Ada Ventures is pleased to announce its investment in JigsawStack, bringing JigsawStack’s total raised capital to $1.5 million, including a previous $500,000 investment from Antler and nine months since its beta launch.

The cost of developing software has plummeted over the last two decades. This is thanks to open source, cloud, and frameworks that abstract away the complexity of writing code.

Building a full-stack web app in the 1990s could easily cost millions of dollars and demand a team of dozens of engineers. Today, it’s at least 1,000x cheaper, and you can build multi-billion-dollar businesses with a handful of engineers.  

Because software development has become more accessible, today, there are over 20 million developers globally versus a few hundred thousand in the 1970s. With generative AI, we expect another step-change in the cost of building software. It is going to get massively cheaper, and more people are going to build software.  

We are already seeing this with the success of products like GitHub Copilot, Cursor, Loveable and others. In the next decade, we could have up to 1 billion people building custom software, roughly the same number of people who know how to use Excel today. That opens up a ton of opportunities for economic empowerment for everyone, not just the technically inclined.

However, builders today face several technical barriers. When we built AdaGPT, for instance, we had to navigate a maze of cloud and database products across various providers. This patchwork approach makes it harder to build software. Consider someone who wants to create an app with multiple genAI features. They would need to manage all the following:

‍

• Fragmented integrations: AI‑powered apps typically require multiple APIs for tools such as web search, data extraction, text processing, image generation, translation, and more. Developers must juggle disparate services—each with its own authentication, rate limits, and response structures. This increases the surface area for bugs and makes debugging a nightmare.

• Manual orchestration and vendor management overhead: Even with code‑generation tools that help users write code, just calling an API isn’t enough. A functional AI app often requires chaining multiple tasks—e.g. scraping content, structuring raw data, analysing sentiment, summarising text, feeding it into an LLM, transforming outputs into speech or images, and storing results. Doing this manually and with separate vendors introduces complexity and operational overhead.

• Scaling and security: Finally, handling high‑volume AI workloads is nontrivial. Speech‑to‑text, text‑to‑image, and real‑time search tasks require infrastructure that can handle spikes in demand while ensuring low latency. Additionally, securely managing API keys, handling user data privacy, and maintaining compliance across multiple AI services is a burden most developers don’t want to deal with.  

‍

These reflections underpin our conviction and investment in JigsawStack, our latest AI investment that directly addresses these challenges for anyone building an AI-powered app.

What Does JigsawStack Do?

JigsawStack helps developers, business owners and teams build and ship AI products faster. The platform provides a suite of small, fine-tuned custom AI models for everyday AI tasks.

For example, rather than juggling infrastructure for web search, AI scraping, text summarisation, translation, image generation, or sentiment analysis, JigsawStack consolidates everything into one platform with easy-to-use SDKs or APIs.  

Here’s a brief example of a feature commonly found in many AI applications: the ability to read web pages. Below, I’ve included code snippets from a script that reads job web pages. The first example uses JigsawStack (just 3 lines of code), while the second uses an open-source library called BeautifulSoup to select components from a webpage manually (8 lines of code!). It’s clear from these two which one is more accessible and quicker to deploy.

‍

JigsawStack is not only focused on ease of use. They are also building for speed. For example, their speech-to-text API is the fastest in the market. Here are benchmarks comparing JigsawStack to companies like Groq, AssemblyAI and OpenAI.

The team are also making AI integrations on the backend seamless. Here’s a video demo from JigsawStack’s founder, Yoeven, showcasing how you can directly integrate translation into a Supabase database.

Although it's in its early days, we can see how JigsawStack will ultimately make it easier for anyone building software to embed multiple genAI services into their product.  

Why does JigsawStack fit the Ada thesis, and what got us excited?

We invested because of two key factors: an exceptional founder and a strong thesis fit with a favourable macro-outlook.

I first met Yoeven in person at an event, and when I asked for a demo, instead of shying away from showing a product that was in alpha, he urged me to get my laptop out to install their SDK and fire up a few API requests right away. Very few founders do this, so I immediately knew we had to dig in to learn more.

‍

‍

Yoeven is also a second-time entrepreneur. He launched Southeast Asia’s fastest-growing prop-tech startup, which grew from $1 million in ARR within the first year to $3 million in the second year before it was acquired. This is a founder who moves with and appreciates speed. Since launching an early version of their platform last summer, JigsawStack have already processed 10 million API requests.

The second reason we invested is that JigsawStack fits our economic empowerment theme. We believe that if more people have access to tools that make it easier to embed AI in software products, the economic value that accrues from AI will spread from the few to the many. Whether that’s a solopreneur building a side project, a small business wishing to build custom AI software, or larger businesses looking to empower their employees with tools that make them more productive, there’s lots to gain when we make it easier for everyone to build with AI.

Looking forward

The world is only beginning to grasp AI's true potential. JigsawStack— positioned at the applied AI infrastructure layer—offers a vibrant spectrum of opportunities for people who are new to code and seasoned developers alike. We are thrilled to be a partner with Yoeven and the JigsawStack team and look forward to supporting their vision: simplifying backend AI integration and empowering anyone to build AI-powered apps without technical overhead.

“Almost anyone will soon be able to build software with AI, yet the backend plumbing still frustrates even the best developers. I’ll never forget how Yoeven, full of energy, got me to grab my laptop, install JigsawStack, and run a rapid live demo. Within minutes, I was sending API requests that could seamlessly integrate multiple AI services—web search, translation, speech-to-text, images, and OCR—without the usual overhead. With more than 100x as many developers today compared to 50 years ago, and with these numbers set to grow even more with AI-code generation, the potential to simplify backend software engineering could not be any more urgent. And so we’re excited to back a visionary founder here who gets what it takes to simplify developer workflows by making small, but powerful AI models accessible to all."

Michael Tefula, Principal and Head of Product at Ada Ventures