We just claimed our .agent domain and joined Agent Community. I find it a great initiative to start shaping the future of autonomous agents . agentcommunity.org/join#SA07… @agentcommunity_

The future buyer on the internet isn’t a person, it’s an AI agent. Imagine tons of new customers flooding in, pockets full of spending power, yet they only discover, evaluate, and purchase through interoperability protocols. That’s not science fiction. That’s literally happening right now. If you're a business without access to agentic infrastructure and your data is not agent-readable, you’re invisible to the fastest-growing buyer segment the internet has ever seen. Think of it.

I just generated a world in real-time on Reactor, it was a really cool experience using world models! reactor.inc/share/478db1d3-0…

Travis is so right here. SMB B2B sales are brutal, and your LTV:CAC is one of the best tools for pressure-testing your business model viability.

Reflecting on another meaningful event and key milestone for @StyleAIofficial team. It was a great collaboration with French Connection (@FCUK) and really energizing to see real customers' response to our product in a live setting when they realize how much friction AI Agents can remove from the online shopping experience. Big thanks to everyone who joined us!

It's about exercising reframing

Most "AI in fashion" content happens online, behind a screen. Next Saturday, April 25th, we are doing the opposite. @StyleAIofficial will be showcasing the new SS26 collection from French Connection @FCUK using our AI Stylist agent in an Afternoon of Style. If you are curious about what retail looks like when human and AI styling sit side by side, come join us!

Retailers are finally seeing AI traffic convert, but I am starting to observe is a tension between the excitement of this new traffic and the reality of the merchant's backend. Most retailers want a game changing AI tool for their storefront, yet their core product data is too messy for an agent to parse reliably. We are moving toward a world where your brand DNA must be machine-readable, because an intelligence layer only works if the underlying catalog logic is sound. If you are seeing this shift in your own traffic, the priority is not just adding a new chatbot. It is ensuring your product's fashion metadata are rich enough to survive the scrutiny of a high-intent discovery engine. business.adobe.com/resources…

For retailers, getting ready for agentic commerce is first about reinforcing the systems that power commerce today, starting by inventories, brand identity, and following with product data and APIs. These strengths improve existing channels today while positioning merchants for new interaction models tomorrow.

here is one thought on the future of agentic commerce: the commerce storefront is not the endgame of online retail. AI agents are starting to shop on behalf of people. not “recommend me something.” actually go out, evaluate, and complete the transaction. when that happens, the storefront stops being the primary surface. the agent is the surface. and the agent needs something the storefront was never designed to give it: the brand’s actual taste logic. a flat catalog tells an agent what exists. it says nothing about what goes with what, who it’s for, or what the brand stands for aesthetically. that’s not a data quality problem. it’s structural. we built commerce infrastructure to serve human eyes, not agent reasoning. so when an agent asks “find me an outfit for a gallery opening that fits this brand’s identity,” a Shopify store can’t answer. a product feed can’t answer. a filter tree definitely can’t, and what answers it is a layer above the catalog that encodes aesthetic logic. queryable, structured, machine-readable taste. the infrastructure shift: from storefront-first to protocol-first. the question stops being “how do we get people to click?” and starts being “how do we make our brand legible to agents?” brands that figure this out early won’t just have better AI features. they’ll be natively discoverable in a world where agents do the browsing. the ones that don’t will be invisible by default.

Great discussions held yesterday at Prosus' Deeptech AI event. Couple takeaways: ▪️How to raise 100M : Investors line up and double down when traction comes along with exceptional Founder-Market-Fit, category-creation potential, and strong market tailwinds. ▪️EU opted for strong regulatory frameworks and protecting workers. US deregulated and made labour conditions more flexible, which, along with a historical risk-taking appetite continues to outsize every other market in fundraising opportunities. Shoutout to Jelle Prins, Sohrab Hosseini and Pieter Kemps for such honest and insightful contributions. Thank you! And of course some pics of a sold out event!

And here we go!! Great news for all those running locally 🙌🫶👏

The agent-to-agent version of every software is what's going to drive a significant part of the technological development in the coming years. At @StyleAIofficial we are building the vertical infrastructure that connects people's & fashion retailer's agents

On point. People are waling up to the fact that it is the cheapest hardware running the largest open source model that will be at the frontier of the AI race

We see in Vertical AI SaaS with domain-specific data (e.g. knowledge graphs) and infrastructure components a solid path to develop moat

This is a great perspective that technical founders should bear in mind when hiring a salesperson

Fully agree with @balajis here: the equilibrium really does point toward open-source models becoming the only ones people truly trust long-term. Centralized American labs are caught in an impossible multi-front squeeze (political pressure from both parties at home, plus the reality that distillation and open weights make "keeping the lead" increasingly futile). They'll make enormous money in the short-to-medium term, but trust, privacy, and local control win out eventually. Running your own model (or a verifiably neutral one) just feels different from relying on something that can be censored, backdoored, or politically pressured overnight. Decentralized/local AI isn't just an ideological preference, it's the practical path when institutions can't be trusted to stay neutral. Excited to be living these times! 🙌

Commoditization is not collapse. Rather, it is how technology lowers costs and expands access. The personal computer commoditized computing, the internet commoditized distribution, cloud commoditized infrastructure, and AI is commoditizing cognition. 👇