Deployment strategists sit between the software and the people who'll use it. In practice, that means figuring out what the customer is really trying to decide, finding the data that bears on it, building the workflow, and then watching people use it and fixing what doesn't fit.
Insurance: reading a hundred million documents
The first customer was an insurer. I built extraction and classification agents for a pipeline covering more than 100 million documents. Together with the rest of the team's work, that surfaced more than $50 million in addressable premium.
I also helped ship the platform the customer's team now works in. More than 60 insurance professionals use it every week, in place of the spreadsheets, email threads and phone calls they had before.
Medical supply: forecasting what to buy
For the second customer I embedded with their team and built procurement forecasting agents. Their analysis is informing roughly $20 million in potential savings.
What the work taught me
- The first version of the question is usually wrong. Most of the early work was finding out what decision a team was actually trying to make.
- At a hundred million documents, data quality is the project.Classification is only as good as your handling of the scans, duplicates and formats nobody planned for.
- People have to be able to check the machine. A tool is only adopted when the person using it can see why it said what it said, and correct it.