Skip to main content

We've been talking for a while about AI and possible impacts on procurement. We use AI in our own service to almost effortlessly categorise hundreds of millions of documents in dozens of languages. We can complete a task in around 6 hours that would take a team of six humans more than four years to complete. That feels impressive but we're not doing using this technology to its full potential (yet), for a glimpse of that take a look at a new service that takes question prompts from tenders and outputs a cogent answer to the question posed.

Want to add a few facts or reference an existing piece of work? Fine, just amend a couple of prompts and Autogen will reprise its answer and include a pertinent fact or insert one of your case studies which it learned about during the set up phase. The algorithm doesn't go over the word limit, doesn't make typos or book holidays that coincide with tender deadlines.

But that's not what's most interesting about, the most interesting part is what it will do to the procurement market. Tools like Autogen could significantly lower the cost of bidding, teams that are used to maintaining a staff of twenty writers are going to need half that number and they'll probably be able to submit more bids. Removing unproductive activity from the economy feels like a good thing, but for governments who are required to run open tender processes this could have a marked impact. Governments can expect many more bids and for those bids to be of much higher quality. So more work and less ability to differentiate between submissions.

In a world where tenders are cheaper to submit, it is easy to imagine that bidders will do what they can to submit a tender for as many frameworks as possible, simply on the basis of playing a numbers game. There's also the question of resource equality, smaller businesses can't yet afford this sort of enterprise style service, simply because it takes a lot of very skilled staff to set up a model for a specific company. So the first mover advantage will inevitably go to the larger companies with large bid teams, for whom the economics of automated bidding are irresistible.

These imminent changes poses substantial questions for governments who need to start exploring how AI will impact procurement and what governments can do to adapt. If you’re thinking “we’ll just use AI to evaluate bids” then you may well be missing the point. If you're a government who'd like to know more about the opportunities and challenges from AI in procurement, get in touch.

Image credit: image created by OpenAi’s DALL-E 2 algorithm, using the prompt: “A black and white photograph of a futuristic train terminus with a train approaching the viewer at speed”

Post by Fiona Hunt
November 8, 2022