Welcome to this week's Tip Sheet on APIs and AI. This week I'm sharing new articles on AI agents by some heavy hitters in tech at Google, Anthropic, and Microsoft. Agents will be critical for AI to effectively use APIs, which is something all the major AI companies are pushing for, but is still a work in progress. There's a signup for a free course on AI agents from HuggingFace. I also have a couple of tech book recommendations you might be interested in. Let's get started!

The Latest Thinking on Agents (and APIs)

Many are calling 2025 the "Year of AI Agents". There is a ton of work being done by LLM providers and everyone else in the industry to figure out how to make autonomous AI agents ready for prime time.

Google recently released a whitepaper about their view of Agents. Here's a key quote that API providers should pay attention to:

As you read about AI agents, if you are an API developer or provider, focus on TOOLs. That's where the action is at for the API-focused readers. APIs are the tools that Agents use to interact with the real world.

Anthropic is one of the top LLM providers including the popular Claude model, and has written their own guide on Agents. Their article links to their earlier tool usage article.

Valentina Alto from Microsoft wrote a good blog post that gives a big-picture view of this interaction: AI Agents — A Practical Implementation.

Free HuggingFace Course on Building Agents

HuggingFace is planning a free course for building agents that will start in February. If you're not familiar with the company, they were responsible for the pre-LLM boom in Transformer models through their Python library. They run a major repository of machine learning models, among other work. Their blog post says "This interactive, certified course will guide you through building and deploying your own AI agents." They say they'll demonstrate LangChain, LlamaIndex, and smolagents (a HuggingFace creation), so it should be a good chance to see different agent frameworks side by side. Full details and signup form here.

New Technical Books Worth Reading

This week I got my copy of Steve Wilson's The Developer's Playbook for Large Language Model Security. Steve led the creation of the OWASP Top 10 for LLMs and has a lot of tips to share about security vulnerabilities. It's a fun read -- the best part so far has been a chapter that uses Independence Day and 2001: A Space Odyssey to demonstrate LLM vulnerabilities.

I also recommend sharing your learning in the open by creating side projects, putting them in public GitHub repos, and posting about them. There's a new book called Writing for Developers by Piotr Sarna and Cynthia Dunlop that tells how to find ideas for technical blogs and tips on writing them. The biggest section in the book lists patterns for useful blog posts, which is something that never occurred to me. They list seven patterns including "How We Built It", "Rewrote it in X", and others. If you're looking for inspiration to get your work out in the world it's a good one.

Both of these books (and thousands of other O'Reilly, Manning, No Starch Press, and others) are available online if you sign up for the O'Reilly Platform. If you'd like to get a 30-day trial of the platform, you can use my coupon code here: https://learning.oreilly.com/get-learning/?code=HOAPI24.

That's it for this week!

Keep Coding,

Ryan Day

https://handsonapibook.com/

​