Slack used to be the scrappy startup and seemed like a low cost choice to keep in touch. Over the years it has become a massive line item on each of their large customers budgets. It is the same story with many SaaS products companies rely on to operate. Chamath Palihapitiya, a VC and co-host on the All-In podcast, started an “incubator” that aims to solve this issue with “AI”. They are calling it 8090 Solutions (8090.inc) in 2024. Chamath claims it will deliver 80% feature complete software for a 90% reduction in price. Sounds like a great solution to balance the per seat budget creep of Slack et al. right?
Getting an 80% feature complete Slack for a 90% discount sounds great. But I don’t think that 8090 as it is envisioned will work. Chamath seems to be assuming if you can offer the same thing for a cheaper price you can make a killing. I’m skeptical of the initial claim of 90% cheaper software enabled by AI, but let’s assume that’s true.
- Step 1: Make something for cheaper
- Step 2: ?
- Step 3: Profit.
What could go wrong?
Let’s step back and try to understand what this type of business model will do for the entire ecosystem of software development. If making software is 90% cheaper, then you can make 10X the amount of software. This is a significant market change. Often we make decisions assuming *all other things remain constant. But you can only exploit 10X productivty for so long before everything starts changing and your previous assumptions are no longer valid.
To consider how these assumption might change, I came up with a few different ways software development might change given this efficiency gain.
Million SaaS Blooming Future
If AI enabled software developers can put out 10X the software, we might see a million SaaS startups blooming. Due to increased competition, each will be smaller than the unicorns of the Web 2.0/ZIRP era. Additionally, these SaaS products will be highly differentiated compared to SaaS today. Most businesses would be able to find a software that matches their exact needs, rather than conforming to the processes of the software.
In this future, you can state your needs and someone in the market probably made it. The price might be comparable to today, but it would be much more tailored to our needs. Developers in this future should expect to be paid less per SaaS, but may be able to offer a portfolio of apps, or work on several side projects and have those as extra income streams.
Discovery will be the biggest issue in this future, so expect that having an audience and/or knowing how to do marketing will be key differentiators. Perhaps a whole new line of AI enabled discovery tools will emerge to solve this problem.
Customized Future
If AI enabled software developers can put out 10X the software, we might see that customization becomes an expected part of any SaaS. Today we have static SaaS that you purchase and updates come out incrementally. A customized future is one where, as part of the purchase price, you can customize features of the SaaS to follow your business processes or logic. If it’s 10X more efficient to make software one way to keep prices high is to offer customization. If Slack wants to justify $15 a seat forever they will need to give you a reason to stay. Customization could give them that justification. Want role based access control that syncs with your ERP? Sure. Do you want a boss god mode so you can snoop on your employees? Why not Mr. Toxic Middle Manager.
This is a lot like how large ERP software was customized for Fortune 500 companies (SAP, I’m talking about you). The big difference is that in this future you just need 1-2 developers instead of a team of road warriors with their inflated salaries and per diems.
Personally I like this future better. Partly because I’ve been living it for several years.
My Own Customized Future is Now
Since 2018 or so I’ve been building internal software for various ventures my wife and I have started. Each of these businesses has unique software needs, from order tracking to finances, payments, etc. Often our processes didn’t match up with the software that was in our budget, so I just started making it myself. This included a factory order management system, student information system for our university, and a certificate tracking system for our jewelry retail business. Where existing software worked well we chose that, including Moodle for a learning management system, and Shopify for a retail POS and inventory system.
I used Anvil for the projects I built. I started with a core set of features (an MVP if you will) and then updated the software to confirm to the processes we developed. This worked well since this software was usually added to a new business, so the software was updated as the business processes were updated. With a new business it is nearly impossible to plan for everything up front, so choosing an off the shelf software would have been nearly impossible.
There was a common theme that emerged with all of this software. When users wanted something changed, I changed it, usually in a day or so. Once they realized they could change things, they kept requesting changes until it was exactly how they wanted it. This was good for the employees because they were able to take ownership of their own tools. After 6 months or so from first launch changes became much less frequest, one a month or so or less.
In a way this has been great because I can launch an internal MVP, get feedback, and then interate to something that is mostly complete. The cost was my time and hosting. But it wasn’t all positive. This method of software development created a situation where I was building a plane while flying it. Schemas change, features are abandoned, and generally everyone feels that it isn’t “ready” until the 6-12 months later.
Personally I feel that a customized software future is preferrable to millions of SaaS applications to choose from. We already have an abundance of choice. It would be better if we had an opinion about what we wanted and then made that a reality rather than trusting an influencer from the internet.