How soon do we see results?

Within a few weeks. We start with one use case and a success criterion set at the outset. Your teams judge the result on their own documents before any wider commitment.

Can you work with sensitive data?

Yes. Your data goes only where you decide: your cloud, a European cloud, your servers, or an air-gapped environment. Our engineers work only with the access you grant them. When data must not leave, we use open models hosted in your environment.

How do you check the AI's answers?

Every answer cites the document and page it comes from, so your teams can open the source and check. When the information is missing, the agent says so.

Our engagements are billed on a time-and-materials basis or at a fixed price per work package.

Why choose you over a large systems integrator?

The engineer you meet first is the one who builds your system. Code and documentation are delivered in your repositories, so your teams can take over the system without us.

AI that acts. On your terms.

Our engineers build AI agents inside your applications, on your data and under your rules, then deploy them where you choose.

Talk to an engineer See what we do

Trusted by

Our engagements

How can we help?

Six ways to put AI to work. Every engagement is delivered by our own engineers, from scoping to production.

Forward Deployed Engineers
An engineer or a full squad works inside your tools, under your rules, through to production.
AI engineer as reinforcement · Project squad · AI tech lead
Learn more
AI in your applications
Your teams keep their tools. The agent works inside them.
Connectors and MCP · Automation without an API · Ask your data questions · Automated tests
Learn more
Agentic platforms
Teams of agents that draft reports, tender responses and regulatory filings from thousands of your documents.
Report drafting · Tenders · Regulatory filings · Long-running processing
Learn more

Hosting and models

Your choice of cloud, your choice of models

Azure, AWS or Google Cloud, with the models you already use. Your agents run in your own environment, under your identity and access rules.

European hosting, for example with Scaleway or OVHcloud, with European open models, when your data must stay in Europe.

On your servers, including in an air-gapped environment isolated from any external network. We install open models there and maintain them with your teams.

Research

Open research that ships.

1.4 million+

Downloads of our open models and datasets on Hugging Face

34 open models and 21 open datasets

Most downloaded

Source: Hugging Face API. downloadsAllTime total (1,491,436) across the paloalma and racineai organisations as of 19 September 2026; per-repository counters as of 12 September 2026. Updated quarterly.

See our research

2026
TW3Cast
A time-series forecasting system, 3rd out of 130 on the GIFT-Eval leaderboard by mean MASE rank as of September 14, 2026, with no agent and no language model.
N. Thierry, A.-L. Rochet
2026
Scoring Without the Engine
A deterministic content score for AI engines, validated end to end: an attacker who games it gains at most 6 points on a 500-source benchmark.
E. Bajemon, A.-L. Rochet
Big Data 2025
CU-1: a detection transformer for computer-use agents
An open, class-agnostic UI detector that beats Microsoft's OmniParser V2 on the WebClick benchmark (70.8% vs 58.8%).
A.-L. Rochet, M. Despujols, L. Appourchaux, N. Brandolini, D. Soeiro-Vuong, P. Lemaistre, G. Réus, B. P. Bhuyan

Our data, reused by

Jina AI (by Elastic)
Jina AI trains its Jina-OCR-v1 document parser on our VDR multi-domain dataset.
Source: arXiv 2609.03181, section 4.1 and table 2
Aalto University
38% of Aalto University's DistilVDR training images come from our open datasets (454,000 of 1.2 million).
Source: arXiv 2608.10636, appendix B
HKUST / Alibaba Cloud
Our VDR datasets are listed among the reference training sets in the HKUST / Alibaba Cloud survey of visual document retrieval.
Source: arXiv 2602.19961 v2, appendix D