Case Study - AI Adoption - Medium B2B Service Company

A company with over 100 employees, with nearly 30 people working in operations for pharmaceutical companies. Within their business lines, they offer several main services:

• Employee recruitment service in the pharmaceutical industry,
• Specialized service: registration of drugs and medical products,
• Sales team outsourcing, 
• Training in the field of Pharmacovigilance, Professional Sales Techniques, and Management of Adverse Events.

To clarify the details and make the development areas more practical, we conducted an AI Workshop in collaboration with the Management Board and Directors, where we analyzed the current issues. Ask ChatGPT

As part of these activities, a number of time-consuming actions emerged within the company:​

• Daily monitoring of new active substances in medications and adverse effects (manual work, internet searching),
• Preparation and verification of documentation according to the standards of the law on drug advertising (Summary of Product Characteristics / Product Information),
• Time-consuming translation of documents related to medications (and their registration),
• Inquiries and citations of specialized knowledge / selected internet portals,
• Construction and reconsolidation of procedures for professional services: drug registration and Pharmacovigilance,
• Analysis and browsing of databases on websites related to medications, e.g., Register of Medicinal Products.

The tasks generated a lot of administrative work during the service delivery, and the client wanted to streamline and speed them up. After the workshop, we analyzed them, assessed their time consumption, and provided suggestions. At the end of this stage, the company received an AI Report, in which:

• We narrowed down the current issues to the 3 most important ones,
• We presented solutions for solving individual problems,
• We presented the business justifications for each of these solutions (investment, maintenance cost, time savings, and ROI).

See the details of the AI Report and the proposed solutions below. 

Each of the identified problems received two recommended potential solutions. See the summary and assumptions below.

• sav. - estimated time savings
• Hours - hours in the week dedicated to the task

For manual document translation, we proposed two options:

• Utilization of the existing AI product - DeepL (200 documents per month cost 220 PLN) and partial support for preliminary translation. Monthly savings of about 15,000 PLN (75 PLN / h), 
• Document translation platform - definition of templates, translation directions, and the use of specialized knowledge in translations (approximately 50,000 PLN investment cost + maintenance cost of about 100 PLN). Monthly savings of about 24,000 PLN (75 PLN / h),  
• The client decided to go with the AI Product - DeepL based on these estimates. 

For monitoring medical articles, I proposed:

• Implementation of AI through the development of automation that will monitor news based on the existing portfolio of drugs and selected portals, and inform employees via email daily. The budget for the investment was 5,000 PLN and the monthly maintenance cost.
• Utilization of the existing product - Perplexity, which provides daily searches for free. The paid version is 20 USD / user / month.
• The client wanted to minimize the risk of their mistakes (forgetting, overlooking) in some way and to standardize the reporting format for new adverse events and changes in active substances, so they decided to implement AI.

The last problem - time-consuming verification of leaflets. Due to the limitations regarding the effectiveness and memory of Large Language Models and the investment cost in AI Automation (> 50,000 PLN), we recommended to the client to build a tool that supports this process and provides insight into what to pay attention to. The client agreed, and we planned our own GPT in ChatGPT.

Based on these proposals, we decided with the client to take a dual approach.:

1. First, we will go through AI trainings for companies and we will train the specified team on the products and support the other ideas. The client has decided on the products: ChatGPT, DeepL, Perplexity, NotebookLM, and Manus AI. 
2. After the training, we will move on to AI implementation and we will start with monitoring adverse effects in drugs / active substances.

The training was prepared based on an interview with the management and an assessment of potential solutions. As part of the training, employees learned about 3 AI products, completed 7 tasks, and became familiar with the basics of data security and AI risks. The training took place at the client's office, with nearly 20 participants. Before the training, we assessed the time consumption of selected tasks based on a "self-assessment" survey and reported time. 

As part of the training, we covered the following areas: 

• 🤖 Introduction to AI and discussion of main risks, assessment of AI usage by participants, Regulations regarding the use of AI in daily work
• ⚖️ Introduction to ChatGPT
• 📄 Task 1: Generating ready-made content
• 📝 Task 2: Internet searching and searching specialized documents (Perplexity, ChatGPT)
• 📝 Task 3: Building your own GPT and your own projects (ChatGPT)
• 📚 Task 4: Translation of Word and PDF documents
• 📧 Task 5: Searching through: medications databases using Manus AI (together, as the solution is not available to everyone)
• 🌐Task 6: Analysis of websites and PDF files, creating a mind map with NotebookLM
• 🗣Task 7: Generating training materials and presentations with AI (ChatGPT, Canva AI)
• Tips and tricks for everyday work (at the end, a presentation of interesting features such as voice cloning or video cloning) Ask ChatGPT

Process audit

• Based on the workshops, we assessed the problem and defined the requirements (in this case, an analysis of three active substances from Pubmed: ibuprofen, paracetamol, pseudoephedrine)
• We have defined the platform to be used (here we chose N8N in the cloud version), 
• We have planned the budget and the plan for the next 2 weeks.

Check budget and ROI

Pilot

• We have prepared automation for the client,
• We conveyed how it will be used on a daily basis,
• We left the client for a week for testing. 

Rollout

• We trained the client so they could continue to develop this process,
• We have prepared a short video documentation,
• The client received support in the form of 1 hour of free consultation within a month,
• After a month, we assessed the time consumption (a decrease from 7.5 hours per week to 1.5 hours)

Check demo of this automation

After successfully completing the first AI implementation, we moved on to the next ideas from the AI Workshop list and have been collaborating in new areas ever since. 

After 3 months of our collaboration, we conducted an AI productivity survey and reviewed the results. Over 80% of employees participating in the training use AI Products daily in their work, increasing their productivity by an average of 25%.

When it comes to monitoring articles about medications, the client has managed to reduce the time spent from nearly 1.5 hours per day to around 10 minutes. Employees now devote this time to other services or focus on actively developing the business.

Added value from the collaboration:

• The client is able to independently automate simple processes using N8N,
• They have become familiar with new digital tools - such as Canva AI, NotebookLM, and DeepL - which they now use in other tasks as well, not just those related to training,
• Employees have become more comfortable working with AI and have started to see it as a tool for increasing productivity rather than something that will replace them,
• We have planned additional, practical AI training sessions—for consultants as well as for sales representatives.
• The company has a long-term plan for building AI competencies, which it is systematically implementing internally or in collaboration with AutomAIze.