AI and automation · Pharmaceuticals · Multiple markets
A dedicated AI model by Coblit pulls tenders from an aggregator, scores fit and fills in the bid and documents. At a pharma company: from over 10 people to 3.
About the project
We designed an AI system that pulls every tender from the aggregator, scores how well it fits the company's offering using a dedicated model trained on the company's own data, recommends which ones are worth entering, and then fills in the bid and the required documents.
Project journey
An international pharmaceutical company was analysing hundreds of tenders across many markets. Notices from an aggregator had to be reviewed and assessed by hand and — for selected procedures — the bid and documents had to be filled in manually. More than 10 people were doing this work.
A huge amount of operational work and difficulty making bid decisions.
We designed an AI system that pulls every tender from the aggregator, scores how well it fits the company's offering using a dedicated model trained on the company's own data, recommends which ones are worth entering, and then fills in the bid and the required documents.
Work that used to require more than 10 people is now done by 3 — and the team only works on the tenders with the best chances.
Key value
This is not a support tool — it is a system that replaces part of the analytical work.
The team doesn't analyse everything — it works on the best opportunities.
The system reflects the company's real business decisions.
The team doesn't start from a blank form but from a ready bid to check.
Summary
The project shows how a dedicated AI model can take over most of the tender work in a large organisation — from selecting notices to preparing the bid — and cut the team needed for these tasks from more than 10 people to 3.
Who it’s for
We see similar challenges in many companies. If you recognise yourself in this list, this project shows what we can do for you too.
Large construction and infrastructure companies
Manufacturers and suppliers of medical equipment
IT companies and integrators bidding in public tenders
Service providers to government and local authorities
Corporate groups operating in many markets
Bid departments overloaded with tender documentation (SWZ, the Polish terms of reference)Why it works
Fit scoring shows which tenders are worth entering, so nobody wastes time on procedures that don't fit.
The assessment is based on the history of its tenders, bids and decisions, so the recommendations reflect its real strategy, not generic data.
The system fills in the bid, declarations and attachments, and the team only reviews and signs them.
Tasks that required more than 10 people are now done by 3, and 100% of tenders from the aggregator are analysed automatically.
The process handles hundreds of tenders, can be extended to further markets, and over time the model selects procedures more accurately.
How we ran the project
This is how we worked together on the International pharmaceutical company project: from understanding the processes, through wireframes and architecture, to rollout and further development.
We talk to the people who will use the system. We map out how processes work today, where time and money are being lost, and how they should work a year from now, once the company has grown.
We build what solves a real problem — not a list of features nobody will use.
We design screens and user journeys in Figma and test a clickable prototype with future users before the first line of code is written.
You see and test the solution at the outset — changes to a mock-up cost a fraction of changes to a finished system. You can order the analysis and mock-ups as a separate stage.
We choose technologies to match the goal, the budget and the systems already running in the company. We design the architecture, integrations and data security so that the system can grow together with the business.
No costly rework a year later — the system is ready for more users, more data and new features.
We work in short stages and show a working version every few weeks. We launch the features that deliver the biggest savings first.
You have constant visibility of progress, a say in the direction and see the first results sooner.
We test the system, migrate the data, launch it in the production environment and train the team in day-to-day use.
A smooth start without downtime — from day one the team knows how to work in the new system.
After launch we monitor operation, respond to tickets under the SLA and extend the system with further features and automations.
One partner from idea to ongoing development — you are never left alone with the system.
How we work with every client Technologies we build with
Ready for your own success story?
Facing a challenge similar to International pharmaceutical company? We’ll show you how quickly you can go from idea to a working solution.
Book a short callCompleted projects
Coblit built a scraper and web app for a law firm that automatically collect content from forums and websites into one searchable database.
See the projectCoblit's AI system runs recruitment without HR: it talks to candidates on WhatsApp, reads CVs and documents, and closes the contract with a SignSafe signature.
See the project
