Prisma Next – data contracts, migration graphs, agent DX
Hey I'm Will from the Prisma team, engineering manager and also the lead developer on Prisma Next. I'd like to introduce you all to the next version of Prisma: a full rewrite in TypeScript that builds on the established patterns in Prisma and comes with a family of skills that integrate it into whatever AI tooling you're using in 2026. (Read the announcement on our blog here: https://pris.ly/pn-ea) The three topics in the title are brand new concepts in Prisma Next so let me give you a quick rundown. 1. Data contracts If you've used Prisma before you've seen the…
In plain words
Prisma Next is a complete rewrite of the Prisma database toolkit in TypeScript, designed for developers building applications with database layers. It introduces data contracts that define relationships between applications and databases, migration graphs for tracking schema changes, and integrations with AI agents to improve developer experience. The tool is built for teams using AI tooling in their development workflows and maintains compatibility with Prisma's established patterns while adding new capabilities for modern development.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hey I'm Will from the Prisma team, engineering manager and also the lead developer on Prisma Next. I'd like to introduce you all to the next version of Prisma: a full rewrite in TypeScript that builds on the established patterns in Prisma and comes with a family of skills that integrate it into whatever AI tooling you're using in 2026. (Read the announcement on our blog here: https://pris.ly/pn-ea) The three topics in the title are brand new concepts in Prisma Next so let me give you a quick rundown. 1. Data contracts If you've used Prisma before you've seen the `schema.prisma` file. That's a contract between your application and your database: it spells out exactly what your app relies on and what your database promises to persist. Your contract drives everything in Prisma Next, like queries, migrations, error checking and correctness guarantees. It's hashed to give it an identity similar to a git commit and we use that hash to sign your DB: if the DB is signed with your contract's hash, your app knows it's compatible. 2. Migration graphs, guardrails and verifications A migration transforms your database from one contract's structure to another. That makes migrations the edges in a graph of data contracts. In Prisma Next we don't store migrations as a sequence of alphabetically ordered SQL files that break the second you merge two branches together. Instead, we store them as a graph, similar to how git stores your filesystem changes as a graph of commits. And we don't store migrations as black box SQL files. Each migration's on-disk structure is an `ops.json` file with a reviewable TypeScript file alongside it. Each operation has a precheck (verifies preconditions before executing) and a postcheck (verifies successful execution and provides idempotency). The migration promises to move you from contract A to contract B, and the resulting schema is verified against the destination contract before the migration is committed. They're the safest migrations you'll ever write and they're generated for you when you change your contract. 3. Agent DX Data contracts and verified, hashed, deterministic migrations are strong enough primitives to safely delegate this work to an agent. Prisma Next helps pry the footgun out of our collective fingers and gives you a set of reliable, predictable tools and also a family of curated skills with guardrails on every operation. This helps you focus on building your app instead of double-checking everything your agent does. Today we're releasing it in Early Access. It's early, there are gaps, but it's a different shape of ORM than you're used to and we want to get feedback to help us shape its future direction. The blog post goes into more detail than I have here but the best way to get a feel for it is to run `npm create prisma@next` or just clone the repo and check it out.
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, May 2026
the whole month →
Parallel agents, diff reviewer, and multi-model comparisons
Dev tools · May 2026 · kilo.ai
- NW
Hey HN, Henry here from Cactus. We open-sourced Needle, a 26M parameter function-calling (tool use) model. It runs at 6000 tok/s prefill and 1200 tok/s decode on consumer devices. We were always frustrated by the little effort made towards building agentic models that run on budget phones, so we conducted investigations that led to an observation: agentic experiences are built upon tool calling, and massive models are overkill for it. Tool calling is fundamentally retrieval-and-assembly (match query to tool name, extract argument values, emit JSON), not reasoning. Cross-attention…
Life & fun · May 2026 · github.com
- FM
Dev tools · May 2026 · github.com