Скачать или смотреть Shipping AI Products at Speed – lessons from Daydream and Microsoft Amazon days By Matt Fisher

"Shipping AI Products at Speed – Lessons from Daydream and Microsoft/Amazon Days"

​Speaker: Matt Fisher

​We're thrilled to have Matt Fisher with us. Matt is the CTO and co-founder of Daydream, an AI-powered fashion discovery platform that raised a $50 million seed round in December 2023. Before Daydream, Matt led Data and Applied Science at Microsoft and managed measurement and marketing science for Amazon Prime Video. His experience spans building scalable AI systems and leading data science teams at the forefront of consumer technology.

​Moderator: Lan Tian

​Joining us as the moderator is Lan Tian, co-founder of Peak Mojo, an AI-native platform that automates sales team hiring and assessments. Lan's background includes engineering roles at Uber and a focus on building AI tools for high-growth teams. He's also an angel investor and community builder in the Seattle startup ecosystem.

​​​Our Partners:

​​​Hosted by FlickBloom: End-to-end social media marketing for startups.

Introduction & Background
Matt is an experienced entrepreneur and tech veteran in the Seattle community for 26 years. His journey is unconventional:

Non-Traditional Start: He dropped out of law school and joined an early startup where he taught himself to code (Python, C).

Corporate Experience: He worked on massive-scale projects at Microsoft (Bing's monetization algorithm) and Amazon, diving deep into data science and machine learning.

Entrepreneurship: He left corporate life 3.5 years ago to found Daydream, an AI-powered personal shopping assistant. He has since stepped away and is now involved in two new ventures:

A commerce stack focused on dynamic pricing with Sebastian Thrun (founder of Udacity).

A new media company with D.VO Harris (co-founder of GOOD Music) leveraging AI to create and distribute interactive, personalized content like movies and video games.

Key Insights on AI & Technology
The Critical Importance of AI Memory:

Problem: Current AI has a "flat," linear memory, leading to poor personalization and "memory leaks" where it forgets previous interactions.

Biomimicry Solution: Inspired by how human brains store memories (fragmented and associative, triggered by senses), Matt developed a new graph-based memory structure.

Benefits: His open-source research claims this approach is faster, cheaper on storage, and more accurate for memory recall. He believes advanced memory is the key blocker for the next leap in AI.

The Future of User Experience (UX) is Multimodal:

Text is Dying: Matt predicts that text-based input for AI will be largely dead within 3 years. The future is voice and visual (video) interfaces.

Beyond the Chatbox: The UX must shift from a predictable, linear design to a responsive "canvas" that can dynamically adapt to rich, multimodal inputs (e.g., using smart glasses for visual context).

Companies with Impressive UX: He highlighted Cursor (IDE), Walmart, and Home Depot for their dynamic and personalized AI-driven experiences. He was critical of Amazon's "Rufus" for being a basic, unhelpful Q&A chatbot.

The Challenge of AI Agent Authentication:

Current State: It's a "wild west" with emerging protocols like MCP. Standard OAuth and JWT tokens are used but are vulnerable to spoofing.

Future Solution: He believes the solution lies in cloning agents within your own ecosystem rather than making authenticated calls to external ones. He cited Arcade.dev as a company doing interesting work in this space using a hashing/chain structure for secure agent identity.

Advice for Founders & Hiring
Hiring Challenges: He identified two major hiring problems:

Spam applicants making it hard to find real candidates.

Finding people with both the depth of skill and the aptitude to learn quickly in a fast-paced startup environment.

The Age & Experience Balance: He applauds young founders for their passion and speed but warns they often lack the experience to handle scale and production-ready systems. The ideal team has a mix of young talent (for bold ideas) and experienced veterans (for guidance and stability).

Key Lessons from Corporate Experience: The three most valuable skills he brought from his corporate career are:

How to handle scale.

The ability to write effectively and form structured opinions.

Learning how to balance work and life and trust teams.

Personal Insights & Inspiration
Sources of Inspiration: He gets UX and product ideas from:

His children (for their literal, preconception-free thinking).

Sci-fi books and movies.

The art world, following the principle of "steal like an artist" from the book of the same name.

Most Rewarding Part of Founding: The adrenaline rush of going from 0 to 0.1—seeing an idea become reality and used by people—and learning to trust his gut when steering the company.