• The Angle
  • Posts
  • The Age of Artisanal Software is Over

The Age of Artisanal Software is Over

The Angle Issue #229

The age of artisanal software is over
David Peterson

A few weeks ago, Chris Paik from Pace Capital set the startup world abuzz with a short post titled “The End of Software.”

In the piece, Chris argues that, as the cost of code goes to zero, software development will be disrupted in much the same way journalism was by the Internet driving the cost of content creation down to zero in the 2000s. He ends with the pithy line: “Majoring in computer science today will be like majoring in journalism in the late 90’s.”

I think Chris is directionally correct, but wrong on the specifics (which, in a variant of Cunningham’s Law, is why I think the piece went so viral). But where Chris and I agree is that there will be an explosion in the creation of software.

This has been a long time coming. Software production has been too expensive for too long. Why is software so expensive? Because software engineers are expensive. For all the tech sector’s obsession with productivity, the production of software is a decidedly artisanal act. There have been improvements in software production pipelines, sure, but let’s be honest…it is still just highly paid humans typing away on keyboards. That does feel a bit old-fashioned, doesn’t it?

Whether or not we have fully internalized it yet, LLMs have changed all that. It’s uncontroversial at this point to say that LLMs are surprisingly good at writing code. Is the code as elegant or performant as the code written by an experienced software developer? No. Could you ask an LLM to write a custom piece of enterprise-grade software? Also, no. But even today LLMs are good enough to empower non-technical people to write small snippets of code - tiny, trivial, seemingly insignificant lines - to solve problems which they previously thought impossible to solve by themselves. And that is more meaningful than it seems, because it has the potential to shift the clearing price of software itself.

As I argued this past week, we may not be witnessing the end of software, but we’re definitely witnessing the end of the MVP. The bar for software that people are willing to pay for is getting higher. And I think this is one of the reasons. As non-technical people are empowered to write more of their own code at the margins, their willingness to pay for new software will go down. They’ll only pay for something that they couldn’t build themselves. And, for the first time in decades, software won’t be the one causing deflation in other industries…prices will start to drop right at home.

What will that look like? It’s hard to know. I would imagine that the willingness-to-pay for that Shopify app with 50 clones will trend down sharply. My instinct is that the deflationary impact of LLMs will be most muted for products that embody some expertise that is very difficult to reverse engineer (e.g. deeply technical niche vertical software) or core pieces of infrastructure (e.g. systems of record or databases/data infra). I think there will also be an opportunity to build the platform on which a lot of this new LLM-prompted code is written. Perhaps the systems of record will own this themselves, but that might also look like an AI agent platform that enables users to orchestrate agents to do work for them across existing systems of record. These are all areas we’re spending a lot of time on as of late.

I wish I could end this post with a bold prediction, but while the immediate future seems clear enough, it’s still so early to pontificate on the second and third order implications of this shift. (And don’t look to investors for predictions anyway). All I can say is if you look across history, you’ll notice that every wave of technological innovation has been catalyzed by the cost of something expensive trending towards zero. It happened with Gutenberg and books. Calories. Transportation. Electricity. That suggests to me that we’re on the cusp of something monumental. So, let me ask you: what do you think will happen when code is free?



Revenue Durability in the LLM World
Everything about LLMs seems to make revenue durability more challenging than ever.

A Digital Fabric for Maritime Trade
Why we invested in Portchain.

Three Keys to the Kingdom
The sometimes-competing and sometimes-aligned goals that early-stage founders must manage.


France / GenAI. Mistral raised €600M, led by General Catalyst, for its open-source foundational model.

Germany / GenAI. Cognigy raised $100M, led by Eurazeo, for its AI-powered contact center automation platform.

UK / Batteries. Echion Technologies raised £29M, led by Volta Energy Technologies, for scaling up the production and go-to-market of its niobium-based fast-charging battery material.

UK / SaaS. Visibly raised €7M, led by Creandum, for its compliance-as-a-service software platform.

Germany / DevTool. Restate raised $7M, led by Redpoint, to build its lightweight “workflows-as-code” startup, together with a team of founding contributors to stream-processing framework Apache Flink.

Germany / Marketplace. Cardino raised €4M, led by Point Nine, for its peer-to-peer electric vehicle marketplace.

Switzerland / SaaS. Rivia raised €3M, led by Speedinvest, for its AI-powered clinical trial data collection and comprehension platform.



The future of big tech. CB Insights dove into the current big tech landscape. Key takeaways include 1) The six big tech companies, Apple, Microsoft, Google, Amazon, Nvidia and Meta dwarf the private tech sector. Combined these six companies’ market cap is nearly 3x that of the entire billion-dollar unicorn club. 2) Big tech has returned to running efficiently by reducing headcount and slowing hiring “with a focus on keeping expenses in check”. And 3) Lastly, “in the US, big tech companies’ R&D investment outpaces that of overall US venture funding by a wide margin. This underscores the scale these companies play in driving tech innovation”.  

