Looking back to move forward
David Peterson

With each passing week, it becomes clearer just how much of an impact generative AI will have on the future of software, and how little we know about what that future looks like.

For those of us who live and breathe technology, this is both wonderfully exciting and wildly disconcerting.

As an example, I was catching up with a former Airtable colleague a few weeks ago. He’s an extremely talented engineer. Has a killer product instinct. And I’ve never seen him more excited than when he’s talking about the latest LLM-powered app he’s hacked together. But he is, unabashedly, incredibly anxious as well. Why? Because he doesn’t know what to build. No idea seems defensible these days. New applications can suddenly become features. Infrastructure can be easily subsumed by incumbents. How can you decide what to build when the ground is shifting underneath you so quickly?

In times of dislocation, such as this, I find it helpful to go back and read the contemporaneous thoughts of people I respect during previous moments of uncertainty…which is how I found myself reading old posts from Fred Wilson blog over the weekend.

There’s a lot of great content from A VC in the mid-to-late 2000s. The rise of the cloud meant the cost of producing software fell precipitously (sound familiar?), and Union Square managed that uncertainty better than almost anyone.

There’s one post I keep coming back to. It’s called “The Dental Software Story” and in it, Fred recounts why Union Square’s very first thesis was all about investing in “networks.”

He tells the story of a software entrepreneur that builds a comprehensive dental office management system called Dentasoft. Dentasoft is expensive, but it takes off, hits $100M ARR, and goes public. A few years later, some young entrepreneurs built a competitor to Dentasoft called Dent.io. Dent.io is modern, mobile-first, and only one-fifth the price. Dent.io starts stealing customers from Dentasoft. Around this time, an open source community crops up to build an open source version of dental office software. A hosted version pops up and becomes very popular with forward thinking dentist offices. And so on.

The moral of the story is that software alone is a commodity. As Fred wrote: “[t]here is nothing stopping anyone from copying the feature set, making it better, cheaper, and faster. And they will do that…[Back in 2003, we saw the cloud coming but did not want to invest in commodity software delivered in the cloud. So we asked ourselves, “what will provide defensibility” and the answer we came to was networks of users, transactions, or data inside the software. We felt that if an entrepreneur could include something other than features and functions in their software, something that was not a commodity, then their software would be more defensible.”

There’s something so comforting about the way history rhymes. Yes. Software is a commodity. And at this moment, as the cost of building software trends toward zero, never has that been more true.

So, what matters now is what has always mattered: First, do you have a unique customer insight? And second, can you leverage that insight to build a business that has line of sight to some sort of durable, competitive advantage?

There are many paths to building a sustainable competitive advantage. Network effects work. You might also benefit from cost advantages. Or perhaps high switching costs or high barriers to entry. There isn’t just one way to get there.

But I think we all convinced ourselves, over the past few years, that you didn’t need to do it at all. Maybe a good customer insight is enough, we’d say. Perhaps product sense is all that is needed, we’d argue. Indeed, the cost of building software had grown so much that we convinced ourselves that capital itself was a moat. I’m not sure if that was ever true. But it’s certainly not true any more.

So if, like my friend, you’re feeling a bit anxious, go read some old posts from Fred. He’ll remind you that we’ve been here before. And sometimes the best way to figure out how to move forward is to look back.

David

EVENTS

May 31 / US Immigration Best Practices
Jennifer Schear, Founding Partner, Schear Immigration Law Firm

FROM THE BLOG

LLMs and the Future of Customer-built Software Design
How will LLMs change software development and design?

Navigating AI’s iPhone Moment
A venture perspective on LLMs and what’s next…

Principles for AI Product Design
Or how we could all learn a little from Google’s conversion optimizer.

Has Everyone Gone Chatbot Crazy?
Large language models have taken the (tech) world by storm, and all of a sudden it’s 2016 again.

