Dust partner: Idoption deploys your AI agents in production

Choose a partner that can prove deployments in production, with figures validated by its clients. Four checks are enough: published case studies, some of them hosted on dust.tt; adoption that is measured rather than estimated; governance set up from the start, with inherited permissions, traceable requests and European hosting; and a provider able to say when Dust is not the right fit. Dust lists three partner types: services, resellers and integrators. At Idoption, two deployments are documented by case studies published on dust.tt, those of Applium and Apiday. The 45-minute diagnostic then checks whether your use case deserves to be deployed, before any commitment. A costed action plan, delivered within two weeks, lets you compare offers on the same basis. Ask the same of any provider before you sign anything: verifiable references and sourced figures.

Four weeks in general, from zero to autonomous teams. The 45-minute diagnostic leads to a costed action plan within two weeks. The first agent is in production in the first week, on your real data. Weeks two to four are for adoption: we bring in the most motivated users, measure usage in real time and adjust the agents based on feedback from the field. The target is 90% adoption on day 29, tracked on dashboards. At Applium, 97 of 120 users were connected and registered twenty-nine days after deployment, with 60% regular users. Each week has its deliverables: kick-off and hackathons in week 1, version 2 of the agents in week 2, a Flash ROI Diagnostic in week 3, keynote and presentation of the return on investment in week 4. Nothing built during the pilot is thrown away, and the full method details the program.

Use case by use case. We track time saved, success rate and volumes handled, and usage can be read in real time on dashboards. At Apiday, prospect enrichment for a campaign went from seven days to two hours, which is 97% less, and the sales team uses the agents every day. At DataValue Strategy, meeting notes save about 70% of the time and are delivered the same day. At Applium, automatic meeting notes save 20 minutes to 1 hour per meeting. A Flash ROI Diagnostic in the third week allows adjustments before the final presentation. These figures are provided by the clients and detailed in each case study. If the return on investment is not there, we do not deploy: a systematic approach also means recognizing the cases where AI does not bring enough value to justify the project.

The agents inherit the permissions of your tools: each person reaches the data they already have the right to see. Every request is traceable and hosting is European. We set up governance before the pilot expands: security frames the deployment from day one. The agents work on your real data, in a production environment, connected to tools such as Teams, SharePoint, Slack, Google Drive or SAP. For Dust's contractual commitments on hosting and data use, the reference remains the documentation published by Dust and its terms: check them before signing, whichever provider supports you. Security is a starting condition of the pilot. You also choose which models are allowed on sensitive data. The MCP protocol also connects your systems directly to AI models for tailored uses, and the pilot runs on your real data rather than on a test dataset.

It depends on your tools, your data and the use case, and the diagnostic settles it. We are vendor-agnostic: Partner of Dust, Microsoft, OpenAI and Anthropic. Consultants certified on Claude, Dust and Microsoft Copilot. We select the solution that fits your situation, because every company is different. Many organizations already use several tools, with Copilot in one department and other solutions elsewhere, with no shared governance and no measurement of usage. Our work therefore starts by looking honestly at what you already have: your processes, your tools, your teams. If a use case is not worth it, we say so, even if that means we have nothing to sell you. That is why the first conversation is a diagnostic and not a sales proposal.

Idoption is an independent company and a Dust partner. Consultants certified on Claude, Dust and Microsoft Copilot. It is also a partner of Microsoft, OpenAI and Anthropic, and picks the tool according to the use case. Idoption SAS is based in Paris and was created on 9 November 2025 by Baptiste Romé and Dimitri Mechai, whose experience spans SAP, IFS, Payfit, Dust, Dataiku and Salesforce. Dust publishes two case studies written with Idoption, the Applium one and the Apiday one, which you can read on dust.tt. You can also check the partner status with Dust, through its partner page. Idoption SAS is registered under SIREN 994300853, so it can be checked in the official registers. Its offer covers diagnostic, activation, training, Anchoring, performance and development. The founders and the rest of the team are introduced on the team page.