For most large companies the obvious answer is to "build it". But for enterprise adoption and scaling, "buy it" is the superior choice that will play out over the next 18 months.
I have a strong view on this one, so this isn’t going to be the politicians answer of “it depends!!” - I am happy to explain why.
It seems to me that everyone is building it today…
I spoke to one industry expert who is very well connected in procuretech and he is convinced that 5 out of the top 50 US companies are now building their own bespoke procurement AI agents. Anecdotally I have heard the same. “Buy it” is rejected by IT departments because of the risks, AI control, data security, hallucinations, interoperability, etc. etc.
You can accuse me of being defensive given I obviously have a vested interest in my clients choosing to “buy it” not “build it”. TBH, one of the hardest questions I have to answer about Nibble is, “why can’t I just use Chat GPT for negotiations?” And honestly, the tech is really really good now. It can deliver a very convincing negotiation as a co-pilot.
Just prompt it to rely heavily on well researched academic theory for negotiation tactics, coach it in a tone of voice which suits your corporate culture and it can now do “deep research” for arguments and justifications for your desired pricing (e.g. looking at market commodity prices and searching the internet for similar contract benchmarks). Within a few hours anyone (even me and I am not a techie) could code up a negotiation bot which would come across very convincing. But it would not be reliable enough for scale use cases and this is the crux of my argument.
At Nibble, we have done nearly 2 million negotiations today and we have been subject to many fun attempts to prompt engineer Nibble and some more serious hacking attacks – one memorable time when a Google engineer prompted the whole of X to try and break us over a weekend. I am proud to say Nibble didn’t succumb!
Until you have enterprise quality reliability you cannot move away from co-pilot to truly autonomous and until you have an autonomous agent working on your behalf your ROI on the project will be capped. Your ability to scale will be limited by your team capacity.
There are plenty of counter-arguments but I truly believe we will end up with a handful of highly specialist agents you buy in for your corporate needs (of which AI negotiation is one) and those will deliver significant ROI. Alongside that we will have thousands of personal efficiency agents you might build yourself, helping individuals plan their day, meet their KPIs, answer emails and deliver more with less… BUT these alone are not going to deliver financial pay-back. They are just going to be a new way of working.
If you are sceptical, have a look at this UK government report which sought to assess the impact of AI co-pilots on one government department in a controlled trial over 3 months from September 2024 to December 2024. The participants were positive and generally liked using the tool and felt they saved 26 minutes per day from mundane tasks like taking meeting notes and summarising documents but there was no evidence that this actually increased productivity and – based on my reading about the activities they did – we are a long way off AGI. It was struggling to be effective for these users on powerpoint and excel tasks.
Find out more from Nibble's experience negotiating 100,000 times a month here.
Interested in Nibble?