AI as a feature, not a product. Apple's latest WWDC highlighted its vision of GenAI as a feature, not a stand-alone product. Unlike other companies creating dedicated AI products, Apple is integrating AI into existing apps and OS features, enhancing usability. New tools include Writing Tools, Image Playground, and Genmoji, alongside a more advanced Siri. According to TechCrunch, “Apple’s AI, Apple Intelligence, is boring and practical — that’s why it works. Instead of trying to overwhelm users with too many AI features to count, the Cupertino tech giant is carefully rolling out AI where it believes it could be useful. That means the tech won’t be included where it could be much of a threat to the carefully crafted consumer experience of using an Apple device.” The shift reflects a broader industry trend, with practical AI applications gaining prominence over flashy, independent AI gadgets and products.

Flying cars. Last week we featured a thread on Waymo, keeping the future of cars theme going, this week we’re covering electric flying cars. Electric vertical takeoff and landing (eVTOL) aircrafts are “rising to the challenge of delivering clean and cheap air taxi services”. They’re substantially quieter than helicopters as they use multiple small propellers that spin half as fast as a chopper’s rotor — avoiding the annoying, low-frequency sound pulses created by the big whirling blades. “Noise reduction isn’t the only perk of this propulsion approach. Many propellers offer safety redundancy that helicopters, with only one or two rotors, cannot, Moore says. If one propulsor fails, others can take the load. And unlike helicopter turbine engines that spew fossil fuel exhaust, he notes, “electric motors are low- or zero-emissions.”” Several eVTOL builders have been founded in the past decade, and a few are nearing commercial certification from the US Federal Aviation Administration or its European counterpart. If and when these “flying cars” are approved for flight, they will likely be used for short flights, typically between 18-25 miles. “These short, high-speed hops could carry commuters between city centers and airports or transport cargo and packages. Militaries may want eVTOLs for casualty evacuations or logistical supply. Other potential uses include air ambulances, donor organ delivery and police transport, as well as scheduled shuttles and ecotourism trips—and, of course, personal flying cars.”


The end of the MVP. Angular’s David Peterson opined on why Reid Hoffman’s advice to “launch so early that you’re embarrassed by your product” no longer works... It may have worked in 2009, but those days are gone. “I’m not saying you should build your product alone in the garage for years. By all means, talk to customers. Get feedback. Build something people want! But don’t expect to win with something that’s just good. Because that’s not good enough anymore.”

Seat based pricing. Jamin Ball from Altimeter shared another fantastic post in Clouded Judgement on seat based pricing. He argues that if “AI delivers on its promise, it may spell the end of the SaaS business model as we know it.” He continues “...One belief I have is that the future of application software in a world of AI agents probably looks a whole lot more like database software…These worlds are merging (database and applications). At the end of the day, most software today is just a UI on top of a database. We need the UI for human consumption. But we don’t need it for AI agent consumption. I’ve long STRONGLY believed that data and data infra are one of the core beneficiaries of AI. Most of my venture career has been spent investing in data businesses. One of which, Tabular, was just acquired by Databricks for >$1b. Others, like dbt Labs, Clickhouse, Airbyte, Prisma, etc are seeing really strong growth. The value of differentiated, high quality data infra has never been higher. But let’s take this one step further. Both Databricks and Snowflake have heavily foreshadowed (or released products) that enable applications to be built on top of them. I think this is clearly the future. Applications built more for AI agent consumption, built on top of leading data platforms. The data platform companies will BECOME the application providers themselves. I’m excited for this future!”

Co-founder partnerships. Harvard Business Review looks at an issue we’ve unfortunately seen frequently plague startups - co-founder partnerships failing. In fact, “up to 43% of startup founders ultimately buy out their cofounder due to interpersonal rifts and power struggles”. The authors identified the root cause often stemms from a mismatch between lead founders (the founder with the idea) and co-founders (founders considering joining the lead founder). An interesting finding the authors discovered was that at the onset of the co-founders meeting, “some lead founders focused almost solely on the business idea and the needed skillsets during their conversations. Others, instead, focused primarily on interpersonal factors (i.e., getting to know the person, discussing mutual interests, etc.), and devoted only a small amount of time during the initial conversations to discuss the business idea and/or needed skillsets. We found stark differences between these two groups in terms of business partnerships actually being formed. It may seem counterintuitive that a focus on interpersonal topics rather than business specifics is more persuasive, but our data provides compelling evidence that this is the case. Those who focused primarily on interpersonal factors during the initial conversations were 70% more likely to find and stick with a business partner. This does not mean that lead founders can never discuss skillsets; many of the successful conversations in our sample did include a discussion of skillsets, but those that were successful also established an interpersonal connection by, for instance, discussing shared interests, hobbies, communities, contacts, or values.”


VCs quiet quitting. Pitchbook recently reported that 13% of VC firms are not planning on raising another fund - “double the rate in H1 2023, when 6% said they had no plans to raise another fund.” Many of the VCs winding down their firms were emerging managers, GPs who got into the venture industry in the early 2020s when there was substantially more LP appetite for the asset class. Serial entrepreneur Joe Procopio argues that “ultimately, this is good for startups in the long term, because it ends a 10-to-20-year cycle of cheap money propping up a system that hasn't placed innovation at the forefront of its thesis in a long time. In business, when innovation is forced to survive on its own merit, you get maximum innovation.”


CruxOCM’s CEO Vicki Knott recaped the American Petroleum Institute Pipeline Control Room and Cybernetics Conference 2024.

Reco’s CEO Ofer Klein shared how Reco is transforming the identity-centric security industry.



or to participate.