Learning Library

← Back to Library

AI-Driven Customization Transforms SaaS

Key Points

  • AI is shifting SaaS from a “one‑size‑fits‑all” model toward **customization at scale**, letting providers embed personalized, workflow‑aware intelligence rather than just generic chatbots.
  • The **cost of intelligence is approaching zero**, dramatically increasing the supply of AI‑driven insights and making traditional predictive features a commodity rather than a differentiator.
  • Because AI makes both mass customization and cheap intelligence feasible, **new entrants can more easily build differentiated SaaS products**, while incumbent firms face significant challenges retrofitting their legacy platforms.

Full Transcript

# AI-Driven Customization Transforms SaaS **Source:** [https://www.youtube.com/watch?v=JpC_N4aBOBI](https://www.youtube.com/watch?v=JpC_N4aBOBI) **Duration:** 00:07:31 ## Summary - AI is shifting SaaS from a “one‑size‑fits‑all” model toward **customization at scale**, letting providers embed personalized, workflow‑aware intelligence rather than just generic chatbots. - The **cost of intelligence is approaching zero**, dramatically increasing the supply of AI‑driven insights and making traditional predictive features a commodity rather than a differentiator. - Because AI makes both mass customization and cheap intelligence feasible, **new entrants can more easily build differentiated SaaS products**, while incumbent firms face significant challenges retrofitting their legacy platforms. ## Sections - [00:00:00](https://www.youtube.com/watch?v=JpC_N4aBOBI&t=0s) **AI-Driven Shifts in SaaS** - The speaker explains how AI is fundamentally reshaping SaaS by turning previously discouraged customization into scalable, AI-powered personalization, altering the industry's economics and growth dynamics. ## Full Transcript
0:00how many times have you heard sas's debt 0:03software as a service isn't going to 0:04keep going anymore with AI or on the 0:07other hand no SAS is fantastic AI is 0:10going to make SAS work even better it's 0:12going to transform the industry and give 0:14us New 0:15Growth I'm not interested in the bull 0:17case I'm not interested in the bare case 0:19I am interested in the fundamental 0:22Dynamics in software and how they've 0:24changed with AI and I think that's 0:26particularly true for software as a 0:28service and I want to lay out three of 0:29them in this video that as someone who 0:31works in that space I have seen start to 0:34come into play over the last couple of 0:36years and they're just getting more 0:39prominent all right number 0:41one customization at scale it used to be 0:44a bad word in fact coming up in product 0:47in the 2010s in software you were taught 0:51not to do customization customization 0:53was a bad word as recently as 2020 0:562021 you would fail product questions 1:00if you 1:01proposed too much customization for big 1:04customers the idea was software should 1:07tastees like chicken it should be 1:09consistent it should be somewhat 1:10flavorful maybe it's not everyone's 1:12favorite but it works to put food on the 1:15table and that was so consistent and so 1:19true that the actual unit economics of 1:21software were also consistent so a 1:24private Equity Firm could roll up a 1:26bunch of different SAS companies treat 1:28them all the same and get similar margin 1:31and similar cash flow out of all of 1:33them that is going to 1:36be different going 1:39forward so the fundamental change that 1:42AI has enabled is customization at scale 1:46that's much bigger than a chatbot 1:48fundamentally instead of this Dynamic of 1:50we build the same software for everyone 1:53AI is driving customization expectation 1:56and customization capabilities so 1:59fundamentally customers are used to now 2:01getting personalized answers whenever 2:02they want them from the chatbot they 2:04expect more the bar has been raised this 2:06is good at the same time AI is 2:11enabling software providers to deliver 2:15customization in their apps and that's 2:16much more than just slapping a chatbot 2:18onto the website it's actually being 2:20thoughtful about how you wrap in AI so 2:23the customization Works in ways that 2:26enhance customer workflows and sort of 2:28provide wraparound experiences rather 2:30than a one-size fits-all flat 2:33experience so that's lever number one I 2:36think companies that are able to take 2:37advantage of that will win and my 2:40suspicion is that this is actually