Skip to content
YC Root AccessYC Root Access

Sim: The Visual, End-to-End Agent Builder

Sim is building the visual, end-to-end agent builder — a platform where developers can design, test, and deploy real AI agents that work in production. Founded by Emir Karabeg and Waleed Latif, Sim grew from a small San Francisco apartment to a community of 60,000 developers and 18,000 GitHub stars, recently raising a $7M Series A. In this interview with YC’s Aaron Epstein, Emir shares how they’re building the infrastructure for the agent era, and what it takes to create AI systems that reason, retrieve, and act safely at scale. Learn more about Sim at https://www.sim.ai. Chapters: 00:00 – Building the Visual Agent Builder 00:36 – What Sim Does and Why It Matters 02:05 – From a SF Apartment to 60,000 Developers 04:20 – Finding Early Users and Product-Market Fit 07:05 – Designing an End-to-End Platform for Agents 10:30 – Powering the Open Source Agent Ecosystem 13:25 – Raising a $7M Series A 16:20 – Building Agents That Actually Work 20:10 – The Infrastructure for the Agent Era 23:40 – Advice for Builders in AI

Aaron EpsteinhostEmir Karabegguest
Nov 11, 202525mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

Sim.ai builds Figma-like canvas for end-to-end AI agents workflows

  1. Sim.ai is a visual, AI-native agent builder that lets users create end-to-end workflows via drag-and-drop plus natural-language prompting.
  2. The company pivoted from a misguided marketing product to Sim after experiencing firsthand how painful and complex it was to manage many interdependent agents in code.
  3. Sim’s growth strategy centers on being truly open source and repeatedly orchestrating launches to hit GitHub Trending, driving both stars and cloud signups.
  4. Monetization comes from a hosted cloud platform where developers pay for inference and get an easier setup than running locally.
  5. Karabeg emphasizes founder execution habits—shipping on a fixed cadence, staying close to what you understand, and optimizing for long-term learning and endurance.

IDEAS WORTH REMEMBERING

5 ideas

Build where your pain is strongest and your understanding is deepest.

Sim emerged from the team’s prior product where coordinating “hundreds of agents” was internally confusing; the pivot worked because they leaned into what they already knew (LLMs/agents) instead of unfamiliar industries.

Differentiate by being AI-native, not just “workflow software with an LLM step.”

Karabeg argues legacy automation tools lack AI-first primitives like easy iteration, natural-language construction, and scaling an agent to run many times with synthesis—capabilities already common in code but absent in older platforms.

Open source works when it’s genuine and paired with a clear platform vision.

Sim’s bet is that code won’t be the moat in an era where models can replicate apps quickly; the moat becomes product vision, velocity, and an ecosystem that builds on the foundation.

Treat GitHub Trending as a growth surface you can deliberately engineer.

They coordinate simultaneous announcements (Hacker News, Product Hunt, Twitter) to create a burst of attention that drives Trending placement, which then produces hundreds-to-thousands of stars per day.

Convert OSS interest by making the hosted path the easiest path.

Sim links its cloud product directly from GitHub and observes a strong correlation between star spikes and new cloud users; the cloud offers convenience and paid inference versus local setup.

WORDS WORTH SAVING

5 quotes

Sim is a platform to build AI agents… a Figma-like canvas with Cursor style side prompt.

— Emir Karabeg

I basically only know things about LLMs and building agents.

— Emir Karabeg

If we get this to 100 GitHub stars, I’ll be super happy… Actually you’re right. So if it gets to 10, I’ll be satisfied.

— Emir Karabeg

Code is no longer the proprietary part of a company anymore as it used to be.

— Emir Karabeg

We don’t go live because it’s ready, we go live because it’s Saturday night.

— Emir Karabeg

Figma-like visual canvas for agentsAI-native workflow orchestration vs traditional automationPivoting and founder-market fitOpen-source licensing ethos (MIT/Apache)GitHub Trending launch orchestrationCommunity building (HN, Twitter, Discord)Cloud monetization via inference

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome