Airtable CEO: This Is What the Top 1% Do With AI | Howie Liu

📌 Launch your dream business with the best designs the world has to offer — Design.com: https://go.design.com/cd5msoz Howie Liu co-founded Airtable — used by over 80% of Fortune 100 companies. Now he's building HyperAgent. He asked me: if you could hire as many people as you want for almost zero cost, what roles would you fill first? I didn't have an answer. We covered the two skills that will make you superhuman in the agent era, how to build a virtual twin of yourself so you stop being the bottleneck in your own company, and why builders are going to win the next few years. He also showed me his real setup live: an agent that monitors X 24/7 and pushes him only what matters, and a full billboard campaign his agents ran end-to-end. If you've been reading about agents but haven't deployed one yet, Howie gives you three steps to start today. 📌 Limited BONUS: Hyperagent is giving away $1,000 in free credits to the first 1,000 subscribers who sign up through my link — claim yours here: https://hyperagent.com/invite/MARINA-HA?utm_source=youtuber&utm_medium=influenceer&utm_campaign=MARINA-HA *Timestamps:* 00:00 Intro 00:43 The real state of AI right now 03:06 The evolution no one mapped 07:58 Claude vs. ChatGPT vs. HyperAgent — how to actually choose 12:12 Building a virtual twin of yourself with AI 15:39 Live demo: Howie's real productivity setup 20:22 How to develop the builder mindset 24:58 Fun agent use cases 29:05 First 3 steps to start with agents today 31:15 The two skills that will make you superhuman *Links:* 📩 Follow my Newsletter: https://siliconvalleygirl.beehiiv.com/subscribe?utm_source=youtube&utm_medium=video&utm_campaign=futureproof-sub&utm_content=HowieLiu 🔗 My Instagram: https://www.instagram.com/siliconvalleygirl/ 📌 My Companies & Products: https://Marinamogilko.co

Marina MogilkohostHowie Liuguest

May 27, 202632mWatch on YouTube ↗

CHAPTERS

From chatbots to agents: closing the loop on execution
Marina frames the core shift in AI: moving from prompt-and-pray chatbots to agents that can learn your preferences, act more autonomously, and improve over time. Howie sets the stage for what “agentic” work looks like in practice and why it changes personal and company productivity.
The real state of AI: agent hype was early, capability is catching up
Howie describes the last few years as phases: pre-ChatGPT, chatbot maturation, then premature “year of agents” hype. He argues model capability has now reached a threshold where more human-like autonomy is realistic, even if full company automation isn’t.
Managing a fleet of agents: the new managerial form factor
Using developer tooling as an analogy (Copilot → Cursor → parallel agents), Howie explains the emerging workflow: orchestrating multiple agents in parallel, including overnight. The human role shifts toward management, prioritization, and review rather than direct execution.
Sponsor break: rapid AI design workflow (Design.com)
Marina shares a tool recommendation for creating brand assets quickly with AI. The pitch focuses on generating, refining, and maintaining consistent design across channels without long revision cycles.
Humans in the loop—yet dramatically more leveraged
Howie argues the near-term reality isn’t total automation but an increasing percentage of the workflow handled by agents. Marina confirms this in content operations: more output without proportionally growing the team, with humans focusing on review and direction.
Choosing tools: chatbot products vs frontier agents (Claude, ChatGPT, Cowork, HyperAgent)
Marina asks how to pick among rapidly changing tools. Howie proposes a simple taxonomy: distinguish “chatbot-era” products from “agent-era” systems designed for hours of autonomous work, then choose based on autonomy needs and team workflows.
Building a ‘virtual twin’ + Telegram as an agent hub
They explore the idea of creating an agent that mirrors your preferences and context—effectively a “virtual twin.” Telegram emerges as a practical deployment surface: easy bot ergonomics, group-chat collaboration, and learning from ongoing feedback loops.
HyperAgent productivity demo: always-on assistants and notification pipelines
Howie walks through a demo-like setup: agents can run on schedules or in always-on “heartbeat” mode, pushing updates through Telegram/Slack/email. A highlighted example is an agent monitoring X/Twitter and alerting only when information is relevant.
Democratizing agents: builder mindset and the tinker-to-fluency path
Howie connects AI’s trajectory to personal computing: increasing power can either become specialized and inaccessible or broadly enabling. He argues “builders will win” by tinkering, learning emergent best practices, and focusing on outcomes rather than old activities.
Fun and ‘luxury hire’ agents: billboard scouting, video concepts, cars, travel
Howie shares creative and personal agent use cases that feel immediate and high-impact: end-to-end billboard selection and mockups, generating ad concepts and videos, plus lifestyle automations like car monitoring and points-optimized travel planning. The key idea is “hiring” roles you’d never justify as full-time humans.
Getting started today: three steps + the two ‘superhuman’ skills
Howie gives a practical onboarding sequence: pick a friendly tool, identify problems worth solving, and keep it fun to build passion and fluency. They close on two compounding skills for the agent era—problem finding and judgment—positioned as the real differentiators as execution becomes cheap.