September 28, 2026
One screen defines the whole interface: Peng Zheng's bot extends it across every screen
On Sep 28, 2026, the Grok Bot team outlined six techniques for working with bots. Peng Zheng creates the first 5% of an interface and one keyframe, then a design bot uses Figma MCP to extend it across every screen. Lauren Tan hands large projects to Matcha, an eng-lead bot: it breaks work down for other bots and does not write code itself.
Peter Yang
@petergyang
6 things I learned from @poteto and @pengzheng_, the engineering and design leads at Grok Bot, about getting the most out of your bots: 1. Build a design system and one keyframe, then let the bot scale the design Peng's design bot edits Figma directly through Figma MCP and uses a skill that knows his files and design system. Peng creates one keyframe, then asks the bot to continue the design across every screen in the flow. 2. Create an engineering-lead bot to manage the other engineering bots Lauren's engineering-lead bot never writes code itself. It divides projects into small tasks for her engineering bots, which launch agents for cloud-based coding. That is how she manages “really huge swarms of agents.” 3. Let bots check their own work Lauren's bots check every change before merging, and the Grok Bot codebase is designed so agents can make changes safely. “Sometimes I do not even look at a PR until it has already been merged.” 4. Put product manager, designer, and engineer bots in one group chat When Peng wants to test a product idea, he adds product manager, designer, and engineer bots to one chat. They challenge him and help put together the requirements. 5. Run through a task manually with a bot, save the successful approach as a skill, then schedule it to run regularly Lauren watches the first run, fixes mistakes, and turns what worked into a skill. As soon as a skill works in one run, she schedules it to run regularly and stops babysitting the bot. 6. Name bots after food or something fun Lauren's bots are named Daifuku, Gyoza, Matcha, Crumb, and Katsu. It keeps her hungry all the time. Just kidding. I had to make her the image below. You are welcome, @poteto 🙂 📌 Full episode: https://youtu.be/xZ5TEaleUdg Text version: https://creatoreconomy.so/p/grok-bot-team-14-best-bots-peng-zheng-lauren-tan
"Everything I touch with my keyboard and mouse, I try to delegate to my bots." Here's my new episode with @poteto and @pengzheng_, the eng and design leads for Grok @bot, where they showed me the 14 bots they use for work and life, including: → A design bot that turns one keyframe into a full user flow → An eng lead bot that manages a team of eng bots → How to trust your bots with more of your work Some quotes from both: "I like to call it the Michelin kitchen…when you say software factory, it has this connotation of mass manufactured slop." "Sometimes I actually don't even look at the PR until after it's landed and then I'm like, 'Oh, okay. Yeah, that looks good.'" "I think it ultimately comes back to trust. First, watch your bot work and correct it. Turn what worked into a skill. Once it nails the task in one shot, make it a routine." 📌 Watch now: https://youtu.be/xZ5TEaleUdg Thanks to our sponsors: @meetgranola: AI meeting notes that don’t suck https://granola.ai/peter @RiversidedotFM: All-in-one AI studio for podcasts and video https://creators.riverside.com/PeterYang
· 2K views
Peng Zheng creates the first 5% of a design and one keyframe. His design bot knows the file structure, colors, typography, and spacing of the design system. It then uses Figma MCP to expand this example across the entire flow.
Work becomes a habit. Lauren Tan watches the first run, fixes mistakes, and saves the successful approach as a skill. When a skill works in a single run, she turns it into a routine.
For Codex, Figma recommends installing the plugin through Plugins → "+" next to Figma → Install Figma and completing OAuth. Remote Figma MCP works on all plans and does not require the desktop app. Writing to the canvas requires a link to a Design file or a selection, plus edit access.
The next step in Lauren Tan's setup is already defined: a successful skill becomes a routine, after which the bot works without constant supervision.
