Ajay Kumar.
Case study

What I built for Project Services Global

Project Services Global is a Calgary company that bids for contracts, public and private. Two parts of their week were being done entirely by hand.

Someone was searching the government tender portal every day to find the contracts worth bidding on. And when a long document came in, somebody read it end to end to answer one question about it. Both jobs were necessary, both were slow, and both happened before the real work of preparing a bid could start.

I built one automation for each. Neither one makes a decision. They take the searching and the reading off a person so the day starts further ahead.

The daily tender search

Built for Project Services Global

I set up a search that runs on their keywords against the government portal every morning and writes the matches into one sheet by 6am. Instead of someone opening the portal and working through it, the shortlist is there when the day starts.

It does not decide what to bid on. It puts the candidates in one place so that call can be made over coffee instead of after working through the portal by hand.

Every matching tender in one sheet by 6am.

The daily tender search, running on a schedule and delivering the matches to one sheet at 6am. Built for Project Services Global. Watch on YouTube.

The document advisor

Built for Project Services Global

The second build loads a long document once and then answers plain-English questions about it, pointing back to the section each answer came from. A question that used to mean reading the whole thing now takes a few seconds.

There is a live version you can try, loaded with a sample document. It is the same build.

Answers from a long document in seconds, not an afternoon.

A long document loaded once, then answered in plain English, with each answer pointing back to its source. Built for Project Services Global. Watch on YouTube.

Live demo: pdf-chatbot-ajay.netlify.app

6+ hours a week, back

Between the two builds, more than six hours a week that used to go on searching and reading now goes on the bid itself.

The productised versions are on the Diligentic site: the tender assistant and the knowledge base assistant.

All of the work