{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"We Built It Because We Had To - Tech Founder Backstories","title":"The AI Worker With Her Own LinkedIn Page (Daniele Bernardi)","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/eeec348e\"></iframe>","width":"100%","height":180,"duration":2888,"description":"One of Toolhouse's AI workers named itself, chose its own pronouns, and now runs a LinkedIn page its founder does not control. That was not in the prompt.\nDaniele Bernardi spent his early career as an engineer at Meta and later at Twitter, where he picked up the definition of trust he still runs the company on: consistency over time. In this episode he explains how that single idea shaped Toolhouse's entire go-to-market. The big AI labs want six figures of annual spend before a business gets real support, which leaves small and mid-market companies rebuilding everything themselves and quietly giving up on AI when the bill arrives. Toolhouse fills that gap by building narrow, single-purpose \"AI workers\" instead of generalist agents, so they do not drift and do not need a growing pile of context to stay useful. Daniele walks through the two-person wine club in San Diego that fired its marketing agency and now runs its whole weekly campaign through one worker, how Toolhouse earns access to systems like Gmail and Salesforce by sandboxing every decision before it touches a live account, and what it took to automate a European Central Bank stress test where the output has to match human work exactly across fifty consecutive runs.\nWatch the full episode on YouTube: https://www.youtube.com/watch?v=1Ye-s3Fe73Q\nCHAPTERS\n01:12 What Toolhouse actually does: turning one-off chats into repeatable AI workers\n02:02 Where the 90 percent cost saving comes from\n04:15 Why they are called workers, not agents\n06:16 Compliance and guardrails, from GDPR basics to enterprise policy\n07:40 From engineer to solutions engineer at Meta\n09:54 How Meta segments advertisers, and the gap that taught him everything\n12:27 Why the big AI labs cannot afford to serve you\n14:22 Finished work, not drafts, not notifications\n17:29 The San Diego wine club that fired its marketing agency\n20:41 When founders want AI to cut payroll instead of raise output\n24:26 Drift, and why you still need someone who knows...","thumbnail_url":"https://img.transistorcdn.com/qSYc3cBdFcsAT9NoNJB-b1WOXf_eKr0fZMdzan3rr64/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS8xNDI0/Y2Y5ZGEyNjlhMWQ4/OGRiNDdjNTc5YjQ3/YTQzOS5qcGc.webp","thumbnail_width":300,"thumbnail_height":300}