Better mouse traps in a mice-less world

Some conversations leave you energized and make you want the weekend to be over already so that you can go cracking. Nitin Jayakrishnan and I met today at Sashwatha Cafe – the new-ish food joint where you go for 1:1 conversations eavesdropped by 100 other people cramming in for the macrodose of ghee.

Nitin runs Pando - the logistics AI company. Go check them out.

This essay touches on topics we both discussed and some threads in my head that are worth putting out.

1. When Product marketing is the differentiator, the product is a commodity: As many of you know, I am a big believer in product marketing. Between 2021-2023, if you recall, we had been talking about 'product marketing' as the next battleground for SaaS startups. And then we started talking about AI-first. We were in the Coco-cola vs. Pepsi phase of SaaS, battling out on benefits and value props that were getting ever more frivolous. Luckily we didn't hire celebrity brand ambassadors. But we were close. Anytime marketing becomes a differentiator, the underlying product is a commodity. A SaaS company enabling a persona with some exoskeleton productivity wrappers, is a commodity.
2. Mousetraps in a no-mice world: SaaS companies are so used to selling better mousetraps. There are companies that are selling the idea of a world there is no mice. Better mousetraps are not good for a world with no mice.
3. Product managers' love for total submission to user interviews is fatal: Product managers are set on the current ways of building products. They vaguely understand workflow compression and productivity boost coming from AI and they apply AI to the current ways of solving problems (for the user). What you get is "forms replaced by chat bots." I will give you an example: In travel management systems, travel approvals are a function of financial controls. But travel itself is a revenue generating activity. A good AI product is not about eliminating forms with chat or reading bills with OCR. Instead a good B2B travel management software should recommend which conferences to attend and negotiate why a spend overage coming from meeting four additional prospects on an approved tour is worth it and where else the budgets could be cut. The problem is the lack of process-thinking. We still think in tasks and workflows, because we still do user interviews. We don't think in outcomes and we certainly don't re-engineer processes. Users have a vested interest in not compressing workflows that define their jobs. It was useful to understand what users want when we were replacing software for them (On prem to cloud, SaaS to another SaaS with better UX). Now we are replacing them. Its time for time & motion studies, and not persona-centered value props.
4. Who cares if AI works today or not. Assume it will. We can discuss if LLM works, if a particular process needs ML or Generative AI, and whether RAG works or not and if knowledge graphs + LLMs are a better way. But customers don't care. They are open to AI not for AI's sake but true reduction in cost or increase in revenue. Stable AI components can fall in place in their own pace but process re-engineering can start today. What does it mean to sell outcomes? It means we are becoming process consultants. It means we will start doing managed services to guarantee outcomes until those services can become truly agent-led.
5. TAM & Category - They do not matter. Every category has billion dollar SaaS companies. It's a waste of time deciding on one category over the other based on TAM and competition. If you are an early stage company and a VC is asking to justify TAM, they probably haven't understood this platform shift or you haven't done a good job of explaining what can happen. The numbers don't matter because no one knows the value capture potential of labor replacement. But there is no disagreement that its bigger than SaaS.

That is it. But most conversations such as the one I had today also inform our thesis at Moative. Did you read the thesis behind Moative? Read it here. Also do subscribe to my memo. I mean this one, right here on Linkedin (Pfft, such hawking, much wow!).

Do you know that build AI-first products and signing 10-year contracts is not a pipedream anymore and that customers don't care if its ML or GenAI? Listen to what Jim Fitchett said in the first episode of 'The Useful AI Hour' Podcast.

And finally, we are building AI roadmaps for SaaS companies and enterprises. We are building our own micro-apps. We work with business services companies to re-engineer their processes. Message me if you had a brain wave.