The Future of AI Agents Just Accelerated (Again)

What. A. Week.

It’s hard to remember the last time I've been this excited about multiple developments happening simultaneously in the AI and agent space.

In today's newsletter, I'm breaking down three massive developments that happened this week. Each one individually would be newsletter-worthy, but having them all happen in the same week is something special:

• agent.ai hitting a major milestone (500,000+ users!)

• Our integration with OpenAI's new o3-mini model

• A mind-bending experience with OpenAI's new Deep Research feature

—Dharmesh

The Power of 500,000

First, some incredible news: agent.ai just crossed 500,000 users. That's 10x growth in just six months.

I want to express my deep gratitude to everyone who's been part of this journey:

To the 13,000+ of you who've taken the time to rate and review agents — you're making it possible for others to discover the best tools for their needs. Every review helps build a more valuable ecosystem for everyone.

To the 5,000+ builders who've used our low-code platform to create agents — you're the innovators making this platform truly useful.

And to everyone on social media who's helped spread the word — your support is helping us build something bigger than we imagined.

But what makes this growth most exciting to me is that we're seeing a virtuous loop forming:

• More users attract more builders.

• More builders create more useful agents.

• More useful agents bring in more users.

It’s not just about hitting a number — it's about reaching the critical mass needed to make AI agents truly useful for everyone. Every new user, every new builder, every new agent adds to the network's value for everyone else.

A Cheaper, Smarter Model for Everyone

The first big news from OpenAI this week: they released o3-mini, their new cost-efficient reasoning model. And we've already integrated it into agent.ai.

If you haven't heard about o3-mini yet, here's a quick rundown:

• It's OpenAI's first reasoning model available to free users on ChatGPT

• It matches or beats o1's performance in technical tasks

• It's 24% faster and 63% cheaper to run

• It comes with adjustable 'reasoning effort' settings (low, medium, high)

That said, what really matters isn't just the specs — it's what this means for builders. Starting today, anyone can build agents using this powerful new model on agent.ai.

Here's what we're working on:

1. Building Goal-Oriented Agents: Instead of needing to explicitly define every step, we're exploring ways for agents to plan and execute tasks based on high-level goals. Think of it as moving from "get this done by doing this and then that and then this other thing" to "here's what I need accomplished, figure it out"

2. Creating Network-Aware Agents: The ultimate goal is developing agents that can understand and work with hundreds of other agents in our network. Imagine an agent that knows exactly which specialists to bring in from our ecosystem to get things done.

It's like LEGOs for the LLM world. Every agent is another piece that can be used to build something bigger.

Mindblown moment: OpenAI's DeepResearch

The most mind-boggling announcement from OpenAI this week: Deep Research.

If you have access to OpenAI's Pro tier ($200/month), you'll be able to try this now in ChatGPT. And it’s worth trying out for anyone who loves going down research rabbit holes like I do.

First, some primer: I've spent 30+ years in the CRM software industry. Not just watching from the sidelines — I've been on the court, playing the game. From my first vertical CRM startup to co-founding HubSpot, I thought I had a pretty good handle on things.

Then, I tried Deep Research.

I asked it to create a detailed research report on the CRM industry — covering everything from competitive analysis to product strategy and AI vision. What it produced was an 11,000-word masterpiece.

Data… Citations… Tables… And genuinely great insights that made me think differently about an industry I've lived and breathed for three decades.

But here's what really has me excited: This isn't just about generating impressive reports. It's about what comes next.

These comprehensive outputs can become inputs for other agents in a workflow. Because the future is about agent composability — being able to pull together pieces and put them together into a larger whole.

This is exactly what I've been dreaming about for years. We're not just seeing glimpses of the future anymore — we're watching it unfold in real-time.

DISCLOSURE: I'm a small investor in OpenAI (and also a big fan).

Agent Builder of the Week

Before I conclude this post, I’m trying something new.

Each week, I’m going to be highlighting someone from our community who's building something interesting on agent.ai, for other’s to draw inspiration from. This week's feature:

Why I love this: In just one week, a Shaalin created something that helped thousands of people. They launched in a day, reached 4,000 users, and maintained a 4.48 rating. That's the magical moment for builders — when they put something out there and other people start seeing real value from it.

Want to be featured? Share what you're building and tag Agent.ai on LinkedIn so I can check it out :)

What a Time to Be Building

Massive developments, 500,000+ users, and builders creating agents that help thousands of people in just days.

This is what excites me most about agent.ai — watching people take these powerful new tools and create things that help others in ways we never imagined.

The future is unfolding in front of our eyes in real-time. And it's all happening faster than even I expected.

As always, excited to hear your thoughts.

—Dharmesh (@dharmesh)