WORTH READING

ENTERPRISE/TECH NEWS

The AI agent abstraction. James Detweiler at Felicis Ventures takes a look at the LLM/GenAI space and concludes that perhaps the best mental model is actually “agents” as opposed to models. We’ve already begun to see this play out with the rapid rise of AutoGPT but James' framing and analysis is very helpful because it frames the current technology wave in its proper - broader - context: the next natural step towards workflow automation. He also provides a roadmap for winning in the space: “The winners of this race will put a fleet of agents in your pocket and on your desktop. Five prerequisites for getting this right include privacy preservation, on-device inference, agent-to-agent interaction, persistent memory, and quality controls.”

What’s a model without the data? Wired reported that Stack Overflow joined Reddit in announcing that it would charge AI giants for training on its data. This makes a ton of sense, reflects the reality of where the value is really being created - and suggests that the future business dynamics of the AI ecosystem will be complex and focused on value capture.“Every AI developer is seeking to bring down the huge costs of developing large scale AI systems, which take enormous amounts of expensive computers to power. Having to pay for data they once grabbed for free could extend the already unclear timelines to turning a profit on their emerging technologies.”

MSFT is building an AI chip. According to The Information, Microsoft is joining Amazon, Google, and Facebook in building their own proprietary silicon to support their AI efforts. “Microsoft has another secret weapon in its arsenal: its own artificial intelligence chip for powering the large-language models responsible for understanding and generating humanlike language. The software giant has been developing the chip, internally code-named Athena, since as early as 2019, according to two people with direct knowledge of the project. The chips are already available to a small group of Microsoft and OpenAI employees, who are testing the technology, one of them said. Microsoft is hoping the chip will perform better than what it currently buys from other vendors, saving it time and money on its costly AI efforts.”

HOW TO STARTUP

When is ARR not ARR? Dave Kellogg at Balderton put out a very helpful reminder that “ending ARR” is really not enough to understand how a business is doing. In short, he reminds us that the change in ARR in a giving period is the result of a number of events and efforts - and that all of them must be understood in isolation and together to really understand what is going on. It’s well worth a read. The reason founders need the level of clarity and transparency that Dave is arguing for is clear: “Look, bad quarters happen. I’m not angry about the bad quarter. I’m angry when people try to pretend a bad quarter was a good one. Or, even more scarily, at the prospect that someone might actually believe that a bad quarter was a good one.”

Long sales cycle? Deal with it. Jason Lemkin argues that for many companies, long sales cycles are an unavoidable part of life. He (like us!) is not afraid of the long sales cycle, and argues that companies need to plan for that early on. “...as you get into Year 3 and beyond, you’ll need lots of irons in the fire. You’ll get better at predicting when and how bigger deals close. These longer sales cycles, so long as they are shortened by a great VP of Sales as much as possible, will help you start each quarter with some good stuff already ready to close.” Also, there is this gem: “But don’t let short-term thinking, here again, add stress. You are a tiny vendor, even at $20m ARR, let alone at $0.2m ARR. It’s entirely fair for a Big Company to ask for a pilot to see if it’s really going to work. More importantly, it’s how they buy from new vendors. If you try and change the way a Big Company buys, then best case, you add risk and time to a deal. And worst case, you simply break it. So when folks tell you not to do Paid Pilots or similar deals, don’t listen. At least, not in the early days. Do what it takes to get the customer. No one has heard of you. The prospect is taking a huge risk working with you. Take a tiny bit of risk back. Best on yourself, and deliver the service the way that makes the customer comfortable.”

HOW TO VENTURE

European venture fell off a cliff. According to Bloomberg, “Europe’s venture capital firms are on pace for their most meager year of funding since 2015, a worrying sign for technology startups already facing a sharp slowdown in deals and exit opportunities. Venture funds based in Europe raised €3.4 billion ($3.72 billion) from limited partners during the first quarter of 2023, down from €7.4 billion in the same period of 2022, according to new data from research firm PitchBook.”

PORTFOLIO NEWS

Januar, crypto financial services provider, received full Danish payments license.

Planable launched Planable AI, the ultimate writing wingman, on Product Hunt.