IgnitionRAG vs Dust
Dust is a polished, well-funded French agent platform. IgnitionRAG is a dev-first RAG backend, IA gérée par la plateforme, billed by usage.
Dust (Paris, well-funded) is a 'multiplayer' enterprise agent platform with good connectors, an API and MCP support. It's a credible French competitor on sovereignty. The differences come down to the model: Dust is proprietary, per-seat and bundles the models; IgnitionRAG is dev-first, IA gérée par la plateforme and billed by usage.
The differences that matter
Dust bills per user (€29/mo on Pro) and bundles the models (not IA gérée par la plateforme); EU hosting is reserved for Enterprise (from 100 members). IgnitionRAG is platform-managed AI with bounded credits, hosted in France from the start, with an SDK, configurable RAG pipelines and natural-language building.
Who is each for?
Pick Dust for a ready-made collaborative agent workspace, per seat. Pick IgnitionRAG if you're a technical team, agency or IT services firm wanting a IA gérée par la plateforme RAG backend, integrable via SDK/API/MCP, hosted in France and deployable dedicated.
At a glance
Public sources, June 2026. Dust is a strong French player; this compares the model, not the quality.
| IgnitionRAG | Dust | |
|---|---|---|
| Model | Managed SaaS + dedicated/on-premise (Enterprise) | Proprietary SaaS (no self-host) |
| Data hosting | France from the start; dedicated/on-premise | EU hosting reserved for Enterprise (≥100 seats) |
| Pricing | Solo €59/mo or Team €249/mo, credits included | €29/user/month (Pro) |
| Platform-managed AI | Yes, no markup on tokens | No, models bundled/billed by Dust |
| Configurable RAG | Chunking, hybrid, reranking, eval exposed | RAG over connected sources |
| Natural-language workflows | Yes (in-builder agent + MCP) + visual builder | Configurable agents (no workflow builder) |
| Microsoft 365 connectors | Native (SharePoint, OneDrive, Teams…) | Generic connectors (Drive, Notion, Slack…) |
| API / SDK / MCP | REST API, TS/Python SDK, MCP | REST API + MCP (no dedicated SDK documented) |
| Governance | Native: audit, GDPR, AI Act, IA gérée par la plateforme | SOC2, GDPR, no training on your data |
FAQ
Are Dust and IgnitionRAG both French?
Yes. Dust is in Paris, IgnitionRAG in Troyes. The difference isn't origin but model: Dust is proprietary, per-seat, models bundled; IgnitionRAG is dev-first, IA gérée par la plateforme, usage-based, hosted in France.
Can I bring my own LLM keys?
With IgnitionRAG, yes (IA gérée par la plateforme, no markup on tokens). Dust bundles and bills the models in its plans, which is not a IA gérée par la plateforme model.
See your first enterprise AI assistant come to life
Book a 30-minute demo. We'll map your documents, your use case and the fastest path from enterprise knowledge to a production assistant.