CHAPTERS
Sim.ai in one sentence: a Figma-like canvas to build AI agents
Amir explains Sim as a visual, end-to-end agent builder: a canvas where you can drag-and-drop or use natural language prompts to assemble agentic workflows. The goal is to make building agents feel as intuitive as designing in Figma, while still being powerful enough for real automation.
- •Figma-like visual canvas plus prompt-driven building
- •Drag-and-drop workflow composition for agents
- •Agents can run inside products or as background automations
- •Designed for developers, but aiming to broaden who can build agents
Early traction: 60,000 cloud developers and rapid open-source growth
The conversation opens with Sim’s current traction metrics and positioning. Amir highlights both cloud adoption and open-source momentum as core indicators of progress.
- •Crossed ~60,000 developers on the cloud platform
- •~17.5K GitHub stars and growing quickly
- •Positioned as a fast-growing open-source workflow builder
- •Momentum coming soon after YC participation
Founders’ origin story: meeting at Berkeley and the leap to SF
Amir shares how he met co-founder Waleed through a Berkeley roommate connection and how their friendship became the foundation for the company. The turning point was a decisive call where Waleed quit his job and moved to San Francisco to build together.
- •Met at Berkeley; became close friends over several years
- •Decided to start a company right after graduating
- •Waleed quit Amazon and moved to SF after a phone call
- •They hadn’t built projects together before—trust came from friendship and respect
The first startup idea (and YC reality check): personalized landing pages
Before Sim, they pursued a dynamic landing page product tailored to individual visitors using browsing/CRM data. They struggled because they didn’t understand the sales/marketing customer deeply, and a YC interview exposed how unclear the product was.
- •Initial idea: personalized, per-visitor landing pages
- •Tried selling into sales/marketing without domain understanding
- •YC interview forced them to confront confusion and weak clarity
- •December became a month of searching for a better direction
The spark for Sim: internal pain building hundreds of agents
While building the prior product, they ended up assembling many “agents” chained together—research, copy generation, image prompting, image generation, and page assembly. The complexity revealed that orchestration, reusability, and collaboration around agents was the real problem they cared about.
- •Previous project required chaining many small agent components
- •Hard even for the founders to track what agents already existed
- •Agent orchestration became the true bottleneck
- •Motivation shifted toward a visual system for building/understanding workflows
January prototype and conviction: combining Figma + notebooks for workflows
Amir describes the early decision to prototype a Figma-like agent canvas, despite skepticism from Waleed. A five-day experiment turned into full commitment once Waleed joined, and they began building continuously without looking back.
- •Amir’s thesis: merge Figma-like UX with notebook-like iteration
- •Initial goal was tiny (10–100 GitHub stars) just to validate interest
- •Waleed was skeptical but joined after the prototype push
- •Once aligned, they committed fully and iterated daily
Differentiation: making workflows truly AI-native (not old automation tools)
Amir contrasts Sim with legacy workflow automation: Sim is designed for AI-first orchestration, including repeated agent runs, synthesis, and natural-language workflow creation. He argues workflows are a new programming paradigm as abstraction continues to rise (like Cursor for code generation).
- •Traditional workflow tools exist but aren’t “AI-native” in key ways
- •Needs: run an agent many times, aggregate/synthesize outputs, build via natural language
- •Belief: workflows become the next layer above code for software construction
- •Historical framing: programming modalities evolve toward less friction
Open-source strategy: philosophy, distribution, and community trust
Sim’s open-source approach is both ideological and strategic: code is less defensible in an era of powerful models, and adoption/network effects matter more. Amir emphasizes developers value authenticity—being truly open source and engaging actively earns trust and contributions.
- •Rationale: code is less proprietary as models can recreate systems quickly
- •Goal: become a foundational workflow engine others customize by domain
- •Developer trust increases with truly open licensing (e.g., Apache/MIT)
- •Distribution via where developers are: Hacker News, Twitter, community channels
Getting to GitHub Trending: coordinated launches as a growth lever
Amir explains that GitHub Trending was a major driver of star growth, and the team learned to plan launches around it. Coordinated bursts of attention across multiple platforms can push a repo onto Trending, triggering large daily star influxes.
- •Most stars arrived while featured on GitHub Trending
- •Trending can yield hundreds to thousands of stars in a day
- •They plan launch schedules to maximize trending probability
- •Coordinated releases: Hacker News + Product Hunt + Twitter at once
Monetization path: converting open-source users into cloud customers
Sim ties open-source interest to its hosted cloud product: the GitHub repo funnels users to Sim.ai for easier setup and managed execution. Growth in GitHub stars correlated strongly with new cloud users, and revenue comes from paid inference/convenience.
- •Cloud product (Sim.ai) linked directly from the repo
- •Stars and new cloud users historically tracked closely
- •Cloud offers convenience vs. running locally
- •Monetization primarily via paid inference/usage on the hosted platform
Operating cadence and product culture: shipping every two weeks
The team maintains urgency by adhering to a launch calendar with a strict two-week release rhythm. Amir adopts the mindset that shipping is a schedule commitment, not a perfection milestone, extending the YC habit of frequent public launches post-batch.
- •Launch calendar as a forcing function for execution
- •Biweekly releases regardless of perceived “readiness”
- •Quote-driven philosophy: ship because it’s time, not because it’s perfect
- •They grew faster post-YC by continuing the launch momentum
Hiring profile and the ‘agent era’ infrastructure thesis
Amir frames the future as an engineering challenge—bringing agent-building capability to more people, not just specialists. Sim looks for highly curious builders excited by AI’s trajectory, with strong drive and willingness to ship over narrow stack-specific expertise.
- •Belief: biggest remaining work is engineering, not science
- •Mission: democratize agent creation beyond expert developers
- •Hiring for curiosity, passion, and execution over specific languages
- •Team values builders who want to shape the emerging AI toolchain
Founder lessons: long time horizons, sustainable intensity, and honest self-awareness
Amir reflects on the shift from scrappy early days to building for a 10–15 year journey. He stresses compounding through consistent effort, maintaining health to sustain pace, and choosing work you’d still do in the hardest moments years later.
- •Adopting a 10-year mindset (compounding daily effort)
- •Balancing high intensity with health (gym, nutrition, sustainability)
- •Being honest about strengths—returning to programming when it’s the best fit
- •Optimize for long-term learning and conviction, not short-term hacks
