Getting started with AI on a budget: Tools you can use and what’s coming next

Tips for developers exploring code assistants, function calling, and open source AI tools.

Fran Hinkelmann, engineering manager at Google, sat down with the All Things Open team to share her perspective on the fast-moving world of AI and the tools she’s excited about right now. She sees competition and constant change as positive forces, pushing better products and new possibilities. For Fran, the unknown isn’t a risk, it’s an opportunity to explore what’s next.

That excitement also comes with challenges. Fran highlights open questions around security and ethics, and stresses the importance of continued work in those areas. For developers new to AI, she recommends jumping in with hands-on learning. Whether you’re using PyTorch to build a model, entering a Kaggle competition, or exploring open source Jupyter notebooks, there are low-cost, accessible ways to begin.

Read more: How to build an AI agent in 10 lines of code

One of Fran’s favorite AI moments came from function calling in large language models (LLMs). This feature allows LLMs to go beyond conversation and actually trigger real actions, such as calling APIs or interacting with devices. She describes it as the moment where AI becomes truly useful. Fran also encourages developers to regularly test different code assistants. Tools like Gemini and others are evolving quickly, and she’s found that trying out new versions can lead to big productivity gains.

Key takeaways

• You don’t need a credit card to start learning AI and open source tools and notebooks make it easy to explore.
• Function calling lets LLMs move from chat to action by connecting to APIs and executing real tasks.
• Code assistants are improving fast, try different ones and re-evaluate them often to find what works best for you.

Conclusion

Fran’s approach to AI is grounded in curiosity and a willingness to experiment. From working with low-level math to high-level tools, she shows that there’s no single way to start learning AI. The key is to dive in, stay flexible, and take advantage of the open source tools and communities that are making AI more accessible every day.

More from We Love Open Source

• How to build an AI agent in 10 lines of code
• How AI will change frontend development in 2025
• The missing metric in open source: Who’s using your code?
• 6 best practices in software architecture
• Why AI won’t replace developers