The Startup Podcast

The traditional GitHub-style pull request review is a waste of opportunity cost for most code changes in the agentic era. Yet simply removing it without a replacement invites complexity death spirals and unbounded product behaviour that agents cannot unwind. James McGill and Haziq Nordin join Yaniv to argue that engineers should stop reading diffs line-by-line and instead scan for architectural complexity, shift accountability to spec quality, and build new verification patchworks of end-to-end tests, fuzzing, and product review. The conversation is for any engineer or founder trying to extract real leverage from AI coding tools rather than bottlenecking them with cargo-cult process.Key takeawaysLine-by-line diff review is dead for most changes; the value now lies in catching complexity death spirals that agents propagate across prompts.Data mutability, not code mutability, is the hard constraint—stateless frontend changes need far less scrutiny than anything touching persistent state.The engineer's role is shifting from writing code to steering agents, writing specs, and verifying behaviour through tests rather than inspecting implementation.Accountability structures must change too: if organisations remain blame-distribution systems, they will hedge to maximum review and forfeit AI leverage.Compliance inertia (SOC 2, ISO) still mandates 'thorough review', creating first-mover risk for teams redefining what review means.Codebase malleability is higher than ever, but unbounded complexity accumulates as context for future prompts and eventually collapses agent reliability.Chapters00:00 Introduction00:38 The dead and the messy03:28 Projector reviews to GitHub05:31 What code review was for07:09 Complexity as the last reason to read diffs10:21 Code malleability, data hardness15:51 Code review as accountability mechanism18:50 Product review over code review25:39 LLMs as spec compilers32:48 Verification patchwork36:40 Accountability and leverage44:11 Where to find James and HaziqLinks and tools mentionedGitHub: https://github.comEnhance Labs: https://www.enhancelabs.comActuals: https://www.actuals.aiAbout James McGill & Haziq NordinJames McGill is a founding engineer at Actuals, building AI-native finance tools; Haziq Nordin is founder and CTO at Enhance Labs, working on product guardrails for agent-generated code.This episode is presented by SparkLaunch and Dam Secure.SparkLaunch (https://sparklaunch.io/wizardry): Applied Wizardry listeners get an extra 20% off.Dam Secure (https://damsecure.ai/wizardry): Applied Wizardry listeners get their first month free.CreditsJustin McArthur, Executive Producer and Video Editor: https://www.linkedin.com/in/justin-mcarthur/Steph Hefferan, Editor: https://www.linkedin.com/in/steph-heff/

Companies now use an average of 12 models per month, yet the biggest hidden variable isn't which model you pick—it's the inference provider serving it and whether your cache hit rate stays above 85%. Peter Walker from OpenRouter explains how the same open-weight model can have wildly different cost and performance depending on the provider's caching policy, latency tail, and whether a stealth provider is reselling your prompts. The conversation covers why agentic workloads broke cache economics in February 2026, how ZDR policies vary even among "zero data retention" claims, and why the frontier labs are now competing across price and intelligence spectrums simultaneously. For engineers choosing production models, this is a practical map of the dimensions that actually move the bill.The inference provider layer—not just the model layer—now has 200+ competitors, and provider choice directly determines effective cost, latency, and data sovereignty.Cache hit rate is the dominant cost lever for agentic workloads: a cache miss can cost 50-100x more per token, and median agentic cache rates have climbed from ~50% to ~85% in 2026.Most organisations use a 'hub and spoke' model architecture with one 'daily driver' taking the majority of tokens and spend, plus specialised models for departments like customer support or sales.Zero data retention claims vary in trustworthiness by jurisdiction and provider size; some stealth models on OpenRouter are free specifically because they are not ZDR.The gap between frontier and open-weight models is widening in raw capability but narrowing in perceived value, as daily tasks saturate below frontier intelligence.Reasoning-only models and jagged intelligence are driving demand for fast decision models like Jev and OpenAI's decision model for latency-critical paths.00:00 Introduction00:38 The New Complexity04:07 Provider Layer Competition06:09 Choosing a Provider08:42 Data Retention Realities12:39 Model Portfolios15:23 Latency and Reasoning18:29 Cache Economics23:35 Cost-Performance Sweet Spot26:24 What Comes Next29:13 Open Weight vs Frontier32:02 Specialisation and Product34:14 Where to Follow PeterOpenRouter: https://openrouter.aiOpenRouter Data: https://openrouter.ai/dataDeepSeek: https://www.deepseek.comAnthropic: https://www.anthropic.comOpenAI: https://openai.comGoogle AI: https://ai.googleCerebras: https://www.cerebras.netCoreWeave: https://www.coreweave.comElevenLabs: https://elevenlabs.ioArtificial Analysis: https://artificialanalysis.aiGroq: https://groq.comPeter Walker is Head of Insights at OpenRouter, where he analyses global model usage patterns across the inference layer.This episode is presented by SparkLaunch and Dam Secure.SparkLaunch (https://sparklaunch.io/wizardry): Applied Wizardry listeners get an extra 20% off.Dam Secure (https://damsecure.ai/wizardry): Applied Wizardry listeners get their first month free.Justin McArthur, Executive Producer and Video Editor: https://www.linkedin.com/in/justin-mcarthur/Steph Hefferan, Editor: https://www.linkedin.com/in/steph-heff/

Follow Applied Wizardry on its new feed to keep getting every episode.Dave Slutzkin, co-founder of Antiburn, joins Yaniv to break down what actually drives token costs in agentic development and how to bring them down. They cover how caching differs between Anthropic and OpenAI, when to reach for which model and thinking mode, and why the habit of compaction matters more than any other optimisation. If you're spending real money on AI-assisted development, you'll come away with concrete tactics for cutting token burn without cutting capability.Key takeawaysContext window depth is the dominant driver of token cost; compacting actively at 200,000-300,000 tokens matters more than any other single action.Cache expiry behaviour differs materially between providers: Anthropic's cache expires at exactly 1 hour, OpenAI's Codex cache is best-effort and often drops in 5-20 minutes, but OpenAI subscriptions do not charge for cache writes while Anthropic subscriptions effectively do through quota.Using a frontier model with a simple prompt to delegate to cheaper subagents is a practical pattern that balances capability and cost without requiring complex harness engineering.High thinking modes often produce worse results than medium or low because models overthink, invent requirements, and fix unrelated bugs; the benchmark sweet spot is typically medium.Skills files, MCPs, auto-memories, and system prompts silently bloat context windows; regular clean-up every month or two reduces both cost and quality degradation.Fast mode costs 2.5x per token but can improve output quality through tighter interaction loops; it is best used deliberately rather than left on by default.Chapters00:00 Introduction00:38 The Token Burn Problem03:12 Three Stages of Adoption07:03 Model Choices Multiplying11:14 The Cardinal Mistake: Depth18:13 Worst-Case Usage Patterns22:13 Picking Models by Feel30:23 Thinking Modes and Overthinking35:23 Fast Mode Trade-offs48:46 Caching: Anthropic vs OpenAI52:29 Context Window Ceiling56:32 Cleaning Up Context Bloat56:46 Where to Find DaveLinks and tools mentionedAntiburn: https://antiburn.comAnthropic Claude: https://www.anthropic.comOpenAI Codex: https://openai.com/codexHarvey: https://www.harvey.aiLegora: https://legora.comAbout Dave SlutzkinCo-founder of Antiburn, a company that helps developers understand and optimise their token spend for AI-assisted coding.This episode is presented by SparkLaunch and Dam Secure.SparkLaunch (https://sparklaunch.io/wizardry): Applied Wizardry listeners get an extra 20% off.Dam Secure (https://damsecure.ai/wizardry): Applied Wizardry listeners get their first month free.CreditsJustin McArthur, Executive Producer and Video Editor: https://www.linkedin.com/in/justin-mcarthur/Steph Hefferan, Editor: https://www.linkedin.com/in/steph-heff/

After nearly 300 episodes of The Startup Podcast, Chris Saad is bringing everything he's learned about building world-class products and product strategy into a single tool: ScaleStudio. In what may be our final episode (unless a white knight sponsor appears), Chris walks Yaniv through the product that could fundamentally change how founders approach strategy; from problem statement all the way to positioning, roadmap sequencing, and org design.Chris demonstrates how this one AI tool, housing his vast knowledge of Silicon Valley startup strategy, can give an overwhelming process a navigable structure, allowing founders to plot a deliberate course through the chaos.In this episode, you will:Discover how Scale Studio turns Chris's 30 years of product strategy expertise into an AI tool that guides you from problem statement through to roadmap and positioningSee the complete strategy development workflow in action: from mapping all possible problem domains and solutions, to building the org structure required to executeLearn why mapping the entire solution space upfront prevents the "thrash" that derails most startups when customer requests pull the team in conflicting directionsUnderstand the customer terrain framework that reveals which customer segments are easy wins and which are death traps, allowing you to carve the narrowest viable wedge into your marketSee how Scale Studio integrates with your development workflow through exports and MCP connections, turning high-level strategy into actionable execution roadmapsExplore how AI can now handle the heavy lifting of strategy generation and iteration in around an hour, while keeping guardrails in place to maintain quality and coherence across every sectionTimestamps00:14 On Today's Show: The Last Startup Podcast?01:39 Turning Our Advice Into A Product (ScaleStudio)03:18 Homepage And Templates04:40 Problem Statement05:36 Copilot And Collaboration08:10 Solutions And Editing09:49 'How It Works' Stage11:22 Building Strategy From Seed13:32 Mapping The Solution Space18:53 Customer Segments and Personas20:46 Wedges And Customer Terrain23:07 Customer Roadmap View26:32 Org Design28:03 How Does AI Change Hiring?31:18 Roadmap To Unlocking Cohorts33:51 Market Sizing37:25 Planning the Future of ScaleStudio… in ScaleStudio?39:08 The Future of Agentic Orgs42:11 Closing Thoughts: 4 Years of TSPResources mentioned in this episodeScaleStudio (Chris’s AI-powered product strategy tool): https://scalestudio.io/Vera (Yaniv's Startup): https://vera.guide The PactHonor the Startup Podcast Pact! If you have listened to TSP and gotten value from it, please:Follow, rate, and review us in your listening appSecure your official TSP merchandise at https://shop.tsp.show/Follow us on YouTube for full video episodes: https://www.youtube.com/@startup-podcast Give us a public shout-out on LinkedIn or anywhere you have a social media followingKey linksThis episode of the Startup Podcast is sponsored by .tech domains. Forget weird prefixes and creative misspellings; the availability for .tech domains is simply way better than .com. For a clean name that highlights your tech credentials, get a .tech domain at your favorite registrar.The Startup Podcast website: https://www.tsp.show/episodes/Learn more about Chris and YanivWork 1:1 with Chris: http://chrissaad.com/advisory/Follow Chris on Linkedin: https://www.linkedin.com/in/chrissaad/Follow Yaniv on Linkedin: https://www.linkedin.com/in/ybernstein/Producer: Justin McArthur https://www.linkedin.com/in/justin-mcarthurAssistant Producer: Steph Hefferan https://www.linkedin.com/in/steph-heff/Intro Voice: Jeremiah Owyang https://web-strategist.com/

A few weeks ago, Rob Snyder came on the pod to introduce his PULL framework for identifying real product-market fit. Now, Yaniv turns the tables and subjects his own startup, Vera, to the same scrutiny. Yaniv and Rob strip away the noise around Vera's mission to help adult children care for aging parents and get brutally honest about where the product actually has pull, where it doesn't, and what needs to change immediately. They identify a specific, unavoidable moment in the customer journey and pinpoint that is where the real business begins, and discuss more macro issues, such as AI vs. human expertise, founder blind spots, and more. In this episode, you will:Understand the difference between an aspiration and a project, and why this distinction is vital for PULLLearn how the PULL framework cuts through startup theory to identify the exact moment when customers will actually pull your product from your hands, before they have to wait for a crisisDiscover why Vera's original problem was too abstract, and how narrowing to a specific market segment with intense, unavoidable need becomes your path to revenue and growthHear the truth behind what really pushes your customers to adopt earlier: reality does the convincing, not your marketingUnderstand the challenge of reputation in AI: how do human-shaped applications build trust the way human experts do?Timestamps00:00 Coming Up01:38 On Today's Show: Yaniv Applies PULL to Vera04:42 Defining PULL and Projects05:54 Vera's Customer Goals08:00 Unavoidable Moments and Crisis11:25 Real Users and Traction16:58 Mission vs Push Marketing24:07 Why Isn't Google Enough?28:21 Nailing the Specific Project34:02 Testing Market With Ads37:06 Bill Gates And AI Shift38:47 Human Shaped Software Design43:45 Trust And Reputation In Bots48:48 About The BookResources mentioned in this episodeThe Power of Pull by Rob Snyder: https://www.amazon.com/Power-Pull-Customer-Successful-Founders/dp/1541705955Rob Snyder's website: https://www.robsnyder.orgVera (Yaniv's startup helping families care for aging parents): https://vera.guideClub Sandwich, Vera's podcast and community for the sandwich generation: https://clubsandwich.communityWhat the sandwich generation is (and why it's Vera's market): https://clubsandwich.community/sandwich-generationThe PactHonor the Startup Podcast Pact! If you have listened to TSP and gotten value from it, please:Follow, rate, and review us in your listening appSecure your official TSP merchandise at https://shop.tsp.show/Follow us on YouTube for full video episodes: https://www.youtube.com/@startup-podcast Give us a public shout-out on LinkedIn or anywhere you have a social media followingKey linksThis episode of the Startup Podcast is sponsored by .tech domains. Forget weird prefixes and creative misspellings; the availability for .tech domains is simply way better than .com. For a clean and memorable name, go to https://⁠get.tech/tspThis episode of the Startup Podcast is sponsored by Vanta. Vanta helps businesses get and stay compliant by automating up to 90% of the work for the most in demand compliance frameworks. With over 200 integrations, you can easily monitor and secure the tools your business relies on. For a limited time offer of US$1,000 off, go to ⁠⁠⁠⁠https://⁠www.vanta.com/tsp⁠⁠⁠⁠⁠The Startup Podcast website: https://www.tsp.show/episodes/Learn more about Chris and YanivWork 1:1 with Chris: http://chrissaad.com/advisory/Follow Chris on Linkedin: https://www.linkedin.com/in/chrissaad/Follow Yaniv on Linkedin: https://www.linkedin.com/in/ybernstein/Producer: Justin McArthur https://www.linkedin.com/in/justin-mcarthurAssistant Producer: Steph Hefferan https://www.linkedin.com/in/steph-heff/Intro Voice: Jeremiah Owyang https://web-strategist.com/