Beyond Finance

Why I built Beyond The Breakers. Two decades in open water swimming and I kept noticing the same gap. Coaches, race directors, marathon swimmers, open water swimmers, support crew — all deeply invested in the same sport, all operating in silos. There was no room where everyone sat down together and actually talked about it. So I built one. Beyond The Breakers started as a bet that this community deserved its own summit. Not an expo. Not a lap swim clinic. Something built specifically for the people who take open water seriously — in whatever role that looks like for them. Year 2 proved the bet right. People showed up from around the world. The conversations in the hallways were as valuable as anything on the stage. Year 3 is November 14–15, 2026 at Bentley University in Waltham, MA. If you're an open water swimmer, marathon swimmer, coach, triathlete, race director, support crew member, masters swimmer, collegiate swimmer or just starting out, there's a seat with your name on it. We built the room. Come fill it. #OpenWater #MarathonSwimming #Triathlon #EventFounder #SmallBusiness #BuildingInPublic #OpenWaterSwimming #EnduranceSports #BTB2026 #BeyondTheBreakers

Heading to the CCN conference next week. Hope to see you there. My office assistant, Bowie, will be handling things while I’m away. By that I mean, sunbathing and snoring.

Vibe coded a financial RAG system to see how far existing frameworks could take me. The idea was simple: a multi-agent workflow that takes any quarterly or annual report and turns it into something actually readable. Retrieves the data, extracts key metrics, runs analysis, surfaces risk signals — all through a Streamlit UI where you can watch every step of the pipeline in real time. The pipeline held, the UI made the workflow legible, and seeing it work end-to-end on real filings was genuinely exciting.But the rough edges were where it got interesting. Retrieval got noisy on dense filings. Multi-step memory degraded in ways I didn't anticipate. A few responses were more confident than they had any right to be. None of this was surprising for a fast prototype — but it made something very clear: Agentic structure makes a system look smart. It doesn't make it trustworthy. Key takeaways I'm carrying forward: → Answer quality matters more than architectural complexity → Each agent needs one job, a defined output, and an explicit failure mode → Memory and state management across multi-step pipelines is genuinely hard — plan for it early → Without evaluation criteria, you can't tell if you're improving or just moving things around → Current frameworks will get you to impressive fast — reliable is still on you The most valuable part wasn't what worked. It was what broke, and how quickly building fast surfaced it. Still building, still testing, still learning.

#KaliAnastasi #OrthodoxEaster #GreekEaster2026 #XristosAnesti #AlithosAnesti American Hellenic Council American-Hellenic Chamber of Commerce American Hellenic Institute Consulate General of Greece in New York Greek Orthodox Archdiocese of America Nathanael Symeonides Cal Alumni Club of New York Harvard Club of New York City Greek Orthodox Archdiocesan Cathedral of the Holy Trinity Annunciation Greek Orthodox Church Elpidophoros Lambriniadis Hellenic Heritage Foundation The Hellenic Initiative