David Zeng didn't set out to build for creators - a chance meeting with an e-commerce entrepreneur at an event pointed the way, and a Stanford PhD turned out to be the price of entry for a problem ChatGPT now solves in seconds
In this AI Central Voices interview, we sat down with David Zeng, Co-Founder and CTO of Beacons - the AI-powered platform helping more than 10 million creators run and grow their businesses online
Zeng is the person who builds and maintains everything a Beacons user can see or touch. Before Beacons he earned a PhD in Electrical Engineering from Stanford, developing machine learning and signal processing algorithms for medical imaging
His throughline is simple and a little uncomfortable for anyone selling software: creators don't want better tools, they want fewer problems - and that gap is exactly where he thinks AI wins where software never could
Key takeaways:
Build from a real customer, not a thesis - Beacons' creator focus came from one entrepreneur with one urgent problem, not a grand plan
Sell outcomes, not features - creators want fewer problems, which is why AI beats software here
Design for users who can't debug you - consumer AI has to work out of the box, at high accuracy, because creators lack the context to fix it
Go AI-native on purpose - Beacons rebuilt how the whole company works around AI, and the leverage compounded
🔗 Connect with David
In two sentences - who are you and what do you do?
I'm David, I'm cofounder and CTO of Beacons. My job is to build and maintain everything that a user can see or touch :)
You went from a Stanford PhD to co-founding Beacons - what problem did you see that everyone else was missing?
I'm going to be honest, this wasn't something that we magically divined. The way we got started with the current version of Beacons is when one of the cofounders randomly met a serial e-commerce entrepreneur at an event. She used entirely influencer marketing and wanted us to build AI to look at creators' image and read their comments to find the right creators for her. Back in 2019, you really did need a PhD to work on this problem (in 2026, ChatGPT can do this trivially). We knew we wanted to build an AI company and this was a very real problem with a very real customer in front of us. The rest of the story is not that linear, and the company today isn't doing exactly that, but that's how we got started with creators.
The kind of AI that we're building for the creator economy needs to work out of the box with high accuracy.
Walk us through one concrete way Beacons changes the day-to-day for a working creator - a real workflow, not the pitch
We're building an AI teammate that can help creators email with brands and agencies as they do brand deals. Right now creators that monetize through brand deals spend a lot of time in their email inbox, replying to messages, vetting brands/agencies/opportunities, and negotiating on very limited information. Our AI can vet deals and draft emails for creators, saving them a lot of time and getting them better deals in the end.
What's the most common thing people get wrong about the tools creators actually need?
Creators don't want better tools - they want fewer problems. This is why AI can be so transformational for creators where software couldn't.
We did flip the whole company upside down to redo the way we work.
What's in your AI stack? What is the one AI tool or workflow you personally rely on every week?
I still use ChatGPT as my daily chatbot (despite its reputation for now being boomer). I pin it to Thinking models only to get my value haha. I like ChatGPT because it has better product sense and decisiveness. I think it's actually still the smartest. Claude Code is great for doing real work though because it can actually interact with everything, like code, MCPs, browsers, etc. Using the best models on extra high thinking is still such a great deal for the cost. The cost of labor, especially in the US, is so expensive, AI cost for us is still a fantastic deal with how fast and how many things it can accomplish at once. We did flip the whole company upside down to redo the way we work to be AI-native. It was a change, and now it's incredible how many things we can do.
You're building AI for 10M+ creators, most of whom aren't technical - what does AI infrastructure for the creator economy have to get right that infrastructure for engineers doesn't?
Unlike engineers who can go and investigate and fix problems that the AI creates (this was part of their job pre-AI too), creators often won't have the context to do that even if they have the desire. That means the kind of AI that we're building for the creator economy needs to work out of the box with high accuracy.
Where is your field in 12 months - one specific prediction you'd put money on
I'll answer my field as engineering. I feel like today the AI models are around 4-5 years of experience, so a low-end Senior Engineer, without much product sense. I still believe that AI is distilling down every field into its purest forms of expertise and removing a lot of tedium. I think that engineering will reward senior engineers that are experts of systems and architecture. I do worry about junior engineers and current college students. I have an intern that I think is amazing right now, so I don't think it will be impossible, but the bar is much higher all around for new grad hires, not just on technical skills, but also soft skills because I think the bar is increasing on every dimension.
Where should readers find you, and what's the first thing they should join or read?
My creator goal is to post on LinkedIn every day that I work. Find me on LinkedIn at linkedin.com/in/davidyzeng and please read my daily musings haha.
AI Central Voices is where the AI Central team sits down with the founders, executives, and builders shaping AI - going behind the scenes of how they operate, what they're betting on, and where the industry goes next.
Want to be featured, or have an event you'd like us to cover? Reach out at [email protected]
