
CV Podcast
AI Coding Is Getting Expensive. This Spokane Startup Is Fixing It | Code Lexica
Why does your AI coding bill jump 5X overnight? Code Lexica co-founders Carter Bastian and Tristan Rees explain the hidden economics of AI coding tools like Claude Code, Codex and Cursor, and how their Spokane startup pivoted from codebase intelligence to cutting AI coding costs.On Episode 43 of the Cowles Ventures Podcast, Blake Gibson sits down with Carter and Tristan to unpack how a platform built to help engineers understand massive codebases turned into a tool that tracks and reduces runaway AI token spend for engineering teams.In this episode: How Carter started building Code Lexica right after GPT-3.5 launched, and met Tristan at a Spokane AI meetup Why their first product, a tool for understanding any codebase, wasn't a painful enough problem to sell The 3 things they got wrong about the AI coding market: how fast the tools improved, how fast adoption happened, and how much trust teams now give AI agents Why an Anthropic bill can jump 5X once a team crosses roughly 30 engineers and moves to enterprise pricing How feeding AI coding agents the right code context cuts both errors and token costs Why stealth mode is a mistake and shipping an imperfect product wins in the AI age Jevons paradox and whether AI will shrink engineering teams or create more demand for code Their long-term vision: a control plane for how companies govern AI coding tools across the whole org If you're an engineering leader, founder, or developer trying to understand AI coding costs, AI adoption, or building a SaaS startup outside Silicon Valley, this one is packed with practical takeaways.CHAPTERS 0:00 Cold open: why your AI bill can 5X overnight 0:43 Meet Carter Bastian and Tristan Rees of Code Lexica 2:06 Building Code Lexica after GPT-3.5 3:50 Finding a co-founder at a Spokane AI event 5:32 Big company vs startup: career advice for engineers 6:45 What is codebase understanding? 9:21 Why the first product wasn't a hair-on-fire problem 11:22 3 things they got wrong about the AI coding market 13:09 AI adoption trends and the governance problem 14:06 The pivot to tracking and cutting AI coding spend 14:55 The two mandates every engineering leader gets 16:07 How code context cuts AI errors and token costs 18:43 Built for developers: zero-friction setup 19:35 Why your Anthropic bill jumps 5X at 30 engineers 20:28 Becoming the control plane for AI coding 21:54 Who Code Lexica sells to, from small teams to enterprise 24:53 Model price hikes and the end of AI subsidies 26:08 What drives AI prices up: regulation, energy and compute 27:44 Will AI shrink engineering teams? 30:00 Jevons paradox and the future demand for code 31:57 Non-technical founders building real software 33:11 Why stealth mode is a bad idea for startups 35:24 Ship an imperfect product, fast 37:58 Building an AI-first team 40:16 Finding message-market fit 42:54 "You'll get tired of your messaging first" 44:48 Where Code Lexica is headed next 48:21 Advice for founders: use 10X more AI 49:15 Hot take: why founders need side projects 50:17 Wrap-upLearn more about Code Lexica: https://app.codelexica.com/ Watch Carter's first appearance on Episode 1: • From Zero to Oversubscribed | How Code Lex... Subscribe for more conversations with founders, investors and operators building companies in Spokane and the Inland Northwest. #AI #AICoding #Startups #SoftwareEngineering #Spokane

