Pittsburgh TechFest 2026

In the age of AI, your product is a commodity. Your community isn't.
When anyone can ship a working product by talking to an AI, product stops being a defensible advantage. Features get cloned in a weekend. What can't be cloned is a room full of people who trust you, use the thing, and tell each other about it. Distribution was always the hard part. AI just made it the only hard part. 

This session talks about how to use community as both the go-to-market and the moat, and shows how startups actually build one: choosing the room before the roadmap, measuring social and business impact, and treating every community event as part of the business plan.

As AV systems become more advanced, the human experience can easily get lost in the technology. This session introduces DONATE—Discover, Observe, Navigate, Align, Test, and Elevate—a practical human-centered design framework created specifically for AV experiences, including interactive touchscreens, software, and integrated environments.


Through real-world examples, attendees will learn how to uncover the right problem before designing a solution, observe actual user behavior, balance technical and human needs, align diverse stakeholders, test ideas before investing in a large-scale rollout, and continuously improve after launch. The session will also explore accessibility beyond basic compliance, methods for managing competing priorities, and ways to create a single source of truth throughout complex projects.


Attendees will leave with an adaptable process that helps reduce risk, improve usability, strengthen stakeholder alignment, and create AV experiences that feel more intuitive, inclusive, and valuable to the people using them.

LLMs appear to vindicate the "bitter lesson": AI researchers succeed less by incorporating domain knowledge and more by scaling data and compute. Taken as prescription, “all you need is scale”. Indeed, practitioners in many domains have moved from traditional ML methods with domain-driven feature engineering to foundation models and other deep learning approaches.
However, theoretical and empirical facts complicate this picture. Data alone, however massive, cannot identify the function we want to learn: identical training pipelines can agree on held-out test data yet diverge sharply outside that distribution.
Despite this, today's deep learning systems generalize impressively well. If training data alone can't determine which of many possible internal models is "correct," where does this success come from? One answer: we never stopped imposing structure; we only made it implicit, built into architecture and learning algorithms rather than features. Tabular data shows what happens without it. No architecture has yet captured its implicit structure the way convolutions capture images or attention captures sequences, so industry still relies on traditional methods and feature engineering other domains moved past. The opportunity is to make that structural choice deliberate, and nowhere is it more available than in tabular data.