AI integration

AI features that do real work in your product: writing, summarising, answering and recommending. Built with checks on accuracy and a ceiling on cost.

What's included

How it works

  1. Choose

    We pick one task where AI will clearly help, and define what a good result looks like.

  2. Prototype

    We build a working version on your real data, so you can judge the output before committing.

  3. Harden

    We add the checks, limits and fallbacks that make it safe to put in front of customers.

  4. Launch and watch

    We release it, track quality and cost, and adjust.

Common questions

Will the AI make things up?

Left alone, it can. That is why we give it the facts to work from and check what it writes against them. Our own product, FFSN, publishes AI-written articles this way.

Which AI models do you use?

We mostly build on Anthropic’s Claude models, and choose the model by the job. Simple tasks run on smaller, cheaper models.

How much does it cost to run?

It depends on how much text goes in and out. We design for cost from the start, set spending caps, and show you cost per feature.

Is my data used to train the AI?

That depends on the provider and the plan. We tell you which service each feature uses and what its data terms are before anything is built.

Other services

Have a project in mind?

Tell us what you're building and we'll come back with next steps.

Start a project