AI for Creation and Maintenance
bolt and v0 for everything else
After playing around with Bolt and v0 the past couple of months, I’ve been blown away at how incredibly easy it is to “build” something today. For $20/month, someone who would never get hired as a software engineer can build basic custom applications. It’s not the death of software engineering, but it will be the evolution of every other knowledge worker.
I’m excited for the next generation of software platforms that help us both create and maintain new digital products.
AI for Creation
The first area I’m excited about is AI for Creation. Bolt, for example, uses AI to remove the technical barrier to designing websites (I present my personal website as evidence: nanduanilal.com). In particular, the non-deterministic nature of generative models allows users to cycle through new ideas quickly and cheaply with full code access as needed. This leads to an important behavior change. Instead of teams exploring just a couple of ideas, they can now cycle through dozens of prototypes in the same time. The goal is not necessarily to one-shot exactly what you’re looking for, but rather explore and iterate through ideas more efficiently. Website generation is one example, but this creation problem exists in multiple domains:
AI for website generation
AI for app generation
AI for architecture
AI for engineering
AI for gam design
AI for media production (film, audio, etc)
Most of these categories have large incumbents, but I think startups have an edge. Incumbents have built both product and go-to-markets focused on a specific user persona, but AI will redefine who should consume these tools. For example, AI for website generation might serve new personas that fall somewhere outside of the Squarespace and Figmas of the world.
Another category where large incumbents serve various personas is AEC. Similar to how product teams coordinate to design, build, and improve digital products, teams of architects, engineers, and contractors need to coordinate to build physical structures. There is an opportunity here to carve out AI-powered features and use cases without having to rip-and-replace Autodesk on day one.
AI for Maintenance
The second area I’m excited about is AI for Maintenance. As more people “create”, maintaining these applications and systems will be important. I think AI is uniquely positioned to help because maintenance and monitoring of complex systems is largely about digesting large amounts of data and making adjustments. AI agents will proactively detect issues, recommend fixes, and even autonomously make adjustments to optimize performance.
As a result, we should have higher quality, resilient applications that counterintuitively cost less to serve. There are a number of domains I that fall into the “Maintenance” bucket:
AI for QA and software testing
AI for cloud management
AI for code reviews
AI for incident response
AI for the data stack
AI for rearchitecting codebase
AI for SEO
AI for security pentesting
Here, incumbents fall into two buckets: 1) services companies and 2) visibility tools.
For the former bucket, security is a market with lots of services in categories like pen-testing or MDR. My favorite fun fact here is that Accenture generates more cybersecurity revenue than Palo Alto Networks does. Companies like Arctic Wolf and Expel have shown you can build venture-scale businesses that still have services, but there’s an opportunity for AI-native companies that can rethink this model entirely. The gold standard here is not just automating human workflows but building autonomous software that maintains itself.
For the latter bucket, we can look at companies like Datadog, Dynatrace, etc. which sit in the cloud observability space. These tools provide rich data on what is happening with your applications and cloud, but rely on people to act on this data. I think we’re going to see this observability + human workflow model collapse as new companies may be able to move directly from telemetry data to a resolution or change.
Conclusion
One of the most important things that AI will do is lower the bar to “create” things. The past couple of waves of technology democratization waves enabled communication (internet) and made software accessible (SaaS). This time, AI unlocks intelligence, allowing us to “create” more and “maintain” less. If you’re building or ideating in one of these categories or have some personal website inspo, shoot me a message!
Nandu