Project context
In early 2024, Mistral AI had become France’s leading AI company.
While its models were best in class, its digital ecosystem (Le Chat, La Plateforme, and Docs) showed fragmented UX and inconsistent navigation.
No full-time designer was yet in place.
I initiated a 3-day research sprint to demonstrate how low-cost, open-source user research can uncover concrete UX opportunities without recruiting users or using paid tools.
OBJECTIVES
Exploring low-cost research methods to identify real UX opportunities
Identify critical usability gaps in Mistral’s ecosystem without traditional research resources.
Deliver actionable design directions to strengthen developer adoption and engagement.
Establish a repeatable low-cost framework for fast UX discovery.
KEY FINDINGS
Merging heuristics and open-source intelligence for actionable insights
01. UX/UI Audit
My audit revealed numerous inconsistencies across Mistral’s ecosystem, notably:
Inconsistent dark/light mode components.
No navigation continuity between Le Chat and La Plateforme.
Duplicated categories (for example “API”).
Impact: fragmented experience and unclear mental model.
02. Mistral’s User Feedback Found on Discord
Developers wanted:
Advanced model testing (temperature, top-p, penalties).
Better conversation management (folders, export, history).
A token counter and easier Function Calling UX.
File upload support (.txt, .md, PDF).
→ Le Chat was perceived as a demo, not a true Playground.
03. OpenAI’s User Feedback taken from OpenAI Forum
Showed similar early-stage pain points, emphasizing:
Interface clarity.
Feature parity between API and playground.
The need for transparency and live parameter control.
SOLUTION
From Insights to Opportunities
Although the initial goal was to document and prioritize UX issues, the research phase naturally surfaced structural improvement opportunities.
Rather than stopping at the diagnostic stage, I decided to synthesize potential design directions to illustrate how the developer experience could evolve.
This led to the creation of three complementary artefacts: a new information architecture, sketched interface concepts, and mid-fidelity UI explorations.
01. Information Architecture Proposal
I restructured the overall hierarchy around three user intentions:
Explore: Understand what Mistral offers (Home, Docs, Models).
Experiment: Play, test, and compare models (Aire de Jeux, Le Chat, L’Assistant).
Deploy: Integrate and scale (API, Cloud, Self-Deployment, Settings).
This framework re-established Le Chat and La Plateforme as connected steps of one coherent journey, rather than separate tools.
02. Sketching Phase
I explored how these insights could translate into interface mechanics:
Unified global navigation linking chat, docs, and APIs.
“Model Settings” panel exposing parameters (temperature, top-p, token count) live.
Folder-based chat organization to better manage multiple projects.
02. User Interface Design Proposals
The final UI mockups illustrated:
A unified La Plateforme entry point connecting Docs and Aire de Jeux.
A reorganized Le Chat Noir with side navigation, live model settings, and token counter.
Harmonized documentation hierarchy for a consistent look and feel.
The result envisioned a transparent, cohesive, and scalable experience aligned with Mistral’s B2B positioning.
OUTCOMES
Demonstrating the value of open-source and time-boxed UX research
This self-initiated project, completed in under three days, produced a full set of insights, diagrams, sketches, and mid-fidelity prototypes based entirely on open data and heuristic analysis.
It revealed how much can be achieved through structured observation and synthesis, even without traditional research resources such as interviews or analytics tools.
The process identified concrete UX inconsistencies across Le Chat, La Plateforme, and Docs, clarified how navigation could be unified, and proposed a clearer information architecture for the entire Mistral ecosystem.
Beyond the deliverables, it demonstrated that quick, low-cost design investigations can generate actionable insight and communicate strategic directions effectively.
Today, as Mistral continues to evolve and mature rapidly, these explorations remain a snapshot of an earlier stage, a moment when the company’s tools were still emerging and opportunities for design clarity were just beginning to surface.