an 2:43area where new companies are going to 2:47find it easier because wrapping that 2:48kind of customization into existing 2:50companies is really hard work it's not 2:52impossible we see companies doing it but 2:56it's 2:57tough lever number two the cost of 3:01intelligence at the end of the day 3:03intelligence is going to Zero from a 3:05cost perspective that means the supply 3:07of intelligence is going up I don't know 3:09a millionfold we'll use a million just 3:12for 3:12working if it's getting that easy to 3:16access human level 3:18intelligence that means that the 3:21predictive deterministic intelligence 3:24that software 3:25provides is only going to be valuable 3:29where where it actually solves a 3:31business 3:32problem and does so in a cost effective 3:35Manner and removes liability from the 3:38company in 3:39question and I'll get into all of those 3:41so traditional deterministic software 3:44software without AI in it it is 3:47intelligence that solves a problem it's 3:49just brittle intelligence it just solves 3:51that one thing it's not general purpose 3:53intelligence in any 3:55sense and so with general purpose 3:57Intelligence coming in the opportunity 4:00is really to enable more effective 4:05solutions for companies across a wider 4:08range of use 4:09cases which sounds a lot like 4:12customization but doing so without 4:15imposing price increases or with limited 4:18price 4:19pressure because at the end of the day 4:22the other Dynamic here with a cost piece 4:24is that there are a lot of ankle biters 4:27or new funded VC companies that that are 4:29going to be able to come into the 4:31space and compete on the basis of cost 4:36because the cost of a general purpose 4:38model is now so low they may not be 4:42actually as effective but the 4:44substitution value may be there if the 4:46price competition is strong enough net 4:49net this is going to be more price 4:51pressure on SAS companies and they are 4:54going to need to find ways to deliver 4:56more intelligence at a lower cost across 4:58a wider range of use cases and they're 5:01going to need to do so while providing 5:03what they traditionally have provided 5:05which is a little bit of a liability 5:06shield no one got fired for choosing 5:10Salesforce and we underestimate the 5:12value incumbents have because they are 5:16incumbents and so in this situation I 5:18would actually give the advantage to 5:19incumbents because they have the 5:20treasury to invest in new AI solutions 5:24they have the margin to absorb some 5:26price pressure and they have that EST 5:29Lish reputation that acts as effectively 5:31a liability 5:33Shield lever number three workflow 5:36breakage so at the end of the day one of 5:40the most unpredictable things about AI 5:42is that it entirely removes certain 5:45niches in the ecosystem that previously 5:47existed I think my favorite example 5:50right now of this is the budding 5:54competition between figma and v0 by 5:58versel so v0 her is an llm driven 6:02front-end application building tool but 6:06the problem is it does such a good job 6:08at coding up front end that you kind of 6:11don't need figma and I'm not the only 6:13one saying that there's a lot of other 6:15folks who are wondering and noticing 6:17they don't just don't use figma as much 6:19as they needed to that's an example of 6:21traditional workflows getting upended 6:24because llms enable such fast paths to 6:29value 6:30and my working assumption here is that 6:32anything that is farther away from the 6:34code base is more at risk because at the 6:37end of the day software is about 6:38delivering code to customers that works 6:42if you have job functions if you have uh 6:46pieces of the value chain tools that 6:48support job functions that are farther 6:50back of the code farther away from the 6:53code I think you're inherently at risk 6:55workflow breakage is going to lead to 6:58huge winners overnight 7:00is people start to eat into traditional 7:02longer workflows and dramatically 7:04compress them I don't know where all the 7:06break points are going to be in 2025 but 7:08they're coming if you thought this was 7:11interesting I put a good solid substack 7:14up on this I think it's something we 7:15need to talk about a whole lot more as I 7:17said I'm not interested in whether it's 7:19a bull case for SAS or a bare case for 7:21SAS I want us to understand the levers 7:23and how we as Builders can use those 7:26levers to build companies that have 7:28value