Emergent agents learn across user sessions using long-term trajectory memory
Madhav Jha · AI Is Unlocking Millions Of New Builders · Y Combinator · Mar 16, 2026 · at 11:45
Emergent co-founder Madhav Jha describes Emergent's continual learning architecture.
0:00 / 0:14exact quote · 14.4s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“We were able to figure out, okay, all the trajectories that we are generating, we can kind of aggregate over time and like sort of build in a long-term memory for the agent, which is very unique in the sense that your agent learns not just from your own session. It learns across the sessions.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Madhav Jha
Jha: Matching build and deploy infrastructure reduces agent deployment errors
“If you give your agents the same in front during the build time and the same in front during the deploy time, Then the sort of like during this like deployment phase, you don't encounter those many problems.”
Madhav Jha: Emergent saves $3,000–$4,000 monthly replacing Asana internally
“And we are also saving like, you know, like 3004 thousand dollars a month in subscription.”
Emergent augments foundation models with fine-tuned verification layers instead of building models
“We don't want to, like, build a Opus 4.5 alternative right away, but we do want to augment it through our custom fine tune verification layers.”
Madhav Jha: Emergent developers use their own agent instead of Cloud Code
“Our coding agent is so powerful that we basically internally use it as a replacement for cloud code as developers, right?”