And it's here to stay.

Let's for a moment go down memory lane. Before "software ate the world", before software platforms took over? How would people access services? Either via word of mouth or a phone book, you would call and get someone to perform that task for you? You want a flight booked? You explain to the person on the other end of the phone what you want and they get it for you. You want tickets for an event? You call and they search what you want and book it for you. You want to find something? Ask a librarian and get right books.

Bottom line is: access to services required customers to explain what they wanted and have the relevant workflow being executed by a professional.

Software as a service

Then came SaaS which cut the middleman: you want something done, you now have full control (within the limits of the platform you decided to use). Example: I want to book a flight? Just log into any popular platform and find the right flight among so many options. You want to find a concert ticket? Any event platform would allow you to book a ticket in a matter of minutes.

The problem with SaaS is that it's heavily reliant on how intuitive the platform is. This required heavy investment in product design, experimentation, feature flag, recommender and information retrieval systems to allow users to perform the intended task with as little friction as possible. This approach is what allowed popular SaaS platforms such as Netflix, Amazon, Airbnb, Uber, ... to grow and offer their services worldwide.

But there are limits to this approach of offering service:

  • Some workflows are complex and technical (e.g. data analytics) they inherently require manual labor even with intuitive platforms
  • Optimising user experience takes coordination between product, data science, engineering and few companies get it right
  • Information retrieval systems have their limits and platforms can require users to invest time in understanding how to find what they need

AI Software as a service

This is a new age of software as a service. It's still software but the emergence of large language models and companies such as Perplexity shows a new way of providing delightful customer experiences.

A side note here is that Google Search is probably the product of the SaaS era that's the closest in terms of experience to the new user experience and maybe that's why the most popular use case for this new age is web search (e.g. Perplexity).

AI software as a service minimalizes the user experience: a chat interface, a few prompts to get you started. This does mean that products can scale with smaller product organisations. The need for A/B testing and user experience optimisation is lesser (though I don't believe it will go away) because of a leaner user workflow.

However, this new shift comes with a few challenges:

  • Exposing complex workflow and their results via a simple interface without the user having a sense of loss of control.
  • Maintaining trust by building AI systems that have high task completion accuracy

A change in the way we build products

This means a shift in our product organisation operate.

Product focus on productivity firsthand rather than on building delightful experiences: the reason is because the predefined chat template removes the flexibility that defined the SaaS era. Product engineering builds features for an AI system that becomes the intermediary between the user and the task to complete and the infrastructure that enable this AI system to access the right data and 3rd party systems to compete the task. AI engineering becomes an integral part of product engineering rather than a data function (which is a trend that we see). Product data science, product machine learning engineering are both functions that are less relevant in this new paradigm.

This is an exciting time to be building products and startups. There are opportunity to disrupt industries but the need to go from 0-1 quickly has never been higher. As competition emerges faster and barriers of entry are lowered the most important question remaining is how do you differentiate yourself when customer experience shrinks to a minimalist version.