SUMM

The tech industry is facing an explosion of AI product launches and intense competition across all software sectors.
Large companies (e.g., Notion, Figma, Atlassian, Anthropic, Google, OpenAI) are releasing overlapping AI-powered features and products at a rapid pace.
Numerous well-funded startups are simultaneously entering every AI-enabled software category.
Some established companies have collapsed very quickly due to this competitive environment (e.g., Chegg declined over 90% in months; Stack Overflow suffered after ChatGPT's launch).

The most important challenge is determining: "What do I build and why will it win?"
Answering this question, based on unique customer insights that others are not acting upon, is the key to competitive advantage.
The current climate has made this question much harder to answer—competitive pressures and technological change are ten times higher than before.

The AI competition is intense, with fast-moving incumbents, large horizontal platforms, and rapid technological shifts.
Simply doing more engineering, project management, or adopting the latest infrastructure is insufficient for success.
Success hinges on finding an under-served "seam" in the market and creating meaningful differentiation.
Traditionally, great product leaders achieved this by identifying and acting on unique, unaddressed insights.

Avoid two traps: 1) reinventing AI infrastructure, and 2) just copy-pasting generic AI features without differentiation.
The recommended approach is to treat AI as modular "Lego blocks," integrating best-in-class AI capabilities with proprietary data and product features.
True competitive advantage comes from combining three things: proprietary data, unique product functionality, and deep understanding of unmet customer needs.

AI capabilities (pre-trained models, task automation, etc.) are not differentiators, as they are accessible to all.
Proprietary data is key, as it adds contextual value to AI outputs—unique data sources can drive unique product experiences.
Valuable data types include real-time, user-specific, domain-specific, human curation, and reinforcement data.
The marginal value of your data (its additional contribution beyond what big models already have) is critical.
Unique product functionality (custom workflows, business rules, integrations, algorithms) defines how AI is experienced by users.
Differentiation comes from stitching unique data and functionality into a self-reinforcing, integrated system.

Granola entered an already crowded field of AI note-taking tools, competing with both startups and incumbents.
Their distinguishing insight: instead of aiming to replace note-taking, they focused on empowering users to take better notes, addressing previously unmet needs.
Used off-the-shelf AI tools for transcription and language models but assembled them in a customized way.
Granola's system combines user-created notes with transcriptions to create unique, evolving repositories, unlocking downstream features like cross-meeting chat and project workspaces.
Integrated deeply with Mac OS and user calendars for context and convenience.
The company's differentiation is maintained by continuously "sequencing" new capabilities—for example, integrating with CRMs and experimenting with self-updating wikis.

Lasting competitive advantage ("moat") is now achieved by stacking a sequence of smaller, temporary advantages.
The lifespan of an advantage has shrunk to 2-3 weeks, down from 6-12 months in previous years.
To stay ahead, companies must rapidly execute and evolve by continually stacking new "Lego blocks" of features and data.

Identify unmet customer problems as the foundation.
Examine what unique AI capabilities can solve those problems in novel ways.
Leverage proprietary data to power differentiated solutions.
Design product functionality that enhances AI with unique user-facing "superpowers."
Success requires constant iteration and creative assembly of unique building blocks.

Speaker invites AI engineers to connect and highlights opportunities at Reforge.
Additional resources and information are available at reforge.com.

Survive the AI Knife Fight: Building Products That Win — Brian Balfour, Reforge