149: I Am The Harness

Ian and Aaron recount their recent trip to Vegas, highlighting the annual KibanaConf gathering, which has been running for over a decade, and their experiences there. Ian discusses his poker game, noting the professional players and his strategic choices, including risky bets and the fun interactions at the table. Aaron adds humorously about the gambling mishaps and the level of their playing, emphasizing the importance of being prepared with cash. They also chat about the specific casino activities like blackjack, roulette, and craps, sharing their perspectives on game entertainment and dealer attitudes, and joking about the thrill and loss involved. They humorously criticize casino practices, including the idea of a gaming level upgrade request, and suggest that gambling should be more engaging and social, possibly at a large resort pool for a big 'LairaCon.' Aaron proposes a fun, controversial plan for a festival centered around a casino environment, which Ian dismisses as impractical.

A significant portion of the episode explores Ian and Aaron’s technical projects, especially surrounding Solo and AI models. Ian shares his work on training a small classifier to analyze terminal grids, capturing data across different platforms for diverse, realistic training inputs. They discuss using AI models like Quinn for internal tooling, including automating onboarding, documentation search, and command execution, leveraging the latest open-source models hosted on Cloudflare. Ian elaborates on how these models are integrated via API, securely, without exposing sensitive keys, and the potential future of tiny, on-device models capable of local inference at high speed. The conversation dives into the technical and strategic benefits of these AI integrations, including improving user experience, automation, and debugging. They emphasize that these models can run across different operating systems and are aimed at making development and monitoring more automated and accessible.

Ian describes his challenging travel experience, including delayed flights, swapping planes, and discomfort on a regular economy seat. Despite the inconvenience, he shares his positive experience with the Rivian self-driving features during the car trip home, highlighting the future potential of autonomous vehicles—an area both find highly promising. Ian discusses using Zooks, a driverless transportation service in Vegas, noting its convenience and futuristic appeal, though recognizing current limitations such as navigating complex scenarios and regulatory hurdles. They reflect on the broader outlook for self-driving vehicles, predicting widespread adoption in ten years, especially for limited-area use like buses or city transit, which could significantly reshape urban mobility.

Throughout the episode, Ian and Aaron express enthusiasm for how AI tools are enhancing their workflows. Ian talks about using AI for code review, generating data structures, and improving software robustness through data structure and algorithm audits. They demonstrate how AI models help identify bugs, suggest structural improvements, and assist with onboarding and documentation, making the development process more efficient. Ian discusses training custom models that assess terminal states, aiming for near-instantaneous diagnostics and decision-making. Both emphasize the importance of good tools and infrastructure, like Tailscale for remote access and secure API interactions, to better manage complex, distributed development environments. They see AI as a transformative force, enabling smaller, smarter models that run locally and facilitate automation, debugging, and onboarding.

In addition to technical deep dives, the episode includes humorous personal anecdotes such as Ian’s travel experiences, including his struggles with poor airline service and his adventures in Las Vegas. They share their thoughts on innovative transportation, including driverless cars and Zooks, and reflect on the societal and regulatory challenges these innovations face. There are lighthearted discussions about the challenges of managing software and AI systems, user onboarding, and UI improvements. The hosts maintain a playful, candid tone, blending industry insights with personal stories, humor, and observations about modern technology, travel, and future possibilities.