How I AI

Owen Williams, a design manager at Stripe with an engineering background, built an internal prototyping tool called 'Protodash' that lets designers and PMs build high-fidelity, on-brand Stripe prototypes without needing deep coding knowledge. The tool evolved from a set of Cursor rules and a React app into a full browser-based studio ('Protodash Studio') with AI-assisted iteration, design review modes, and dev box hosting. Unexpectedly, PMs became even heavier users than designers, transforming how product work is communicated and reviewed at Stripe.

The speaker argues that companies should give their marketers creative freedom or risk losing them. Drawing from personal experience, they left a company that restricted their work, then raised money and built their own business. They predict a broader trend of marketers and non-technical people leaving restrictive companies.

The speaker describes a quirky, highly personal DIY productivity tool — a Raspberry Pi duct-taped to a keyboard that lets them blind-type rough notes in the dark, which an LLM interprets and converts into to-do list items. The device is intentionally unpolished and built solely for personal use, not for scale.

The speaker reflects on the paradox of building beautiful UX while knowing that 'no UX is the best UX.' Drawing on a quote from Ramp's CTO and an anecdote from their lead investor, they argue that the future of software is agent-driven, where users won't need to interact with interfaces at all.

Jason Levin, CEO of Memelord, discusses building a meme generation API that enables AI agents to create contextual memes for brands. He shares his journey from non-technical founder using no-code tools to building a company that generates memes through AI, emphasizing the importance of entertainment in marketing and empowering non-technical team members to code.

A speaker demonstrates GPT 5.5 building an educational app for teaching second-grade subtraction, questioning whether 17+ minutes of AI processing time is necessary for such simple tasks. They argue that most users don't have problems complex enough to justify this level of computational power and suggest the interface may not be optimal for non-technical users.

A user tested ChatGPT's image analysis feature for color analysis, finding it initially categorized them as warm neutral but successfully corrected to dark winter when provided feedback. The AI-generated color palettes and styled images were deemed impressive despite some visual artifacts.

The creator reverse-engineers a Divoom Mini 2 retro Bluetooth speaker/screen by using a packet sniffer to capture Bluetooth traffic from the official app, then feeds those logs to GPT-5.5 (via Codex) to decode the protocol and display custom content on the screen.

A user tests ChatGPT's new image generation capabilities by creating a brand kit for their company Chat PRD. While the initial AI-generated design had good typography and layout, it required reference images from Midjourney to achieve the desired aesthetic that better matched their brand identity.

A developer successfully used GPT 5.5 to solve a complex data migration problem involving millions of rows with unstructured data and edge cases. Previous attempts with other AI tools including Cloud Code and GPT 5.4 had failed to resolve the issue.

A designer tests Claude Design and finds it impressive for incorporating design systems as first-class citizens, but immediately hits credit limits and finds the iteration speed much slower than Figma due to LLM processing delays.

The speaker argues that AI tools like Claude have made it possible to achieve "backlog zero" by dramatically reducing development time. Tasks that previously required months of planning and team coordination can now be completed in a single coding session.

Clara Vel, a product leader and AI enthusiast, shares her hands-on experience testing GPT-5.5 and GPT-5.5 Pro in Codex over several weeks. She highlights the model's superior intelligence and efficiency for complex technical problems, including autonomous security remediation, a multi-million-row data migration, and hacking into a proprietary Bluetooth device. Her core argument is that GPT-5.5's value is best realized by developers with genuinely hard problems, not average ChatGPT users.

A leader at Intercom explains how leadership in the AI era requires giving employees permission to experiment with AI tools while taking accountability for potential failures. They emphasize that all work will eventually become agent-first and leaders must break down barriers to AI adoption.

Claire Valle reviews Claude Design's capabilities for importing design systems and creating marketing prototypes, while also testing ChatGPT's new Images 2.0 model for brand kit creation and layout work. She finds both tools promising but notes significant speed limitations and credit restrictions that impact iteration workflows.

The speaker discusses the future of different API interfaces and emphasizes the importance of providing real-time assistance to agents during their discovery processes. The focus is on enabling agents to complete complex multi-step tasks autonomously without requiring users to handle manual workarounds.

Brian Scanlan from Intercom explains how they doubled their R&D team's pull request throughput in 9 months by implementing Claude Code across their engineering organization. He demonstrates their comprehensive approach including skills development, telemetry tracking, and quality controls that enabled 2x velocity gains while maintaining code quality.

The speaker demonstrates how to create a scheduled AI task that runs daily at 7:30 AM to analyze email, Slack, and calendar data to generate a morning debrief and action plan. The system leverages project context and connected tools to provide personalized daily preparation.

The speaker demonstrates AI orchestration tools that allow users to manage multiple AI agents simultaneously through a dashboard interface. The system shows when agents need permissions and allows users to grant or deny access to various operations.