控、異常行為處置、第三方審計及失控模型關閉等機制。在滿分5分的評估中,OpenAI以3分排名最高,Anthropic和Meta得分最低。

8 小時前7400
何夕2077AI Agent

τ_0-VLA長程操控

τ₀-VLA是一種層次化機器人基礎模型,透過世界模型引導的測試時計算來改善長程操控任務。高層策略在決策不確定時會額外分配計算資源,搜尋替代子任務並預測其視覺結果,而低層策略則在40,115小時的異質真實世界數據上訓練。在包含13到25個步驟的真實長程任務中,封閉迴路成功率從27.5%提升至45.0%。

10 小時前
何夕2077AI Agent

智能體技能合集

VoltAgent 的智能體技能庫近期持續擴充,目前收錄的技能規模已突破千項,相關倉庫在開發社群中獲得約 31.3k 的星標關注。這個持續成長的技能庫,正逐步成為開發者快速組合與部署智能體工作流程的重要資源。 該技能庫涵蓋多種類型的 CLI 工作流程,這些流程具備高度可複用性,讓開發者不必從零開始建構每個環節,而是能直接引用既有技能來加速專案進度。隨著技能項目不斷增加,VoltAgent 生態系的應用範圍也隨之擴大,從自動化任務到複雜的指令處理,都能找到對應的現成模組。

10 小時前
何夕2077AI Agent

ruflo蜂群代理

近期在AI資訊日報中受到矚目的「ruflo蜂群代理」,是一款以多智能體框架為核心的代理管理工具。其設計重點在於讓開發者能夠有效管理複雜的代理流,透過系統化的方式協調多個AI代理協同運作,而非僅是單一任務的自動化處理。這種架構特別適合需要分工、接力或平行處理的應用場景,讓整個執行流程更為清晰且可控。 值得注意的是,ruflo導入RAG(檢索增強生成)記憶機制,讓代理在運作過程中能將經驗與知識「沉澱」下來。也就是說,代理不只是每次從頭開始執行,而是可以從過往的互動或任務中提取相關資訊,作為後續決策的參考依據。

10 小時前