Want To Know How To Calculate Your Enterprise AI Usage? You Can't!

Let’s take a look at some AI pricing models and pause to pay respect to the poor sod whose job it will be to make sense of it all.

Let’s take a look at some AI pricing models and pause to pay respect to the poor sod whose job it will be to make sense of it all when the CTO starts writing blank cheques.

Zapier: Per task automated
Microsoft: $4 per hour of usage (AI copilot for security)
OpenAI: Per input/output token (GPT-4o)
Warmly: Hybrid pricing (features & No of warm leads)
Asana: Based on platform credits (AI Studio add-on)
Cognition: Per agent compute unit (Devin)
Relay.app: Per workflow step
Salesforce: $2 per conversation (Agentforce)
Zendesk: Per successful autonomous resolution (Zendesk AI)

(there’s an even bigger list in my last post about AI Orchestration)

I mean, come on, this is taking the piss, it’s made up nonsense and completely opaque pricing in a lot of cases, much like the black boxes they’re providing.

How any CFO, procurement bot, or CTO is expected to be able to budget is beyond me. Pricing models like “pay-as-you-go,” “per output,” or “per credit” can result in fluctuating monthly bills making it difficult to forecast expenses for starters let alone account for any seasonal or project based surges in usage.

And then you have the difficulty during a tender process of accurately modelling any TCO or cost savings you’re being promised by ejecting a human from their seat for a bot, or making a true comparison between competing vendors in the same space with different pricing models. Nothing is standardised, there’s no available metrics to prove whether anything is worth the implementation, and there’s no real methods of tracking ROI when you’re essentially having to work with multiple vendors here with wildly different cost models.

There’s very real chances here that an enterprise will risks paying multiple vendors for similar AI capabilities if the pricing models obscure true functionality overlap as well.

What are you going to do, use ChatGPT to work it out for you? I mean, sure, it’s supposed to be PhD level and all that malarkey after all. With such varied pricing models, ensuring compliance with financial policies and conducting accurate audits can be time-consuming and those structures may be subject to different taxation rules depending on geography as well.

It’s honestly a scam, nobody knows how to price this shit, they’re making it all up. The vendors have no idea what it’s going to cost them to run the bloody stuff either in the long run because they can’t bank a profit.

It means you either end up with one fucker of a headache to report on, you build or buy yet another procurement solution to help you wrangle the costs, or you look at even costlier AI orchestration middleware to handle it all for you.

The harsh reality is you will not be using just one AI platform, but many, and by the time you get a handle on the true cost of running a business with AI agents bouncing around then I can pretty much guarantee you will see those savings evaporate as a result of the manual effort needed to reconcile and audit it all.

Oh, the irony.

(and if you’re looking for an idea for a low-cost app that will generate a lot more value than an AI platform then it’s sitting right above)