Axonal AI / moving from a chat surface to a real workspace.

Axonal helps pharma marketing teams turn brand briefs into compliant creative work. The platform's first version was an AI chat. We redesigned it as a workspace built around the actual cadence of the team's work.

Client
Axonal AI
Sector
Pharma marketing
Role
UX Strategy, User Research, Product Design
Year
2024 to 2026

Teams were collaborating with AI in separate tabs, losing alignment the moment two people touched the same brief.

Pharma marketing teams were jumping between ChatGPT, Perplexity, and shared docs to draft regulated creative work. Each tool produced its own version. Each handoff produced version drift. The platform that was meant to speed work up was reliably slowing it down.

Axonal needed an AI workspace, not another chat surface. Multiple contributors had to be able to work on the same brief, in the same context, with a stable record of what changed and why. Compliance-bound work cannot tolerate quiet edits and untraceable suggestions.

We redesigned the workflow model around shared workspaces with consistent history, structured updates, and a clear separation between exploratory thinking and official artifacts. The AI participates inside the workspace, not from outside.

Internal pilot results.

  • +45% faster workflow, teams complete research and content tasks materially faster.
  • +60% user satisfaction, measured in the internal pilot before broader rollout.
  • +35% workspace engagement, teams collaborate inside shared spaces more frequently than they did in scattered docs.
  • +50% output reuse, reports, summaries, and insights are reused cross-team instead of recreated.

One context, multiple contributors, one source of truth.

The workspace is the unit, not the conversation. When a team creates a workspace for a New Product Launch, every output, brief, summary, competitive analysis, content draft, lives in that one shared context. Edits remain traceable. Sources stay linked. Teams refine work together.

The artifact updates live when a user asks the model to revise it, the way you would talk to a colleague: "Add a competitive analysis chart at the end." The change is logged. The previous state is preserved. Nobody has to manually reconcile two versions.

Exploratory thinking, separated from regulated work.

Teams kept mixing draft thinking with final, sometimes regulated work. A speculative question in the wrong document became a compliance liability. We designed a separate Quick Insights surface for the half-formed questions and quick checks that are part of any creative process, kept out of the official content path.

The result: people stop being careful about where they ask questions, and start being careful only about what they ship. The error rate on regulated content drops because the regulated content surface is no longer doubling as a scratchpad.

An AI workflow engine, not a chatbot.

Marketing work does not start the same way every time. We added multiple entry points so the team can begin from where their thinking actually is:

  • Continue a recent workspace.
  • Create a new one from scratch.
  • Use a previous artifact as a template.
  • Ask a quick question without generating a full artifact.

This is the move from a tool that answers questions to a tool that runs workflows.

The principle

For regulated team workflows, the AI cannot sit outside the system, it has to participate inside it. Once the AI's context, history, and attribution all live in the same workspace as the team's work, the conversation about trust stops being abstract.

Other case studies

Read what's already published.