February 2020
The AI Software Business Model - Is it Different Than SaaS?

Welcome to The Cirrostratus Group February newsletter.

Previously, I've written about some of the changes required as you incorporate Artificial Intelligence (AI) into your SaaS product such as on-boarding and the importance of managing the associated services. The article below, written by investors in AI businesses, builds on that and talks about several differences in an AI software business model. I strongly recommend you read the article if you are using or considering using AI in your SaaS product. I think one of the biggest implications is the potential of lower gross margins. Most importantly the article has some suggestions of how to potentially address some of the business model differences and challenges.

Gartner's Magic Quadrant for Data Science and Machine Learning Platforms just came out. They include 16 vendors in the evaluation, all the way from the large cloud providers such as Google and Microsoft, to much smaller providers. The vendor offerings and their strengths and weaknesses are quite diverse and provide lots of alternatives. I think it is great that Gartner is focused on the whole solution life-cycle including model testing, deployment,   monitoring & maintenance. These items are critical to get right in the SaaS model because of their importance to customer satisfaction. You can get a free copy of the Magic Quadrant here from Dataiku, one of the vendors in the leader quadrant.

I'm always interested in hearing about your experiences with delivering cloud applications.

Paul Ressler
The Cirrostratus Group  

   The AI Software Business Model - Is it Different Than SaaS?   
This article by Martin Casado  and Matt Bornstein, partners at the venture capital firm Andreessen Horowitz, outlines how AI software appears to be different than SaaS software in some very fundamental ways. Their perspective is as investors in AI software and they believe that AI software will provide investment opportunities and transform businesses. However, there are three fundamental differences in the business model that they have observed:
  • Lower Gross Margins due to heavy cloud infrastructure spending and ongoing human involvement.
  • Scaling Challenges due to edge cases.
  • Weaker defensive moats due to commoditization of data models and challenges with data network effects.
A couple of their recommendations really resonated with me.
  • Avoid model complexity whenever possible - For example, avoid a model(s) per customer and try to have a model(s) that works for all customers.
  • Acknowledge and plan for the higher services content for the business.
  • Understand and manage the variable costs such as ongoing model maintenance.
Although not addressed in the article I believe that there will likely be some new types of pricing models which will attempt to address the variable costs and the challenges of ongoing services to alleviate the lower gross margins.

Resources From The Cirrostratus Group

The Cirrostratus Group Services
Learn more about The Cirrostratus Group's services here.