CASE STUDY · PHARMACEUTICALS · AHMEDABAD
Practical AI training for about 50 pharma staff, in one day
I trained about 50 people at Sundyota Numandis Pharmaceuticals across five functions on 2 October 2026, with one batch at the Ahmedabad office and one at the plant. They left using ChatGPT, Gemini, Claude and NotebookLM for MIS reports, documents, research and analysis, with a clear rule for what never goes into a public AI tool.
By Keyur Patel, AI adoption advisor and corporate Claude trainerPublished 4 min read
At a glance
- Industry
- Pharmaceuticals, Ahmedabad
- My role
- Trainer and programme designer
- Format
- Two sessions in one day, office and plant
- When
- 2 October 2026
- Teams
- Manufacturing and Maintenance, R&D, Medical Affairs, Sales and Marketing, HR, Training and Admin
~50
People trained
5
Functions covered
5
AI tools taught
On this page
What problem were they solving?
Sundyota wanted AI in everyday work across very different teams. Manufacturing writes SOPs and batch documentation, and maintenance tracks plant activity in MIS reports. R&D reads literature, Medical Affairs drafts manuscripts, Sales builds campaign briefs and HR answers policy questions. One generic demo serves none of them.
Pharma adds a constraint most companies do not have. Some documents are confidential and some are GxP-regulated, so “just paste it into ChatGPT” is the wrong answer. The team needed to know which tool to use, and which data must never go into one, before anyone could use AI with confidence.
What did the training cover?
Responsible use first, not last
Every participant got a one-page Pharma Responsible AI Canvas: data classes (public, internal, confidential, GxP-regulated), where a human must check the output, IP considerations and hallucination risk. Every exercise referred back to it.
Live tools on familiar work
I demonstrated ChatGPT, Gemini, Claude and NotebookLM on the material the team handles: MIS reports, formal emails, document drafts, research summaries and data analysis. People worked on their own devices.
Claude Cowork for sensitive files
For SOPs, batch records and other documents that should not be uploaded to a chat window, I showed Claude Cowork, which works on files on the user’s own machine.
Two prompt frameworks, not twenty
APE and RACE, practised until people could write a structured prompt without a template. Both come from the AiPromptsX library, which participants keep access to.
Projects, documentation and troubleshooting
How to set up a project so the tool keeps context across a piece of work, how to document what worked, and how to recover when the output goes wrong. Poonam named these as the parts that made it practical.
What changed?
- About 50 people trained across Manufacturing and Maintenance, R&D, Medical Affairs, Sales and Marketing, and HR.
- Every participant left with the one-page Pharma Responsible AI Canvas for classifying data before it goes near a tool.
- Five tools matched to tasks: ChatGPT, Gemini, Claude and NotebookLM for everyday work, Claude Cowork for sensitive files.
- Two prompt frameworks, APE and RACE, practised until people could write a structured prompt without a template.
- Poonam Miranda, who sponsored the programme, named MIS reports, formal documents and emails, research, summaries and data analysis as the work it applies to.
| Item | Use case demonstrated |
|---|---|
| Manufacturing and Maintenance | SOP drafting |
| R&D | Literature review in NotebookLM |
| Medical Affairs | Manuscript section drafting |
| Sales and Marketing | Campaign briefs |
| HR, Training and Admin | Policy Q&A |
“The practical demonstrations on prompt writing, project creation, documentation, and troubleshooting made the session relevant and easy to understand.”
Read the full feedbackHide the full feedback
Thank you for conducting such an informative and practical AI session. Learning how to use ChatGPT, Gemini, Claude, and NotebookLM for creating MIS reports, drafting documents and formal emails, conducting research, summarizing information, and improving data analysis was highly valuable. The practical demonstrations on prompt writing, project creation, documentation, and troubleshooting made the session relevant and easy to understand. The session will help us use AI more effectively to improve work efficiency and maintenance activities. Overall, it was an engaging and useful session.
What happens next?
This was a first engagement in a regulated industry, and it was built to be one. The responsible-use canvas, the function-specific examples and the two-framework limit were all choices about what a pharma team can adopt the next working day.
The programme was designed with a second phase in mind: paired-function cohorts, a playbook of each team’s top three workflows, and an AI adoption policy. Sundyota plans to discuss that phase next quarter.
Common questions
Can you run AI training for a pharmaceutical company?
Yes. The programme for Sundyota Numandis was built for pharma: a responsible-use canvas covering confidential and GxP-regulated data, Claude Cowork for sensitive files, and use cases from manufacturing, R&D, medical affairs, sales and HR.
How do you handle confidential and GxP-regulated data?
Before any tool is opened. Every participant classifies data as public, internal, confidential or GxP-regulated using a one-page canvas, and the exercises refer back to it. Sensitive files go through Claude Cowork on the user’s own machine, not into a chat window.
Which AI tools do you teach non-technical teams?
Usually ChatGPT, Gemini, Claude and NotebookLM, matched to the task: drafting, long documents, research and working across a fixed set of sources. I teach which tool fits which job, and which data never goes into any of them.
Can you train at our plant as well as our office?
Yes. For Sundyota I ran one batch at their Ahmedabad office and one at their plant on the same day, so plant and maintenance staff did not have to travel. I run sessions on-site across Ahmedabad and GIFT City, and remotely across India.
Programmes behind this work: AI Training for Manufacturing Companies, Corporate AI Training for Employees, AI Governance Workshop. Training in this city: AI training in Ahmedabad.