DFM Platform

Dust: AI Agent Orchestration and Secure Enterprise Data Integration

Technical analysis of Dust’s agentic AI framework. Mapping multi-agent orchestration and secure data integration for enterprise and dual-use C4ISR

14 pages · PDF · 26 December 2025 · Licensed single-user copy, watermarked to the buyer

Tactical Generative Models Large Language Models

Cover of Dust: AI Agent Orchestration and Secure Enterprise Data Integration

€299 excl. VAT — EU VAT calculated at checkout (VAT ID accepted for reverse charge); invoice issued after payment

One click to Stripe — guest checkout, no account. Your download appears on the confirmation page and arrives by e-mail right after payment (link valid 72 hours, up to 5 downloads).

Read the free summary of this report →

About this report

Dust is a Paris-based deep-tech startup building a secure AI “operating system” for enterprises, enabling automated agents to access and act on internal corporate data and workflows .

Founded in 2023 by former Stripe and OpenAI engineers (Gabriel Hubert and Stanislas Polu ), Dust’s platform lets companies spin up specialized AI assistants integrated with tools like Slack, Salesforce and GitHub, processing knowledge bases without manual effort . This vision – moving beyond standalone chatbots to collaborative, task-oriented AI agents – positions Dust at the forefront of the enterprise AI revolution.

Key questions this report answers

Inside this report

Who it's for

Investors screening Dust, competitors and partners assessing their position, and analysts who need a sourced, structured account of the entity and its technology.

Methodology & sources

DFM reports are built from primary and official sources — TED procurement notices, CORDIS and the EU Funding & Tenders Portal, EIB operations, the NATO Innovation Fund portfolio, SIPRI data, official budget documents and company disclosures — read together with the underlying legal texts. Sources are cited in the document; the report reflects them as of its publication date (26 December 2025).

Format & delivery

14-page PDF, watermarked to you (name, e-mail, order number on every page). Delivered immediately after checkout on the confirmation page and by e-mail; the personal link is valid for 72 hours and up to 5 downloads (re-issued on request). Guest checkout — no account required. Single-user licence: see the Terms of Sale.

The full analysis, not the summary

The public briefing linked above tells you what this report found; the report itself shows the full evidence and reasoning behind it, section by section. It is the complete, dated document from the DFM research desk — sources cited, delivered as a licensed PDF you can keep and cite in your own work.

Get the full report — €299 →

Related on DFM

More Tactical reports · Generative Models · Large Language Models · Free summary of this report · All reports

Prefer unlimited access?

Prefer unlimited access? Every report like this is included in the DFM Analysis subscription. See plans →