Case Study: Top-10 Pharma - Daily GenAI use in four months

This team works on early-stage pharma market research. When they get a call wrong, the company can either miss a future blockbuster or spend years and millions chasing a molecule that goes nowhere.
That's the kind of work we trained them to use GenAI for.

The Business Need

The Portfolio & Pipeline Strategy Group's Market Insights team does early-stage market research and sizes the commercial opportunity for new molecules. The cost of getting these calls wrong is significant in both directions.

Missing a real opportunity

If the team writes off a molecule that turns out to be the next billion-dollar drug, the company misses it.

Chasing a dead end

If the team backs a molecule with no commercial path, the company spends years and millions on something that won't pay off.

GenAI could clearly help with this kind of work, and the team already had access to the tools. What they didn't have was the confidence or the practical skill to use those tools on work this important.

Workshop facilitation with the Market Insights team

The Approach

We used a method called See–Do–Teach, which comes from how doctors are trained. The reasoning is simple: a smart person who has read about a procedure can still freeze when it's time to do it for real. They need to watch it done, do it themselves with guidance, and then explain it to someone else before it really becomes second nature.

01

See

We show the team real GenAI use cases, live, using their actual research questions. No toy examples.

02

Do

The analysts try GenAI on their own work in the room: molecule sizing, competitive intelligence, source triangulation. We coach them while they do it.

03

Teach

Team members teach what they just learned to their peers. This is how internal champions emerge and how adoption continues after we leave.

The point is that hands-on practice is what changes how people work. Slide decks about AI don't.

The Engagement

Phase 01 · Pre-engagement

Workflow Audit

We captured how the team spent its time and mapped that against their actual responsibilities: seven focus areas where AI could help, with a clear picture of how each one was being handled today.

Phase 02 · 29 July

First Demo Session

A live demo of four off-the-shelf GenAI tools, each matched to a specific pain point from the audit: a tool for search and citations, one for web-grounded research, a general-purpose LLM, and a model better suited to creative and narrative work.

Phase 03 · Aug / Sept

Internal Tools Showcase

We then walked through the company's licensed internal tools (AlphaSense, GiGi, Jules) for internal research, meeting summaries, and governance prep. These cover cases where off-the-shelf models aren't appropriate, usually for governance or data-access reasons.

Phase 04 · 15 October

In-Room Workshop

A full day in person: live demos of every tool, troubleshooting, best-practice patterns, and analysts teaching each other what was working for them. This was the "Teach" step running at full scale across the team.

The Result

The honest measure of an AI program is whether people are actually using the tools afterward. If next Monday looks the same as last Monday, the program didn't work. In this case, both daily use and overall use went up significantly in the four months after the workshop.

Adoption · Daily

Daily AI Use Among Market Analysts

16%

Before

83%

After · 4 mo

A 67-point increase. Most analysts now use GenAI as part of their normal working day rather than as a side experiment.

Source: PPSG team survey, four-month post-program review.

Adoption · Any Use

Any AI Use Among Market Analysts

31%

Before

100%

After · 4 mo

Every analyst now uses AI at least weekly. That was the baseline the program was designed to reach.

Source: PPSG team survey, four-month post-program review.

Where the Time Went

The audit captured how analysts were actually spending their time before we started, across seven focus areas. Four months in, the mix had shifted: meetings, deskwork, and admin all came down, and analysts spent more of their time on strategic and narrative work. Value-add activities went from 35% of total time to 48%, a 13-point shift.

Time Reallocation

Share of Team Time by Activity - Before vs. After

Before · pre-engagement

35%

Value-Add

After · 4 months in

48%

Value-Add

↑ +13 pts

Blue · Value-Add Activities

Strategic Thinking

Narrative & Storytelling

Analysis

Governance Prep

Grey · Low-Value Time (Compressible by GenAI)

Meetings

Deskwork

Admin & Other

Time-capture aggregated from team self-reports (PPSG, June 2024 baseline). "After" reflects post-engagement allocation. Categories were defined in the audit, and each segment is labeled with its share of total time.

The Audit · Detail

Seven Focus Areas - Where the Team Actually Spent Its Time

Focus Area	% of Time (Before)	What It Looked Like
Deskwork	25%	Manually searching internal and external sources; reading and distilling; manually validating and referencing.
Meetings	23%	Scheduling, follow-ups, agendas, pre-reads, minute-taking.
Admin & Other	20%	Email, travel, expenses, HR tooling, RFP creation, line management, internal initiatives.
Analysis	12%	Excel analysis of desk research; triangulation; forecast models from existing templates.
Governance Prep	10%	Objection handling, governance decks, stakeholder management and education.
Strategic Thinking	9%	Higher-order narrative and decision work, the part the team actually wanted to be doing more of.
Narrative & Storytelling	7–32%	Building narratives, decks, executive summaries. Range reflects current versus target spread.

Pre-engagement time capture, June 2024. Grey square = activity that off-the-shelf GenAI can compress; blue square = value-add activity to protect or grow.

Pain Points Addressed

The audit produced a short list of problems we could address this quarter using tools that already exist, rather than a roadmap for someone to build something later.

Deskwork

Sourcing & Distilling

Finding accurate, up-to-date sources quickly
Distilling many sources into clear summaries
Validating and referencing sources
Surfacing internal information and prior research

Content Analysis

Turning Research Into Insight

Bite-size, readable analysis with clear summaries
Triangulating insight across multiple sources
Sense-checking forecast inputs and assumptions

Storytelling

Narrative & Exec Summaries

Building narratives from multiple research sources
Consolidating into one- and two-page exec summaries
Drafting governance and pre-read narratives

The Tooling

We picked four off-the-shelf models with complementary strengths and matched each pain point to a primary tool plus a fallback. The company's licensed tools (AlphaSense, GiGi, Jules) layered in for the phases where governance or data access required it.

01 · Search & Cite

Highest-Quality Citations

Question-led prompting, accepts multimodal input, uses several LLMs on the backend. Best for source-grounded research where the citations need to hold up to scrutiny.

02 · Web-Grounded

Search-Backed LLM

Up-to-date internet data, includes source links, accepts large inputs, supports multimodal, exportable. Best for fast-moving competitive intelligence.

03 · General Purpose

Jack-Of-All-Trades LLM

Internet access, data analysis, multi-language. The default workhorse, and the model most analysts open first each morning.

04 · Creative & EQ

More Human-Sounding Model

Better at sentiment, stronger ethical reasoning, higher prompt-token limit. Best for narrative drafting and executive summaries.

Outcomes

→Daily GenAI use across the team. The workshop didn't end at the workshop; the behavior carried into normal work.
→A GenAI playbook the team owns, which they can use to onboard new joiners and refresh existing analysts without us being in the room.
→Competency across both internal and external tools. Analysts know which model to reach for, why, and what to fall back to if it's the wrong choice.
→Time reallocated from meetings, deskwork, and admin into strategic thinking, narrative work, and analysis.
→Internal champions. The "Teach" step produced peer trainers who now run the practice on their own.

"The playbook is fantastic and sets us miles ahead of the rest of the org."

Portfolio Strategy HeadUSA

If This Sounds Like Your Team

Your audit is one conversation away. Your team's first session is one quarter out.

Every engagement starts the same way: a workflow audit, a time-capture exercise, and a short list of pain points we can address with tools that already exist. Then we get into the room with your team.

Book a Discovery Call Inside the Lab