# Akumi solutions: pick the problem you are solving

Meeting a residency or GDPR obligation, or shipping a chatbot, a retrieval app or an agent: the same EU-resident platform sits underneath all six.

## AI you can answer for.

Whether you are meeting a residency or GDPR requirement, or shipping a chatbot, a retrieval app or an agent, the same EU-resident platform sits underneath. Pick the problem you are solving.

## The same platform, pointed at a different problem.

The two columns are the two people who end up on this page. One has to sign the thing off and one has to ship it, and neither reads the other's half.

### If you sign it off

- GDPR compliance: Residency, pseudonymization and erasure in one place.

- EU data residency: Data that never leaves the EU, and inference that leaves only when you allow it.

- EU AI Act: Records for the obligations as they take effect.

### If you ship it

- RAG applications: Answers from your own documents, with sources.

- AI agents: Tools and memory an auditor can follow.

- Chatbots and assistants: Customer-facing AI with the guard rails on.

## One platform underneath, whichever door you came through.

The six pages differ in what they lead with, not in what runs. Every one of them is the same EU-resident inference behind one OpenAI-compatible endpoint, with the Firewall, retrieval, the cache and a metadata-only audit trail available on the same request. Nothing is a separate product with its own bill, and nothing is a package you have to be sold.

## The parts that do not vary.

Whatever brought you here, these hold. They are the reason the six pages can be short.

### EU-resident by default

The application, the data and the default models are EU-resident. Routing to anything outside that set is blocked at a guard that fails closed unless you explicitly allow it.

### Provable per request

Every call records the model, the provider and the region that served it, in an audit trail that holds metadata and never prompts or completions.

### Metered per use

Base inference and each service meter on their own line when they run. Nothing runs, nothing costs, and the prices are published.

### One endpoint

An OpenAI-compatible API, so an existing client moves by changing one base_url and the key. First-party SDKs for PHP, TypeScript and Python.

## Questions people ask before they pick.

### Which of the six should I start with?

The one that names the reason you are here. They lead to the same platform, so starting on the wrong one costs you a page, not a decision. If an obligation brought you here, start on the left; if a feature did, start on the right.

### Are these separate products?

No. They are six framings of one platform. There is no solution SKU, no package and no separate bill: you get the same API and the same services, and you switch on the ones your case needs.

### Can one project be more than one of these?

Usually it is. A retrieval app inside a firm with a GDPR obligation is both, and nothing about the platform changes between the two readings.

### Do I have to talk to sales?

No. Signup is self-serve and pricing is published. You need a payment method before running inference, and then a key and one base_url gets a first response in minutes.

### What if my case is not one of the six?

The platform pages describe the nine services directly, which is the better route when your case does not have a familiar name. Tell us what you are building and we will say plainly whether it fits.

## Start with the problem you have.

Add a payment method, create a key, and build on EU-resident AI. Or tell us the requirement you have been handed and we will say whether this meets it.