History Often Rhymes : Phillip Merrick, CEO and CoFounder of pgEdge

"I spent several decades perfecting my craft as a computer programmer. Guess what? Nobody is writing code anymore. At least they shouldn't be."Joe sits down with Phillip Merrick — the founder who took webMethods public at the height of the dot-com boom, co-founded EnterpriseDB to bring Postgres into the enterprise, and now leads pgEdge — for a conversation about what today's AI moment looks like to someone who lived through the last one.Phillip walks through the parallels he sees with the dot-com bubble (revenue round-tripping, the joint ventures, the concentration of capital) and where this time genuinely differs. He shares what happened inside pgEdge when agentic engineering took hold: engineers doubling, trebling and quadrupling their productivity, three new products shipped in nine months — and then the rest of the company struggling to keep up. He calls it a technology digestion problem, and it's the most useful warning in the episode for anyone about to turn this on.The conversation closes on build vs. buy, and Phillip's advice for credit unions is direct: if you haven't historically been an organization that builds software and runs it, maintains it and secures it, don't assume a coding agent changes that.⏱ CHAPTERS00:00 — Introducing Phillip Merrick03:37 — The webMethods IPO04:31 — Three citizenships, every dialect of English05:07 — Dot-com bubble vs. AI: the parallels07:42 — "Nobody is writing code anymore"08:31 — History rhymes: round-tripping and over-provisioning09:16 — Are we over-provisioning compute this time?11:15 — Agentic engineering at pgEdge: the productivity jump12:21 — Three products in nine months13:14 — The technology digestion problem14:30 — When speed moves risk onto the critical path15:36 — Don't build one monolithic agent16:28 — The error-rate math on chained agents18:08 — Toward a composable agentic reference architecture19:12 — Build vs. buy for credit unions20:12 — "Don't kid yourself"21:35 — Prototyping with Lovable and Replit22:37 — Why credit unions should own agentic banking24:24 — Closing thoughtsKEY TAKEAWAYSThe capital going into AI is several orders of magnitude beyond the dot-com boom — but unlike then, there's real revenue behind itAgentic coding tools crossed a threshold in 2025: the output still needs review, but it no longer needs reworkWhen engineering output goes up and to the right, operations, enablement and risk have to move with itMonolithic agents create attack surface, are terribly hard to debug, and degrade as you ask more of themChain enough 97%-accurate agents and the multiplication catches up with you — decompose into specialistsPrototyping platforms can close the old gap between what business users want and what IT buildsCredit unions know banking and member service — that's the ground to win agentic banking onABOUT THE GUESTPhillip Merrick co-founded webMethods, where he pioneered web services and web APIs and led the company through one of the most successful software IPOs of its era. He went on to co-found EnterpriseDB, one of the first companies to make Postgres viable as a fully open-source enterprise database, and later led turnarounds at SparkPost and Fugue. He co-founded pgEdge in 2022, now serving major U.S. government agencies, the German healthcare system, Thales and a major Western European payment system.