The problem
Launch knowledge is spread out: guidelines, processes, documents, market research, and the numbers behind the dashboards.
case study · Enlight · 2023 — now
The LEA assistant: answers for launch teams
I designed and built the AI assistant in LEA, Enlight's launch management platform for pharma and medical companies.
YouWhy is Germany over budget on Product A this quarter?
LEA assistantTwo Q4 purchase orders landed in Q3, and one activity was added outside the approved plan.[1] Budget tracker[2] Approval guideline[3] SAP actuals · Q3
LEA, the platform it lives in · lea.solutions
900+
launches managed
15
launch stages
One
data model behind every dashboard
in four notes
The story
What I built
An assistant inside LEA. Ask about a process, a document or the numbers, and it answers with links to its sources. Agents help collect the data.
Where it lives
On LEA's front page, behind a help icon, in chat, and under AI Insights.
What changed
On one multi-country launch, past market research was already in LEA. With the AI reading it, the team understood the market quickly.
from question to answer
How it works
You ask
About a process, a document or the numbers.
It reads the playbook
Your guidelines, launch processes, documents and market research.
It checks the numbers
In the data model behind the dashboards.
You get the answer
With links to the tracker, guideline or data behind it.
for pharma IT review
Built for trust
- Runs in Azure
- The AI features run in Microsoft Azure, and people sign in with their company account and MFA.
- Never trains on your data
- Customer data is never used to train models, and AI can be switched off per organisation.
- Audited
- Enlight is audited to SOC 1 and SOC 2: security, availability and financial reporting.
My part
I designed and built the LEA assistant and its agents, mostly on my own, with help from Enlight's BI team. I also led Enlight's Azure platform revamp, with SOC 2 built in.