Howie: The AI Secretary | Austin Petersmith, Co-founder and CEO
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
In this episode, Turner Novak speaks with Austin Petersmith, co-founder and CEO of Howie, the AI secretary. (00:16) The conversation explores how Howie launched two weeks ago with a viral video and customer enthusiasm, despite the incredible difficulty of building an AI assistant that actually works. Austin and his team took a highly opinionated approach by focusing on the narrowest possible use case - email-based scheduling - which has led to consistent double-digit monthly growth for the past year. (01:06) The discussion covers their decision to completely rebuild the product with humans in the loop, similar to how autonomous vehicle companies train self-driving cars, their viral launch strategy, and lessons on building AI products that augment rather than replace human capabilities.
- Main Theme: Building focused AI products that actually work by combining artificial intelligence with human oversight, demonstrated through Howie's journey from concept to successful launch in the narrow but universal use case of scheduling meetings.
Speakers
Austin Petersmith
Austin is the co-founder and CEO of Howie, the AI secretary that specializes in email-based scheduling. He previously worked for prominent investor Jason Calacanis and has experience building products since 2013. Austin is also an angel investor alongside his brother through their fund Cough Drop Capital, with successful early investments in companies like Lattice, Superhuman, and Mercury. He splits his time between building Howie and selective angel investing, focusing on products and founders he uses or knows personally.
Turner Novak
Turner is the founder of Banana Capital and host of The Peel podcast. He is an investor in Howie and has been publishing episodes explaining the world's greatest startup stories for over 100 episodes. Turner focuses on conversations with founders and operators building innovative companies across various sectors.
Key Takeaways
Choose Extreme Narrowness Over Broad Capabilities
Austin emphasizes that focusing on one narrow use case is crucial for AI products. (11:41) While it's tempting to build generalist assistants that can handle multiple tasks, the reality is that narrow focus allows for genuine problem-solving. Howie chose scheduling specifically because it's frequent, tedious, affects billions of professionals, and has natural distribution built in since multiple people are involved in scheduling interactions. This narrow focus allowed them to build the architecture and solve one thing extremely well before expanding to other areas.
Embrace Human-in-the-Loop Architecture
Rather than pursuing fully autonomous AI, Austin explains how they implemented humans to correct mistakes before they happen, similar to how Cruise and Waymo train self-driving cars. (23:58) This decision was made because mistake tolerance is extremely low in scheduling - unlike ChatGPT where errors are acceptable, scheduling mistakes can damage important business relationships. The human-in-the-loop approach allows them to deliver a product experience that exceeds current AI capabilities while continuously improving the underlying models.
Optimize for Product Value Before Growth Mechanics
Austin advocates for building inherently valuable products rather than relying on growth hacks early on. (85:36) Howie doesn't send reminder emails, push notifications, or use typical retention tactics - yet maintains 30-40% daily active users among monthly actives. This approach provides ground truth about product value because there are no artificial mechanisms driving engagement. The best products grow themselves through word-of-mouth, and focusing on this organic growth validates genuine product-market fit.
Design Pitch Decks for Three-Second Attention Spans
Based on analyzing DocuSend metrics, Austin discovered that most people spend only three seconds per slide but will go through an entire deck. (83:14) His strategy is to create more slides with less information - maximum 8 words per slide with narrative arc that pulls people through. Think of it like a Twitter thread where each slide contains one digestible nugget of information. This approach ensures your key points get consumed rather than ignored in dense, paragraph-heavy slides.
Build Meta Work Solutions, Not Job Replacements
Austin identifies that professionals spend enormous time on "meta work" - work about work - rather than their actual core responsibilities. (63:54) This includes scheduling, setting reminders, reviewing notes, and mental overhead of tracking tasks. Rather than trying to replace people's jobs entirely, there's massive opportunity in building AI that eliminates meta work while keeping humans in control of their actual expertise. This approach is more realistic and creates sustainable competitive advantages.
Statistics & Facts
- Howie has grown at 30-50% month-over-month consistently, maintaining 30-40% daily active users among monthly active users - indicating strong product engagement. (27:19)
- The company launched with over 1,000 paying customers and schedules over 5,000 meetings per week for users, demonstrating significant scale and usage. (72:01)
- Austin's viral launch video received approximately 1.2 million views across different posts and converted to substantial customer growth and revenue. (46:01)
