How to price AI in SaaS

#2-pricing-questions
1

IMPORTED FROM SKOOL
Original Post was posted by Ulrik Lehrskov-Schmidt on Oct '24

I’m writing a guest chapter on AI pricing for another authors upcoming book. So I’m thinking about this a lot lately, and wanted to give you a sneak-preview of a linkedin post that will come out later next week.

I’ve done maybe 10 pricing projects involving significant AI functionality this year. Here is how I think about it.

AI should be considered a 2-layer stack:

:one: The AI compute ‘fuel’ (i.e. token pricing at OpenAI)

:two: The AI solution (i.e. the value you add on top of AI)

2

Sally Forsyth • Oct '24

Thanks Ulrik we are about to run into the scenario next year but I am doing a price refresh now so I have half an eye on how we will manage the AI pricing. One of my biggest concerns is positioning and how to communicate AI pricing. I actually like the use of ‘fuel’ as a word to use. It would be great to see a couple of good examples. If you have found any.

3

Michał Narkiewicz • Oct '24

Fully agreed with enabling AI usage early on to gain data.

To price for the future the AI has to be an integral part of the solution. It seems that many AI functionalities are slap-ons equivalent of copying and pasting a chatGPT response. To be able to charge for the future, the AI has to do something that would otherwise be too difficult or even impossible to do.

We had two start-up cases with LLMs at the core of the value proposition:

  1. E-commerce solution - the AI assistant had a major impact on k…
4

Fynn Glover • Oct '24

This is a super interesting topic.

I interviewed the founder/ceo of an AI company yesterday who had an interesting take on this subject.

Summarizing in case useful:

  1. he doesn’t believe that value-based pricing will work for the AI companies as many people have argued
  2. he thinks most AI companies — assuming they’re application layer — should price based off seats and maybe have a little usage on top of the seat

Why does he believe this? I need to go back and listen and better synthesize, but the gist is:

  1. c…
5

Roee Hartuv • Oct '24

I think this is great.

For the Li post, you might want to refer to SFDC’s announcement two weeks ago regarding their AI pricing. This will most likely boost engagement.

This is just one of the articles talking about this subject.

6

Christine Carragee • Oct '24

This is changing super rapidly. There are companies like Predibase who can make on prem implementation much more efficient and a lot of upheaval in the major players so Agentic solutions and system designs are being set up to be able to swap from across different LLMs.

The biggest factor I don’t see being considered among these new tools is RISK – there are a lot of unknown or poorly understood ways these tools can be missed and hacked.

There’s a great book by Richard Heimann and Clayton Pummil cal…

7

Nikhil Kotcharlakota • Oct '24

Interesting topic

@Ulrik Lehrskov-Schmidt

. I have had a chance to work with a legal tech company bringing in AI as one of the main features.

We approached it similarly as well.

  • Started with Usage with two tiers
  • The next step was to make it more aligned with the additional usage information we get as more customers use it.

The only challenge I see though in moving from Fuel → Value is the willingness for SaaS companies to iterate on the pricing. Hope it will get better than what it was in the past and c…

8

Rob Litterst • Oct '24

Love the fuel and solution framing. Super helpful to hear how you’re thinking here, Ulrik. :pray: