Developer Tools keep being on-trend
As more and more technie quit their 9–5 to start their own ventures, more Developer Tool startups have been founded since late 2019.
There’re over 30 new Developer Tool startups funded by YC in the last 2 batches. YC also favors Developer Tool as one of their Requests for Startups:
“Software used by large companies is still awful and still very lucrative.” — YC’s Requests for Startups
Collectively, there’re 2 types of Developer Tool startups that investors are more likely to fund:
- Making development easier and cheaper by automating or optimizing part of development workflow: deployment, code review, design, operating, tracking, etc.
- Low-code and no-code solutions to digitalize non-technical industries.
JAMstack enables frontend developers to do more
2019–2020 seeing the growth of JAMstack along with the emergence of new serverless technologies, frontend developers can do a lot more than just a single-page application.
- A static site generator reads Markups, then populates data into UI components, and generates a static multi-page application, which is optimized for both availability and SEO.
- APIs are written and deployed as Serverless functions talking to Serverless Databases, or even are designed completely with no-code CMS like Strapi.
With existing technologies like Vercel, Next.js, Strapi, etc, a modern website can be built in hours and deployed in seconds, enables leaner development for startups, at least at early stage.
Serverless is a trend…, and an obstacle
Facilitating the growth of JAMstack is the growth of Serverless.
Serverless means provider-managed services, no server to manage, pay-per-use or scale-to-zero, continuous and unlimited scaling. Serverless is great but has drawbacks.
However, because of serverless’s simplicity in unit deployment and operation, developers tend to try to move all of their services into serverless functions without thinking of the tradeoffs.
Serverless’s great characteristics are true, and kept as is, as long as it’s “truly” granular, which means it can do a transaction from beginning to end on its own, without calling any other services. Because if a serverless function has to call and depends on others, its simplicity disappears and is replaced by the complexity of communication.
Although the complexity is the nature of distributed system, not just serverless, over-engineering by writing everything as serverless functions is like reinventing compiler optimizations and Database transactions over the network.
Using serverless exhaustedly also creates bottlenecks to development because lots of serverless services are simply not available for local development. Going 100% serverless also means not being able to use open-source, which is meant to be run and managed by developers.
Anyway, serverless’s unique advantage of simplicity stays, and developers will keep going crazy for it, it’s just that some new technologies need to be created to make serverless “truly” simple.
Microservices is great, but not all people have the bandwidth and expertise required to manage it
As serverless, microservices architecture has its own advantages but shares the same complexity of a distributed application.
Microservices eliminates the single point of failure and weightlifting problem, allows teams to move independently, and experiment more. However, the heaviness of managing and orchestrating dozens or hundreds of services is not a thing that every team can suffer.
Again, microservices’s advantage of being able to try more and iterate fast stays, it definitely has value to many types of development, it’s just that some new technologies need to be created to lessen the heaviness.
WebAssembly is not just for the web
WebAssembly (WASM) + WASI offers the same ability that Docker is currently facilitating the whole modern technology world: Sandbox + Portability. It even has better advantage of being able to run on limited-resources metal.
With that ability, there’s a chance to build a new foundation and reinvent how programs are made and deployed.
All code is wrong, no code is true
Every code you’ve written is wrong and costly. All code just works by coincidence, even it works in all existing circumstances, every new feature added increases its likelihood to be wrong.
The only code that’s true, is no code. A product, which can help eliminate fractions of code, can help avoid tons of wrongness, and save tons of dollars.
Open-source is the way to go for Developer Tool startups
2019–2020 sees a lot of new companies growing from open-source projects, notable names are:
- Vercel (formerly ZEIT) successfully raised $20M to optimize JAMstack and enable frontend developers to do more.
- Strapi successfully raised $10M to commercialize its open-source offering.
Most open-source startups open-source their projects to gain loves and trusts from developers, then offer managed services.
Open-sourcing is a great way to achieve $0 acquisition, and also has a great impact on the company’s development growth and feedback. A lot of open-source customers tolerate bugs, then even help contribute and add new features to the product.
AI & Blockchain need more time
2 years ago is the hype of AI & Blockchain when a lot of startups founded just to give AI & Blockchain a shot. Now, despite the amazing progress of several companies like OpenAI, Scale, etc, it’s taking too long to see whether the technologies can create a sustainable business for founders to try while there’re other attractive fields.
Rethink of how users use their data & the rise of Notion
2019–2020 sees the rise of Notion taking down existing productivity giants (Evernote, Trello, etc).
People love Notion because it’s the one tool that everybody can use to solve such a wide array of problems in their own ways. It facilitates all productivity, creations, and sharing. But the secret behind its success lies in its elegant, flexible, and transparent data structures. Most of existing apps are backed by very opinionated and hidden data structures, therefore, limits what its users can do with their data. While Notion allows users to do direct manipulation and combination of different types of known blocks, which seems like what users want the most when they’re working on their data.
Thank you for reading this far! It’s what I found and wanted to share, hope it can help you gain some insight to develop your startup.
Just one last thing! I’m actively looking for a co-founder, if you’re interested, please contact me through my LinkedIn!