Recommendation · BodyBakken advises not doing lactate testing on every session because it is expensive; instead use occasional field tests to cross-check whether training intensity is approximately right. The ear is the most accurate measurement site but impractical; when self-testing without a helper, he used the finger.
Garage Strength · The Genius Who Solved Endurance Training w/ Dr. Marius Bakken
Recommendation · BodyBakken outlines three marathon-specific session models: running part of the training very easy and then going straight into intervals at the end; Italian-type sessions where you go up and down in speed for a long period of time; and simply longer and longer easy running, which can work for quite a few runners.
Commercial context: The creator describes this as their own product.
Garage Strength · The Genius Who Solved Endurance Training w/ Dr. Marius Bakken
Recommendation · BodyBakken endorses that the 45-15 session can be used as an X (event-specific) session for a 5K.
Garage Strength · The Genius Who Solved Endurance Training w/ Dr. Marius Bakken
Recommendation · MoneyThe host endorses negotiating with a give-get framework, where a true commercial give-get means offering something like split billing terms in exchange for signing by a set date.
30 Minutes to President's Club | No-Nonsense Sales · #625: 18 Deal Cheat Codes to Cut Your Sales Cycle in Half
Recommendation · MoneyAt the infrastructure and platform layer, the intelligence plane adds inference, fine-tuning, and continual learning beyond what the control plane's deployment provides.
AI Engineer · Everyone Has the Same AI Models. Here's What Sets Agents Apart - Richmond Alake, Oracle
Recommendation · MoneySupriya Subudhi recommends setting up eval tools to read real AI outputs and see what real users are actually receiving, which reveals whether the prompt is doing a good job.
AI Engineer · 78% Wanted This AI Feature. Almost Nobody Used It - Supriya Subudhi, Bumble
Recommendation · MoneySupriya Subudhi recommends changing AI success criteria from overall green-versus-red results to examining the long tail of data and how different segments of users receive feedback — e.g., whether women respond the same way as men, and whether users in different countries and languages respond the same way.
AI Engineer · 78% Wanted This AI Feature. Almost Nobody Used It - Supriya Subudhi, Bumble
Recommendation · MoneyThe guest recommends requiring human approval for high-risk agent actions, for example any download or upload, or any visit to a new website.
Also discussed by Javier Garza on AI Engineer (unclassified source — not counted until classified). Not counted as independent sources.
The a16z Show · OpenAI: When AI Agents Use Computers Faster Than You
Recommendation · MoneyThe guest recommends adopting WebMCP as a standard so humans and agents can use a site together — the agent does the tedious work while the person browses and finishes transactions like entering payment details, carving out an easier experience for sensitive steps.
Also discussed by Shubhankar Srivastava on AI Engineer (unclassified source — not counted until classified). Not counted as independent sources.
The a16z Show · OpenAI: When AI Agents Use Computers Faster Than You
Recommendation · MoneyThe guest recommends configuring agent sandboxes by explicitly whitelisting which apps and websites the agent may use and blocking the rest.
Also discussed by Anthropic on AI Engineer (unclassified source — not counted until classified). Not counted as independent sources.
The a16z Show · OpenAI: When AI Agents Use Computers Faster Than You
Recommendation · MoneyMasaba Gupta advises naming a brand with a story that flips a negative into a positive, so that people remember the brand far beyond the founder. When consumers ask 'why did you name it Love Child?', there is a story ready — she designs brand names to invite that question.
Figma · Config India 2026: What makes the House of Masaba feel like Masaba
Recommendation · MoneyIf you are trying to get better productivity across your team, Zach Lloyd says you want a view of what is happening from a productivity standpoint, including how automated your factory is. Track cycle time with a DORA-metric-style view of your software factory, and track spending alongside productivity metrics. He recommends tracking all data on what your team is doing, including productivity, automation level, cycle time (Dora metrics), and spending. Warp gives people tools to measure cost per PR and what percent of cost goes toward scoring or benchmarking, and you can extrapolate from cost-per-PR measurements to decide how much you're willing to spend on a project.
AI Engineer · How to Build a Software Factory That Improves Itself - Zach Lloyd, Warp
Recommendation · MoneyZach Lloyd describes using an orchestrator agent that figures out whether work should go to a triage step, a spec step, an implementation step, or code review, and recommends including orchestration in the infrastructure so multi-step workflows are possible.
AI Engineer · How to Build a Software Factory That Improves Itself - Zach Lloyd, Warp
Recommendation · MoneyZach Lloyd advises spec-driven development so changes can be tested for adherence to a spec, and computer use verification where agents verify or produce evidence from the user standpoint that the thing works. He also describes agents reading matching repository skills before implementing.
AI Engineer · How to Build a Software Factory That Improves Itself - Zach Lloyd, Warp
Recommendation · MoneyZach Lloyd recommends measuring cost per PR and what percent of cost goes toward scoring or benchmarking in order to plan project budgets, framing the questions as 'What's your cost per PR?', 'What percent of the cost is going towards scoring or benchmarking?', and 'Are you willing to spend X amount on a project?'. He also says you want productivity, cycle time, and spend data tracked across the team, since without it you cannot measure or improve how automated the factory is.
AI Engineer · How to Build a Software Factory That Improves Itself - Zach Lloyd, Warp
Recommendation · MoneyWhen building an RSI loop, optimize the planning and logic component that orchestrates execution, the context and memory component that preserves session state, the tool interface that interacts with external and internal APIs and datasets, and the artifact storage where outputs and generation logs are stored.
AI Engineer · An Agent That Fixes Its Own Mistakes - Saloni Gandhi & Ziwei Guo, LinkedIn
Recommendation · MoneyDaksh Gupta recommends documenting all the flows through your application as a prerequisite for doing end-to-end QA.
AI Engineer · How to Build an AI Code Review Agent - Daksh Gupta, Greptile
Recommendation · MoneyDaksh Gupta recommends building an internal knowledge base that catalogs all parts of the codebase and indexes how each part works, including flows, diagrams, and every file involved.
AI Engineer · How to Build an AI Code Review Agent - Daksh Gupta, Greptile
Recommendation · MoneyDaksh Gupta recommends combating overactive file reads using a cheap model that looks through all of your context pieces and figures out which ones actually should be included.
AI Engineer · How to Build an AI Code Review Agent - Daksh Gupta, Greptile
Recommendation · MoneyDaksh Gupta recommends maintaining a sufficiently simple build script so the agent can spin up your stack in your environments.
AI Engineer · How to Build an AI Code Review Agent - Daksh Gupta, Greptile