MUXA Design Studio
MUXA Design Studio
We redesigned how pharma teams collaborate by creating a shared AI workspace that keeps complex projects aligned, contextual, and enterprise-ready.
Teams were using tools like ChatGPT and Perplexity separately, which caused version drift and constant rework.
Version drift is what happens when several people each keep their own slightly different copy of the same work and no one can tell which version is current. In pharmaceutical marketing that is more than an annoyance. Every public claim about a drug has to pass medical, legal, and regulatory review, often shortened to MLR, and when the research and drafts behind a claim live in scattered chat tools, the trail of what was actually checked and approved gets lost.
We redesigned Axonal's workflow model so multiple contributors could work in one stable workspace with consistent history, structured updates, and no conflicting edits. This created a faster, more reliable foundation for AI-assisted collaboration, rather than a pile of disconnected chat windows.
Axonal's teams were working in scattered documents and losing alignment as soon as multiple people touched the same project. The moment two contributors edited their own copies, the team no longer had a single answer to a simple question: what is the current version, and who changed what?
We redesigned the workspace experience so every team member works in one shared, reliable place. The system keeps updates consistent, avoids duplicate versions, and maintains context even when people join at different stages, so a colleague who arrives halfway through a project can see how it got to where it is rather than starting cold.
We noticed teams often mix draft thinking with final work, asking a quick exploratory question in the same place they keep approved content. In regulated marketing that is risky, because an unverified, off-the-cuff answer can quietly slip into an asset that is supposed to have passed review.
To fix this, we designed a separate view for quick questions and checks, a safe space to explore ideas without affecting official content. This reduced errors and kept regulated work clean, giving people room to think out loud without contaminating the record of what was formally approved.
A blank chat box is intimidating, and it is rarely how real work actually begins. People usually pick up where they left off or build on something that already exists.
To make Axonal flexible for different marketing needs, we introduced multiple starting options: continuing a recent workspace, creating a new one, using a previous artifact as a template, or asking a simple question without generating an artifact at all. An artifact here just means a saved output, a report, a summary, a compliance check. This approach transforms Axonal from a basic chatbot into a complete workflow engine.
We rebuilt the main conversation page, the screen where users actually guide the system: choosing data sources, picking a model, and directing what the agent should do next.
Our focus was clarity, showing what changed, why it changed, and how the project is evolving. This reduces confusion and helps teams trust the system during long projects, where work passes through many hands over weeks and a person needs to understand a decision they were not in the room for.
Before any polished screen existed, we worked the whole thing out in low-fidelity wireframes. These early sketches helped us define navigation, shared context, and how long projects stay organized before moving into final UI. Getting the structure right on paper is far cheaper than discovering it is wrong in production.
We also explored how users give instructions, how the system updates work, and how multiple contributors stay in sync. These foundations shaped the final interaction model, including how a task moves from a prompt, through the agent's working steps, to a reviewable checklist.
More selected work across enterprise AI, consumer fintech, and complex internal systems.