When Anyone Can Build Anything: The New AI-Native Stack
Overview
| Experience | In Person |
|---|---|
| Track | Artificial Intelligence & Agents |
| Industry | Enterprise Technology |
| Technologies | Databricks Agents |
| Skill Level | Beginner |
The act of building, whether software, knowledge systems or workflows, is being fundamentally democratized. Natural language is becoming the new interface between human intent and outcome. Natural language is becoming the new programming language. Ideas are becoming applications in minutes, not months. This panel brings together the founders of AI-native tools that are collapsing the distance between human intent and running software. Together, they'll explore a near-future where dramatically more people can ship products, and where the primary bottleneck shifts from technical skill to creative vision.
The big questions:
- What happens when the cost of building software drops by orders of magnitude?
- Are we heading toward a world with far more software, fewer traditional developers or both?
- When AI agents can build, test, and deploy autonomously, how does engineering culture evolve?
- What's still hard and what will remain uniquely human in the development process?
- Where do AI agents break down and are we moving fast enough to find out before it matters
Join the founders reshaping the most fundamental act in technology for a candid look at where this lands in the next 2–3 years.
Session Speakers
Ashwin Sreenivas
/Co-Founder
Decagon AI
Jeff Wang
/Windsurf
Patrick Wendell
/Co-Founder and VP of Engineering
Databricks
TR Vishwanath
/Co-founder & CTO
Glean, Inc
Full Summary
Where AI Value Actually Lives: Lessons From the Founders Building on Top of Foundation Models
Early in the foundation model wave, it was unclear whether enduring value would sit with model providers or higher up the stack. A conversation hosted by Databricks co-founder Patrick Wendell with leaders from Decagon, Glean, and Cognition points to a clear pattern: application-layer companies are capturing outsized value by solving the enterprise-specific work that raw models cannot. The discussion aligns with themes Databricks sees across customers, including a shift toward agentic analytics and AI-powered business workflows and the broader question of what AI analytics requires to work on governed data.
FAQ
Models provide raw capabilities, not production systems. Enterprises still need secure connectivity to all systems, permissioning that mirrors existing governance, specialized infrastructure for testing and verification, evaluation harnesses, hallucination safeguards, and workflow tooling. Those are substantial engineering lifts that most organizations prefer to buy rather than build from scratch.
Treat models as a portfolio. Use task-specific evals and route to the best option for performance, cost, or policy constraints. Admin controls and auto routing prevent overuse of large models. For some workloads, small, specialized models can outperform frontier models on accuracy, latency, and cost when narrowly scoped.
Yes, because large language models handle variation off the happy path that rigid decision trees could not. When the AI can solve the issue, customers get fast, personalized resolutions. When it cannot, it hands off with full context to the right human team so customers avoid repeating themselves.
Evidence from adopters points the other way. Agents take on high-ROI, tedious work, which increases per-engineer leverage and allows teams to advance roadmaps faster. Many organizations using coding agents are hiring more engineers to capitalize on the expanded capacity.
Start with measurement. Token usage is typically long tailed, so visibility into who spends and on what reveals high-ROI patterns to scale and areas to rein in. Combine admin controls, auto routing, and model portfolios with clear performance and cost modes. Some vendors back outcomes with ROI guarantees tied to shipped work.