§ WHAT'S DIFFERENT

There's more than one shape of AI platform.

Different shapes suit different missions. We think Writ fits a particular kind of customer — organizations where data sovereignty, security, and independence shape every decision. Here's how Writ compares to the other options, without naming names.

§ THE SHORT VERSION

What's unique.

Writ is the only sovereign AI backend built for vibe coding against mission-critical data. Each of the four shapes below solves a slice of the problem. None of them let a small team stand up an accredited, auditable application at the speed the mission demands — without leaving their own infrastructure.

Public clouds move fast but can't follow the data. Appliances are turnkey but lock the stack. Microservice catalogs are flexible but push integration work onto the mission team. Roll-your-own produces the thing you want — two to three years from now. Writ is the assembled result: one contract, one audit trail, one fence.

§ FOUR CATEGORIES

The shapes of AI platform you'll encounter.

Public cloud AI services

THE SHAPE

You send your data to a large public cloud. The provider runs the models. You pay per use. It works beautifully when you can send your data to their cloud.

WHEN IT FITS

For workloads that don't touch sensitive, regulated, or sovereign data, and where the cost curve works out, public cloud AI is often the right answer.

WHEN WRIT FITS

When your data can't leave your network, when you need to deploy into places the public cloud doesn't reach, or when regulatory posture requires the inference itself to happen on your side of the fence, Writ is the option.

Commercial AI appliances

THE SHAPE

A single vendor sells you an appliance — hardware, software, and support bundled together. It comes with a subscription and is supported only on the vendor's own gear.

WHEN IT FITS

When you want a turnkey experience and are comfortable with a single-vendor relationship and the subscription that goes with it, an appliance is straightforward to buy.

WHEN WRIT FITS

When you want to run on hardware you already have, avoid vendor lock-in, and own the full stack — including the right to modify it, re-distribute it internally, and keep running it if circumstances change — Writ is the option.

Microservice catalogs with a gateway

THE SHAPE

A catalog of specialized AI services, each with its own shape, behind a gateway layer that routes between them. Flexible — and the flexibility is the product.

WHEN IT FITS

When a team has the engineering depth to compose services, standardize auth and audit across them, and maintain the glue — a catalog-plus-gateway assembly is powerful.

WHEN WRIT FITS

When the mission team wants one login, one audit trail, and one toolkit that covers every kind of AI out of the box, Writ is the option — because we did the stitching once so every customer doesn't have to.

Roll-your-own open source

THE SHAPE

Pick the components yourself. Stitch them together. Do the security approval work, the supply-chain work, the post-quantum work, the documentation. Own it end to end.

WHEN IT FITS

For a team with years of runway, strong platform engineering, a dedicated security function, and a mandate to build rather than buy — rolling your own produces exactly the thing you want.

WHEN WRIT FITS

When you want the openness of open source, but not the two-to-three-year integration project, the security-approval package, and the post-quantum wrapper — Writ is the assembled result. Maintained. Signed. Documented.

§ HONEST ANSWERS

Questions we hear a lot.

"We already use a public cloud AI service. Why change?"

You probably shouldn't, for the workloads where that service fits. The question is whether it fits the next workload — particularly if that workload touches sensitive data, regulated information, or an environment the cloud can't reach. That's where the fit question actually lives, not in a blanket displacement argument.

"Is Writ ready for our most sensitive environment today?"

Writ is pre-release. It is architected for high-sensitivity deployment and ships with a full security package. Whether it's 'ready' for a specific environment depends on your accreditor, your existing baselines, and your specific data — not on anything a marketing page can promise.

"Why would we roll our own instead?"

Some organizations have the engineering depth and the mandate to build. If that describes your team and your time horizon works for it, rolling your own is a legitimate path. What Writ saves you is the two-to-three-year integration project — and the ongoing work of keeping the pieces aligned.

"We're a small team. Can we run this?"

Yes. Writ is designed so that a team of two to three engineers, plus an integrator partner, can stand up and run the full platform without dedicated Writ staff. The open-source model is deliberate: you shouldn't need us to keep going.

"What happens if Writ disappears?"

Nothing urgent. The platform is 100% open source. The source code ships with every release. If we disappeared tomorrow, you could keep running the version you have, patch it yourself, and migrate at your own pace. That's the point of the open-source commitment.

"How do you stack up on a specific feature?"

The architecture and limitations are documented in the white papers. Every platform has strengths and weaknesses — we'd rather you read the actual reference than squint at a comparison matrix.

§ READ NEXT

Curious whether Writ is the right shape for your mission?

Read the Papers Journal