Right now every major dating platform is racing to add AI. Tinder has photo selection and a matching tool called Chemistry. Hinge launched AI conversation starters and a recommendation engine that drove a 15% increase in matches. Bumble rolled out AI photo feedback and profile guidance globally in February 2026.

Bumble announced Dates this week, powered by an AI assistant called Bee that learns your values, relationship goals, and communication style through private conversation before it ever proposes a match.

The goal, in Wolfe Herd's own words, is to build the world's smartest and most emotionally intelligent matchmaker. The news triggered Bumble stock to rally by 40% and gives us a glimpse as to the roadmap for Bumble and AI.

Already back in May 2024, Wolfe Herd floated a world where your AI concierge dates on your behalf, screens the entire city, and surfaces the three people worth your time. Industry professionals laughed, but she was simply too early, and the public just wasn’t ready to hear it.

Agent-to-agent dating is coming. When it does, how well you describe yourself to your agent will become extremely important. The agent's output is only as good as what you gave it to work with, and a basic level instruction will no doubt get basic results. A thoughtful input, for example one that’s very specific on a hard to articulate preference might surface that small wine bar in the back of your mind you walked past a hundred times and never tried as a first date location suggestion.

With the rise of agent-to-agent optimization, I can’t help but wonder what happens to the relationship that should never have worked on paper? The person whose attachment style is all wrong, whose photos are not your type, whose opening line was awkward, but who turned out to be the One? An algorithm optimizing on compatibility might filter that person out before you ever knew they existed. One day in the future we may even start to hear stories of some of the best relationships that were initially rejected by a screening AI agent…

Yelp announced a new data licensing deal with OpenAI in February 2026, positioning its 330 million human-written reviews as the trusted local data source AI systems need to answer recommendation questions. As search shifts toward AI, any platform trying to compete with Google needs high-quality local content, and Yelp has plenty of it. CEO Jeremy Stoppelman said this is not the last licensing deal Yelp expects to sign.

As AI surfaces multiply across devices and platforms, a business recommended by Siri may not be the same one ChatGPT surfaces. Yelp is betting that by being present across every major AI surface, it becomes the essential local data source no platform can afford to go without. Whether that translates into visible attribution inside AI-generated answers is still an open question. Nobody knows yet whether AI-driven local search consolidates around one platform, or creates something entirely new.

With its recent $270 million acquisition of Hatch, an AI lead management platform, Yelp is moving to solidify its position by owning the business operations layer, so that regardless of where a consumer finds a business, Yelp is still part of the transaction.

One thing I can’t help but wonder about: If you are a small business, you have no doubt heard about Yelp’s long-standing controversy over alleged pay-to-play tactics which are said to suppress good reviews to encourage paying businesses. Will AI be the one to finally outsmart the algorithm and democratize the review landscape naturally? I suppose it depends on how OpenAI and other partners weight and filter the data they license. I’m suspecting they’ll naturally filter credibility signals, and at that point the dynamic shifts entirely, which is a huge win for both consumers and businesses.

Zillow had a very loud February. It published a position paper declaring its AI advantage "structural" the kind of language companies use when talking to investors. They brought Zillow's home-buying guidance into Google NotebookLM, making it available inside one of Google's fastest-growing AI research tools.

Zillow's approach is a layering strategy. ChatGPT catches the buyer browsing. NotebookLM catches the buyer doing homework. Zillow is building the foundation to eventually add in agents, and they are actively hiring for an Agentic AI team right now.

To understand what this missing agentic layer could look like, I want to re-introduce you to a company I met at CES in January: obo-world, co- founded by Sonia Reis Masclez. Obo-world is a French company building virtual recruiting agents for HR teams and beyond. They are also working in Real Estate. Obo-world is building agents that can schedule tours, handle qualification calls, document follow-up, but what’s special is on the buyer’s side, which can also be assigned an agent that would essentially stay with them for every real estate transaction across their lifetime. It is an ambitious plan, and if done right will help many people buy and sell their next homes.

Zillow is hosting an AI Investor Summit, ‘Leading the Next Era of Real Estate’ on March 24th in New York City and is poised to shed light on the rest of Zillow’s agentic strategy.

Visa's Intelligent Commerce platform has over 100 partners building on top of it. As AI agents begin acting on behalf of humans, they need a way to pay. Visa is building that infrastructure, and real transactions are already running.

The mechanism works like this: you authorize an agent with a defined spend limit tied to your real card. The agent requests a virtual card number for a specific task, executes the purchase, and reports back. No checkout page. No form. Visa's Trusted Agent Protocol adds cryptographic signatures to verify that legitimate agents are acting on behalf of actual humans.

Just last week, Santander and Visa completed Latin America's first live agentic transactions across five markets, including Brazil, Mexico, and Argentina. AI agents purchased chocolates in Brazil and books across the other markets. Today, Visa launched Agentic Ready, a new program rolling out across Europe with Santander, HSBC, Revolut, and Nationwide, designed to get issuers ready for agent-initiated payments at scale.

Visa is not the only agent payment option. MoonPay launched MoonPay Agents in February 2026, assigning AI agents their own wallets and enabling them to transact autonomously on blockchain rails. Unlike card-based systems, this gives agents a decentralized path to payment, including support for x402, the machine-to-machine payments standard built for agent-driven applications.

* For those building on OpenClaw: Visa just partnered with Crossmint to launch Lobster.cash, a virtual card layer built specifically for OpenClaw agents. Worth knowing if you are already in that ecosystem.

Since we’re talking about visibility, I recommend typing this into Claude or OpenAI or your favorite AI platform:

“Act as an AI shopping agent representing a customer.
Analyze my business and tell me:

1. Whether an AI agent could easily understand our products
2. Whether our purchasing process could be automated
3. What information is missing for an AI agent to make decisions about our products?
4. How can our business become more agent-friendly?

[paste your website or product page URL here]”

Paste your website or product page and see what comes back. You may discover a completely new way to think about your business and begin a path to your AI strategy.

You have probably experimented with AI tools and given permission to access your calendar, email, or files. It’s great practice to check which of these tools still has access.

The easiest way to see all the AI tools or platforms you've experimented with that still have connections to your accounts can be found in a couple of places:

Take 5 minutes this week and go to your Google or Microsoft account settings, open "Connected Apps," and revoke anything you aren’t using. An agent with stale access is an open door. Close it for safety- you can always re-connect once you need it.

A quick personal note on what I’m building.

I’ve been quietly upgrading the architecture behind a start-up I’m working on, it’s a fashion tech startup that’s a monster of a project. I’m excited that it has grown the way it has, something I’d never be able to do had it not been for an agentic team. I’ve been working on this for the past few months, and now it’s time to move it out of early prototype tools and into a more serious development environment. This is also a great case study for anyone dreaming of their next startup, and a testament that it IS possible to advance an idea with one person using an agentic team. What a time to be alive.

I’m excited to tell you more soon.