Before we get into my week, a quick ask. I'd like the next few months of this newsletter to be shaped by what you actually want to read, so I've added a one-click poll below to find out what you'd like to see more of.

It takes a couple of seconds, and there's space to leave a note after you vote if there's something you'd like that isn't on the list.

28 Wishes sells ice cream in Los Angeles. Below 70°F, sales fall about 20%. The weather is out of their hands. Rent isn't.

So the owners started putting about $20 a day into Kalshi weather markets, taking the cold side. The days that keep customers away now pay something back.

This is hedging. Big companies have done it for decades, buying protection against bad weather, fuel spikes and rising rates. It used to take a broker, a trading desk, and an order size no corner shop could meet.

Kalshi opens it up. Contracts on weather, fuel prices, inflation, tariffs and regulation, starting at a few dollars. Take a position on the outcome that would hurt you. If it hits, the payout softens it. If it doesn't, the contract expires and the good month was the point.

See how other owners are doing it at Kalshi.com/hedging

A lot changed for me this week, so this edition is part personal update and part lessons learned, from two and a half years of AI search in trading to what a VC in Dublin says investors look for.

On Thursday I stepped back from my day-to-day work at Pepperstone so I can put my time into growing Obsero. Pepperstone was the very first client I signed when I started Be Franc in April 2024, and AI search became a priority there almost straight away, because traders embraced it early and we could see ChatGPT turning up in the "where did you hear about us?" answers on our sign-up forms.

A few things I learned along the way:

Traders use more AI platforms than you'd think. ChatGPT made up the majority of what came through in attribution, but Claude was showing up in the reporting too, and what struck me was how quickly traders adapted to using a whole range of assistants to research the category.

They ask long, detailed questions. The team were early on this, so we ran user research into how our ideal customers actually prompt these chatbots, and it proved their questions about the sector are far longer and more detailed than anything they'd type into a search box.

The sources matter as much as the answers. Citations quickly showed that AI search is both a huge danger to the affiliates we work with and an opportunity for them to work more closely with brands. It also surfaced affiliates I had never heard of that were being cited heavily, which was probably the most useful thing the data showed us.

Risk warnings turned out to be a product and education issue. Every broker has to carry a risk warning depending on its jurisdiction, so compliance shapes everything that gets said about you. For a prompt like "Which CFD trading platforms are suitable for beginners?" I saw in Obsero data that ChatGPT was pulling the risk warning into its query fan-out as a site operator search, which I'd never seen before. ChatGPT is effectively linking a high risk warning with "not suitable for beginners", so the lower the percentage of retail accounts losing money in your risk warning, and the better you help traders understand the category, the more you can win in AI search. That simply isn't how it works in traditional search, and I found it fascinating.

My advice to any finance brand is to look at sources everywhere. Your own site's content matters hugely, especially for branded prompts when people are in the consideration phase, but you can't ignore how important it is to be mentioned in the key places off-site. Start with user research to understand which prompts your customers use and which assistant they favour, whether that's ChatGPT, Grok or Claude.

Once you have that prompt set, set up tracking (ideally in Obsero), find the top sources and check whether you're mentioned there. Anywhere you're already covered is a base to protect, and anywhere you're not is your outreach list.

I'm not disappearing, though. One of my consultants stays embedded in the team and I'll keep working with them at a more strategic level. A big thank you to everyone there for being so good to me, and the comments on my LinkedIn post meant a lot.

It's impossible to compete in the AI search space and run another business at the same time. Even Jack Dorsey was heavily criticised for running Twitter and Square together between 2015 and 2021, to the point where an activist investor pushed for him to go at Twitter. To be clear, Obsero is a very long way from Twitter or Square and I'm a long way from Jack Dorsey, which is exactly the point: if splitting your focus was a problem at that scale, it's a much bigger one for an early-stage startup.

The days are long and the nights are short. I'm up at 5am, down tools at 7am to get the kids ready for the day, then there's the day job, and I pick Obsero back up at night and at weekends. The context switching is just as tough, jumping from one piece of work to another and trying to give each one the attention it deserves.

My focus now is building product and accelerating it even further, helping to solve the real problems marketers are facing in AI search. We're not trying to be something we're not, and everything we build comes back to one thing: how you become the recommended brand in AI search.

On Tuesday I was in Dublin for Kimber's event at the Guinness Enterprise Centre, run with MCS Group and Genese Solution, where I sat on a panel hosted by Alexander Sleigh about building AI on AWS, covering both the investor and the founder side.

It was my first time back since 2012, which is a long gap for someone whose grandparents were born in Ireland and who spent long childhood summers in Kerry and Clare.

That last visit was for my 30th birthday, when Google+ was still in its groove, Facebook had just floated and I gave a disastrous rendition of Frank Sinatra's "My Way" in the Bunker Bar in Ballybunion that finished with a very slow clap and a lot of open mouths.

Amy Neale from Delta Partners gave the investor view on the panel and had the sharpest insights of the trip. The one that stuck with me most was integration. VCs want to know how deeply a product is built into a customer's workflow, because the more tangled up in day-to-day work it becomes, the harder it is to unpick and the more the customer comes to rely on it. She also looks at what a company is building into its core and whether the team has real experience in the market it's entering, with a tether to the real world or a regulated space, "not just for the hype of the current time".

That landed with me, because one question I've been asked by investors a few times is whether it's unusual to be a first-time founder in your 40s, when startups are usually seen as a young person's game. I'd been trying to be modest about it, saying I can be chief product officer for Obsero in the world of AI search but I couldn't be chief product officer for a clothing brand, and Amy's answer made me realise that deep knowledge of one category is exactly what investors are looking for. Good things come to those who wait, I guess.

When I was asked for my one piece of advice to anyone building with AI, I said to get going straight away, because there's never going to be a perfect product, so get selling as quickly as you can and learn from the conversations you have with clients.

Away from the panel, dinner at Dada (highly recommended if you're in town) turned into one of the most useful parts of the trip. I spoke to founders who run their own events, which they've found a great way to generate leads, and it's something I'm now exploring for Obsero. We also talked about how recruitment is changing to meet the demand for AI talent, with leaders from Zapier and Payoneer sharing their insights on hiring in the era of AI.

Thanks to Kimber for having me. Dublin was brilliant and I won't be leaving it another 14 years before I go back.

That is it for this Sunday, until next time.

Have a good one.

Andy