# Better thinking, not bigger machines.

Efficient language models, made in Europe.

[Request access](mailto:hello@placeholder-ai.eu?subject=Access%20request) · [See the API](#api)

- Hosted in the EU only
- No US subprocessors
- GDPR by design
- Built for the EU AI Act
- OpenAI-compatible API

## One model family, four ways in.

### API

Compatible with the OpenAI Responses API. Change the base URL and keep your code.

### Chat

A chat app for your whole company, with your documents and only our models.

### Agents

Agents that work through tasks with tools, with a full audit trail.

### Coding IDE

Code completion that never sends your code outside Europe.

## Made in Europe, for Europe. Without a single US subprocessor.

### Sovereign by design

None of your data is subject to the US Cloud Act, and no foreign government can switch our models off.

GDPR · EU AI Act

[Why sovereignty matters](https://www2.placeholder-ai.eu/sovereignty)

### Lower cost

Smaller models need less compute, so we can offer them for less than comparable models.

[See the savings](https://www2.placeholder-ai.eu/cost)

### Per-turn billing

We count model turns, not tokens, so cost is easy to plan.

### Cloud

Use our models through the API, hosted on European infrastructure only.

### On premise

For security-critical work such as defence, we also deploy on your own hardware.

### No training on your data

We don’t use your prompts or outputs to train our models.

### Data security

GDPR is supported by design, and we are working on our ISO certifications.

### EU AI Act

Our models are built with the requirements of the EU AI Act in mind from the start.

### Training data

Sourced in line with EU copyright and text-and-data-mining rules.

## One line to switch.

Our API is compatible with the OpenAI Responses API. If you already use it, testing our models takes a single line change.

[See all models](https://www2.placeholder-ai.eu/models)

```
import OpenAI from "openai";
const client = new OpenAI({
  baseURL: "https://platform.plhldr.ai/v1",
  apiKey: process.env.PLHLDR_API_KEY,
});
const response = await client.responses.create({
  model: "tbd-0-gp",
  input: "Write a one-sentence bedtime story about a unicorn.",
});
console.log(response.output_text);
```

## More efficiency instead of bigger models

Instead of trying to compete with frontier models in terms of size, we are working on building more efficient models.

[How it works](https://www2.placeholder-ai.eu/technology)

### Expert models

Smaller models trained for one domain. For work inside that domain, they beat general-purpose models.

### Retrieval models

Knowledge sits in a retrieval layer rather than in the weights, so it can be updated without retraining.

### Distributed MoE

Experts run across several machines, and a request only activates the ones it needs.

### Constant context size

The context window stays the same size however long a conversation gets, so context rot and the “lost in the middle” problem don’t occur.

## The founders

### Denise Harrison

Psychologist (M.Sc.), specialised in learning and memory. For the last few years she has worked with European companies on how they operate. She runs Placeholder AI’s business: operations, strategy and go-to-market.

[LinkedIn](https://www.linkedin.com/in/denise-harrison) · [denise@placeholder-ai.eu](mailto:denise@placeholder-ai.eu)

### Nicklas Bocksberger

Previously Head of Data & AI at a European consultancy, where he watched the economics break once pilots hit production volume. He designs and builds the memory and reasoning architecture.

[LinkedIn](https://www.linkedin.com/in/nicklas-bocksberger) · [nicklas@placeholder-ai.eu](mailto:nicklas@placeholder-ai.eu)

## Try our model on your own use case.

Your data stays in Europe the whole time.

[Request access](mailto:hello@placeholder-ai.eu?subject=Access%20request)