An intelligence workspace for a global pharmaceutical company — unifying research, prescribing, referral and account data into a single trusted, AI-assisted experience.
A global pharmaceutical company was building a scalable AI platform to help teams access business and medical insights without navigating multiple disconnected systems.
It brought together qualitative research, quantitative data, field interactions, prescribing behaviour, referral activity and account-level intelligence in one enterprise experience. I designed an integrated platform combining natural-language AI conversation, HCP and account search, prescribing and patient insights, referral-network visualisation, pre-engagement planning, contextual filters and comparisons, evidence and feedback controls, and administrative tools.
The objective: move beyond static dashboards to a workspace where users explore data, ask follow-up questions, and understand the evidence behind every insight.
Medical and field teams worked with information distributed across several systems and formats. Preparing for an engagement meant manually searching, comparing and interpreting from many sources.
Prescribing data, patient activity, account relationships, field notes and referral patterns lived in different places — no unified view of an HCP or account.
Users needed to move fluidly between account-, HCP- and network-level information while keeping the context of their original question.
Referral networks, eligibility, prescribing and time-based comparisons carry huge amounts of information that must be surfaced without overwhelm.
In enterprise healthcare, an answer can't be an unsupported recommendation. Users need its source, scope, timeframe and limitations.
Field, medical and insight teams need different depth — quick pre-engagement prep and deeper analytical exploration alike.
…create a trusted AI-assisted workspace that helps healthcare teams move from a question to an evidence-backed insight without searching across multiple systems?
Existing-experience audit. The earlier experience was heavily form-based — users configured several fields before an answer appeared, with dense, text-heavy results and little support for exploration or follow-up.
Requirements & use-case mapping. Searching for an HCP or account, preparing for engagements, natural-language Q&A, prompt suggestions, conversation history, prescribing/patient summaries, referral exploration, comparisons, contextual filtering, response feedback, and administration — organised into one shared architecture, not independent dashboards.
Information architecture. Three primary layers:
Iteration & cross-functional review. The file carried multiple workflow states and separate annotations for design, development and business stakeholders — so behaviour, data needs and edge cases were reviewed before handoff.
Wireframes established the product architecture and validated the relationship between search, dashboards and AI conversation before applying the visual system.
Key flows recreated and fully anonymized for portfolio presentation — conversation, explainable responses, entity intelligence and the referral network.
Prescribing for this account increased steadily across the period, led by two high-volume specialists. Referral inflow from the regional network also rose.
Growth is concentrated in two indications; referral strength suggests expansion potential.
Every response can expose its sources, timeframe and a usage disclaimer — AI positioned as support, not absolute guidance.
Summary-first cards, collapsible sections, tabs, drawers and detail-on-demand tables keep complexity accessible.
Active filters stay visible so users always know which data is shaping a chart or answer.
The selected HCP, account, filters and timeframe carry across dashboards and conversation.
A component-led foundation — so the platform can extend to new data domains and AI agents without redesigning its patterns.
Described as product & usability outcomes — the file contains no verified production metrics, so no quantified results are claimed.