Autoresearch: Let AI Agents Run the Experiments You Never Had Time For

Autoresearch: Let AI Agents Run the Experiments You Never Had Time For

Can AI agents do more than write code? David Cortés Fulla, Staff Engineer at Shopify, shows how pi-autoresearch lets agents run autonomous engineering experiments against any measurable metric. The talk follows a real developer pain: slow CI and long feedback loops. Instead of manually trying one-off fixes, autoresearch gives an AI agent a metric — like build time, installspeed, test performance, bundle size, prompts, or skills — and lets it iterate: propose changes, run checks, keep improvements, reject failures, and repeat. David explains how this approach helped Shopify speed up backend CI, generate multiple performance improvements, and automate the kind ofvaluable but boring optimization work thatengineers rarely prioritize over feature development. Agents used to code like us, but faster. Autoresearch lets them run the experiments we never had time for. 👉 Subscribe to Itnig for more conversations about real business, startups and brands. 🎙️ Want to join the Itnig podcast or sponsor one of our episodes? Appear on the podcast: https://tally.so/r/wo1Poe Sponsor the podcast: https://tally.so/r/3EERLN ABOUT ITNIG: 🐦 X - ⁠https://x.com/itnig 💡 LinkedIn -⁠ https://linkedin.com/company/itnig 📸 Instagram - ⁠https://instagram.com/itnig 💌 Newsletter - ⁠https://itnig.net/newsletter/⁠ 🌐 Web -⁠ https://itnig.net/ LISTEN TO OUR PODCAST ON: 🔊 Spotify: ⁠http://bit.ly/itnigspotify⁠ 🎙️ Apple Podcast: ⁠http://bit.ly/itnigapple ⁠00:00:00 Intro & welcome 00:00:05 David introduces himself & his work at Shopify 00:01:16 The problem: slow feedback loops in development 00:04:27 First experiment with pi — tackling build times 00:08:00 First results — 50% faster builds with parallel workers 00:09:24 Live demo: real project walkthrough 00:13:00 How pi researches, iterates and self-improves 00:16:04 Storing ideas for future iterations across model versions 00:18:30 The results go viral: trending on X, 1,000+ stars, 160+ forks 00:20:15 50+ successful internal experiments at Shopify 00:21:11 Beyond speed: running autonomous feature experiments in parallel 00:22:32 Q&A

From Magic to Harness: Building an AI Agent for a 25-Product SaaS Company

What happens when you put an LLM on top of a 25-product SaaS company?   Ilya Zayats, CTO at Factorial, shares the story of building Factorial ONE — an AI agent designed to help users work across HR, payroll, time, talent, finance, IT, documents, permissions, and many other product surfaces.   The first approach looked simple: let the model query the product directly. But the product graph was too large, the GraphQL schema did not fit in context, and the model struggled with product entities, permissions, and stable answers.   The talk follows the evolution from “LLM magic” to a production agent harness: semantic tools, skills, smaller schemas, runtime context, deterministic computation, permissions, and browser-based actions.   The main lesson: useful AI agents are not built by giving the model the whole world. They are built by designing the right harness — context, tools, workflows, permissions, and runtime — so the agent can safely understand the product, take actions, and create new work across it.   Recorded at AI Builders Barcelona.   Speaker: Ilya Zayats, CTO at Factorial   Topics: AI agents, Factorial ONE, SaaS, LLMs, context engineering, tool use, skills, GraphQL, product automation, AI platform, agent harness, enterprise AI

MCP Isn’t Always the Answer: 7 Experiments Building AI Agent Interfaces at PostHog

Georgiy Tarasov, AI Product Engineer at PostHog, shares what PostHog learned from seven experiments building interfaces for AI agents — and why MCP isn’t always the right answer. As customers started asking for “all PostHog AI features in my agent,” PostHog had to rethink how its product should work outside the browser: inside Claude Code, Cursor, Codex, CLIs, MCP clients, and local agent workflows. In this talk, Georgiy walks through PostHog’s experiments with handwritten agent tools, traditional MCP, CLI-first interfaces, MCP with a CLI-like shape, code execution, dynamic toolsets, and SQL-based retrieval. He compares the trade-offs across developer experience, context bloat, tool discovery, token efficiency, latency, eval results, and real customer usage. The core lesson: your new user might not be a human in a browser tab — it might be your customer’s AI agent. 👉 Subscribe to Itnig for more conversations about real business, startups and brands. 🎙️ Want to join the Itnig podcast or sponsor one of our episodes? Appear on the podcast: https://tally.so/r/wo1Poe Sponsor the podcast: https://tally.so/r/3EERLN ABOUT ITNIG: 🐦 X - ⁠https://x.com/itnig 💡 LinkedIn -⁠ https://linkedin.com/company/itnig 📸 Instagram - ⁠https://instagram.com/itnig 💌 Newsletter - ⁠https://itnig.net/newsletter/⁠ 🌐 Web -⁠ https://itnig.net/ LISTEN TO OUR PODCAST ON: 🔊 Spotify: ⁠http://bit.ly/itnigspotify⁠ 🎙️ Apple Podcast: ⁠http://bit.ly/itnigapple ⁠00:00:00 Intro & welcome — AI Builders BCN first edition 00:01:04 Georgiy introduces himself & PostHog 00:02:14 The challenge: shipping AI agents on a complex product 00:04:37 First MCP experiments & why they didn't scale 00:05:26 Token optimization problems & early lessons 00:07:08 Rethinking the approach: sandboxing & unified interfaces 00:08:13 Deep dive: how MCP prompts & instructions really work 00:10:26 Optimizing MCP tools for different model providers 00:12:55 Experiment #1 — grouping tools by intent00:14:15 Switching to a resource-based approach 00:18:24 Advantages & limitations of the code execution approach 00:22:30 MCP with native code execution — Georgiy's favourite 00:25:15 How GitHub optimizes their MCP (grouping by intent) 00:25:43 Agent skills: what they are and why they matter 00:26:43 Building skills with regex, access controls & data 00:27:52 Benchmarking: testing with real vs synthetic data 00:29:38 Results & what actually worked in production 00:31:35 Key takeaways & closing thoughts 00:31:43 Q&A 00:36:29 End Recorded at AI Builders Barcelona. Speaker: Georgiy Tarasov, AI Product Engineer at PostHog Topics: MCP, AI agents, Claude Code, Cursor, Codex, CLI, codegen, agent interfaces, developer tools, PostHog, AI engineering

