I'm Abel Samet, co-founder of Troubadour Goods. I spend my time building practical AI — dashboards, agents, and internal tools that a whole company actually relies on, not demos.
I co-founded Troubadour in 2011 and have spent the last fifteen years building it — from a two-person startup into a premium brand with a team of ~25 across the UK and US, customers in more than 150 countries, and 15x revenue growth over the last five years. Along the way we earned four consecutive Wirecutter "best laptop backpack" wins, certified as a B Corp, acquired the Dutch tech accessories brand Mujjo, and built a multi-channel engine spanning DTC, Amazon, wholesale (Harrods, Nordstrom, Bloomingdale's), corporate sales, and a flagship London store.
Numbers were my first language. I started out in M&A at Lazard and private equity at AnaCap, then took an MBA at Harvard Business School, where I graduated a Baker Scholar — all while Troubadour was getting off the ground. That training never stopped being useful: I raised £8.6M of growth capital, structured the Mujjo acquisition so it financed itself through an earn-out, and served a two-year stint as our CFO. HBS now teaches a case on Troubadour, and I go back most years to guest-lecture it.
More recently I found a second craft. At a company with no engineering team, I started building the software I wished we had — first a fleet of agents posting daily sales reports to Slack, then a live revenue dashboard the whole company runs on, then inventory tooling that frames six-figure purchasing decisions. Everything I build gets held to an operator's standard: reconciled against the accountants, trusted by the CFO, and used every single day. I'm always interested in comparing notes with other brands and operators putting AI to work the same way.
I live in the Boston area with my wife and our two boys, and most weekends involve a soccer ball — strictly for fun. I like being outdoors and exploring national parks, and in an earlier chapter I ran my share of marathons and triathlons, including an Ironman. Most mornings you'll still find me on the deck with a coffee while my agents post the day's numbers.
The point of AI is being better informed: reliable, consistent systems that put real-time, correct data in front of you — so decisions get made faster, with confidence in what's underneath them.
A dashboard nobody has reconciled is a rumor. I audit my tools against the accounting system, line by line, before anyone is asked to rely on them.
I build tools I have to live with. When you own the P&L, you don't ship things that waste your team's time — you ship things that give it back.
A live revenue and NPS dashboard covering seven sales channels across two brands, with every currency normalized and pacing tracked against plan. Alongside it, a family of scheduled agents posts daily reports to Slack — yesterday vs. target, month-to-date vs. pace and last year, partner-level breakdowns, and a month-end run-rate projection — for e-commerce, Amazon (including ad performance), marketplaces, paid acquisition, and the London store.
Before shipping it company-wide, I reconciled it against the accounting system with our outsourced finance team — chasing down FX treatment, duplicate orders, and edge cases until the numbers tied out.
An internal decision-making tool managing 113 SKUs across warehouses in New Jersey, Amazon FBA, and beyond. Purchase-decision queues, stock rebalancing between locations, end-of-life management, and factory payment tracking — built to give leadership a clear frame for the holiday production buy. The interface came quickly; the real lift was the data backbone underneath — BigQuery and MCP connections that keep every number real-time and reliable.
BigQuery + MCP data backbone · Lovable front end
A first-principles analysis of Google and Meta advertising that separated genuinely incremental revenue from what the platforms merely claimed credit for. The findings reshaped how the company measures attribution and allocates media spend.
Google & Meta data · Attribution modeling · Presented to the board & leadership team
Led the build of a BigQuery warehouse unifying customer and sales data across Troubadour and Mujjo — moving the company off scattered per-tool reporting and onto one queryable foundation that the dashboards, agents, and forecasts all draw from.
BigQuery · Shopify, Amazon & marketplace pipelines
A savings app for kids and their parents: real ledgers, parent-set interest rates, automated interest postings, and savings goals that teach the magic of compounding. The first version came together in a single evening as a prototype; the app now running is a ground-up build with Claude Code.
Claude Code · React · Supabase · Cloudflare Workers · Live at opifera.co
A daily briefing agent that reads my email, Slack, Asana, HubSpot, and calendar overnight and delivers one prioritized digest: what needs me first, what's coming in the next 48 hours, and what can wait. It ran every day for months — including straight through my vacation.
Claude agents · Gmail, Slack, Asana, HubSpot & calendar connectors
Co-led the AI build session at our company-wide all-hands, teaching a team with no engineers to build their own assistant agents. Ongoing internal champion for AI adoption — from evaluating vendors to sharing what works in a dedicated wins channel.
Claude Cowork · Connectors · Live workshop for the full company