{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"START","title":"START: Benjamin Swerdlow, CEO & Co-Founder, Freestyle: \"Full Linux VMs for AI Agents. Built for complex tasks that run for hours, days, or weeks.\"","html":"","width":"100%","height":180,"duration":1023,"description":"“I don't believe Claude Code will exist in its current form in six months”\nBen Swerdlow, founder of Freestyle, thinks coding agents are moving from local machines to the cloud, where a single task could get the attention of 20 agents at once.\nEach gets a complete copy of your stack, production environment included. Each can spend a week testing and refining its approach. You compare the results and take the best one forward.\nThe economics aren’t there yet. Ben expects cost per task to fall another 99% over the next four years, making that level of parallel work practical.\nFreestyle builds the computers for it: full Linux VMs for tasks that run for hours, days, or weeks. Clone a running machine, memory included, and let agents pursue different approaches from the same starting point. Pause and resume with their state intact.\nInside Freestyle, Ben already gives agents a week to find improvements to its VM technology. Roughly 700 tests and 90 metrics help the team judge whether the work made things better.\nHe calls this goal engineering: define the outcome clearly, give agents time to work toward it, and measure whether they’re making progress.\n\"Full Linux VMs for AI Agents\"\nBuilt for complex tasks that run for hours, days, or weeks\n\n🎙️ Benjamin Swerdlow, CEO & Co-Founder, Freestyle on Fondo START \n‍\n00:57 Why coding agents could move from local to the cloud\n02:22 From harness engineering to goal engineering\n03:08 Giving agents a week to improve measurable results\n04:13 Why defining the problem becomes the bottleneck\n04:48 How early access to o1 changed Freestyle’s direction\n05:51 Why frustrated sandbox users revealed a bigger opportunity\n07:28 Giving agents a computer instead of building custom tools\n08:12 Getting into YC \n11:17 Snapshotting VMs and running agents in parallel\n12:24 The economics of letting multiple agents attempt every task\n13:26 How GPU supply could drive down agent costs\n15:09 Try Freestyle and follow Ben\nLearn more at freestyle.sh","thumbnail_url":"https://img.transistorcdn.com/q-E1hh7K6IS4AfZiNy2p4MYVGcUOO8lQP92h8QbOEOA/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS80NzVj/MDEzNjkxNjU1N2Uy/NDFhMDQ3M2ZhNWI3/NWY0MS5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}