Bitcoin, Tecnología, regulación y dinero: una tertulia con Leif Ferreira

Esta tertulia es posible gracias a Imperia SCM. Las herramientas tradicionales de supply chain limitan tu planificación.SCP Studio rompe ese modelo con una plataforma de Supply Chain Planning con IA nativa que permite adaptar el software a tu negocio.Mejora tus decisiones alineando demanda, producción y compras en un solo entorno. Conoce más sobre SCP Studio: https://imperiascm.com/ En esta tertulia en directo desde Lanzadera, en Valencia, charlamos con Leif Ferreira, CEO de Bit2Me, sobre cómo se construye una empresa tecnológica desde cero, los aprendizajes de emprender durante años sin apenas recursos y el reto de escalar en un sector complejo, regulado y en constante cambio. Hablamos de la comunidad tecnológica en Valencia, de los inicios de Leif como desarrollador, de digitalización, banca, regulación, financiación, seguridad, inteligencia artificial, geopolítica, innovación en Europa, oportunidades para nuevos emprendedores y cómo detectar tendencias antes de que sean evidentes para todos. También entramos en el crecimiento de Bit2Me, la relación con bancos e instituciones, la tokenización de activos, las stablecoins, la custodia de fondos, los riesgos del sector cripto, el caso FTX y el futuro de los productos financieros digitales. Una tertulia sobre startups, tecnología, dinero, regulación y nuevas formas de construir en un mundo que cambia cada vez más rápido. Sigue a los "tertulianos" en Twitter: • Bernat Farrero: @bernatfarrero • Jordi Romero: @jordiromero • Marcel Queralt: https://www.linkedin.com/in/marcelqueralt/ SOBRE ITNIG 🐦 Twitter - https://twitter.com/itnig 💡 LinkedIn - https://es.linkedin.com/company/itnig 📸 Instagram - https://www.instagram.com/itnig/ 💌 Newsletter - https://itnig.net/newsletter/ 🌐 Web - https://itnig.net/ ESCUCHA NUESTRO PODCAST EN 🔊 Spotify: http://bit.ly/itnigspotify 🎙️ Apple Podcast: http://bit.ly/itnigapple

La startup que DETECTA ENFERMEDADES antes de que ocurran

En este episodio visitamos el recinto modernista de Sant Pau en Barcelona para conocer Base4, nuestra última participada de itnig, y entender cómo están usando datos, biología molecular y tecnología para cambiar la forma en la que entendemos la salud humana. Hablamos con Pol Cervera, Dr. JM Soria y Ángel Martinez sobre su modelo fundacional de biología humana, sus relojes biológicos por tejidos y órganos, y cómo una simple muestra de sangre puede ayudar a detectar anomalías antes de que se conviertan en problemas clínicos. También profundizamos en uno de los grandes focos de la compañía: la salud femenina, la fertilidad, la perimenopausia, la menopausia y la necesidad de construir modelos médicos específicos para mujeres. Además, exploramos la historia de Base4, el origen científico del proyecto, la ronda de inversión, sus próximos retos comerciales y regulatorios, y por qué su ambición va mucho más allá de crear un test: quieren contribuir a una medicina más preventiva, personalizada y accionable. 👉 Suscríbete a Itnig para más conversaciones sobre negocios reales, startups y marcas 🎙️ ¿Quieres participar en el podcast de Itnig o patrocinar uno de los episodios?Aparecer en el podcast: https://tally.so/r/wo1Poe Patrocinar el podcast: https://tally.so/r/3EERLN SOBRE ITNIG:🐦 X - ⁠https://x.com/itnig 💡 LinkedIn -⁠ https://linkedin.com/company/itnig 📸 Instagram - ⁠https://instagram.com/itnig 💌 Newsletter - ⁠https://itnig.net/newsletter/⁠ 🌐 Web -⁠ https://itnig.net/ ⁠ESCUCHA NUESTRO PODCAST EN: 🔊 Spotify: ⁠http://bit.ly/itnigspotify⁠ 🎙️ Apple Podcast: ⁠http://bit.ly/itnigapple⁠