Report · BodyIn a 2017 study by Hansen et al., a group of subjects slept in a room at 24 degrees Celsius, then had PET-CT scanning to establish baseline brown fat levels. After one month of sleeping at 19 degrees Celsius — under a duvet and with clothes on — the subjects had increased insulin sensitivity, which the speaker finds remarkable. In the second month, when subjects slept at 24 degrees Celsius, their insulin sensitivity decreased a little bit. In the fourth month, when subjects slept at 27 degrees Celsius — a quite warm room — there was even less activation of brown fat and also less insulin sensitivity.
Andrew Huberman · Essentials: Cold & Heat Exposure for Metabolic Health | Dr. Susanna Søberg
Report · BodyRepeated cold exposure increases brown fat activation and the number and efficiency of mitochondria in brown fat cells, because the body expects you to do it again. With repeated cold exposure, the capillaries in your skin become better at constricting, giving you a better protective shield for the next time. With adaptation, your stress response to cold subsides — you get a smaller increase in catecholamines. Activation of brown fat and muscles increases metabolism.
Andrew Huberman · Essentials: Cold & Heat Exposure for Metabolic Health | Dr. Susanna Søberg
Report · BodyThe 'after drop' is when core temperature decreases even after you get out of the cold water — and it always happens. When you enter cold water, all blood vessels constrict to keep blood in your core and keep vital organs warm. After exiting cold water, blood vessels reopen, warm blood flows out to the cold surface, gets colder, and flows back into the core — decreasing core temperature. The after drop is what induces shivering, and shivering then activates brown fat, which warms the body up again.
Andrew Huberman · Essentials: Cold & Heat Exposure for Metabolic Health | Dr. Susanna Søberg
Creator perspective · MoneyAn agent architecture should include an agent runtime, a memory and knowledge layer, a memory manager sitting on top of a data platform, and a continual learning loop that feeds model updates back into the agent runtime.
AI Engineer · Everyone Has the Same AI Models. Here's What Sets Agents Apart - Richmond Alake, Oracle
Creator perspective · MoneyGEPA's Pareto-based candidate selection maintains a much more balanced search process, eventually converging to a much higher score than an LLM-in-a-loop approach, with almost twice the performance gains and more than half of GEPA's total gains attributable to this strategy.
AI Engineer · Beating RL With Reflection: GEPA and Optimize Anything — Lakshya A. Agrawal, GEPA
Creator perspective · MoneyGEPA keeps a Pareto pool where it keeps every single candidate that wins on even one training example, not just the top scorer.
AI Engineer · Beating RL With Reflection: GEPA and Optimize Anything — Lakshya A. Agrawal, GEPA
Report · MoneyOptimizeAnything is a universal API for optimizing any text parameter. To use it, provide the set of problems you care to be solved along with an evaluator (fitness) function that returns a score plus any available domain-specific side information. The evaluator's side-information return value is an open-ended dictionary that can contain literally anything. Call OptimizeAnything with a fitness function and a set of problems, and it returns an optimized solution.
AI Engineer · Beating RL With Reflection: GEPA and Optimize Anything — Lakshya A. Agrawal, GEPA
Principle · MoneyRoman Ugarte says when the model gets smart enough to do something natively, delete the scaffolding built around it; adding features instead of deleting old scaffolding bloats the product, and hacky pro tips and crazy knobs make it confusing.
Lenny's Podcast · How we built Grok Bot in a month | Roman Ugarte (SpaceXAI)
Report · MoneyA task is essentially a unit test for your model, described as YAML files. In ARIA's tasks, an LLM judge determines what correctness means in the context of the question, a second LLM judge determines if the insights are actually interesting, and a rule-based judge checks if the agent was able to generate a result within a set number of tool calls, e.g., six, for expediency. About 200 tasks are clustered together into an eval suite that runs nightly, tracked in Weave with a listing of every night's eval.
AI Engineer · An AI Research Agent That Runs Your Experiments — Tim Sweeney, Weights & Biases
Report · MoneyThe team's evaluation results are emitted to Weave, where they have a common dashboard for making go, no-go decisions on prompt or architectural changes.
AI Engineer · An AI Research Agent That Runs Your Experiments — Tim Sweeney, Weights & Biases
Report · MoneyThe team's offline research loop involves forming hypotheses, implementing candidate agents, and analyzing evals, with data fed from Weave.
AI Engineer · An AI Research Agent That Runs Your Experiments — Tim Sweeney, Weights & Biases
Report · MoneyThe team runs a live production loop where they trace production logs, review them as humans, use LLMs to complement that analysis, and transform findings into tasks.
AI Engineer · An AI Research Agent That Runs Your Experiments — Tim Sweeney, Weights & Biases
Principle · MoneyJemima Abu recommends telling your agent to default to the best way of writing things rather than the most popular way, and including this instruction in your skill — for example, custom JavaScript accordions are the popular approach but not the best approach.
AI Engineer · Teach Your AI Agent to Build Products Everyone Can Use - Jemima Abu, CAIS
Creator perspective · MoneyJemima Abu discourages throwing an accessibility overlay on a website and acting like that's the fix-all; she says she hates the stick-person overlay widgets in the corner of websites and calls them terrible. Shoving a bunch of ARIA labels into a completed product and calling it an accessible website also does not work.
AI Engineer · Teach Your AI Agent to Build Products Everyone Can Use - Jemima Abu, CAIS
Creator perspective · MoneySeveral anti-patterns emerged: trusting a good-looking AI forecast result (like 4% error) without expert review because errors can cancel out; relying on AI-extracted numbers from filings without verifying them against disclosed figures; applying a growth rate to the wrong population base (US-Canada increase applied to the whole international membership); and accepting AI-invented assumption ranges (5–30%) that miss the real disclosed figure (45%).
AI Engineer · Right Answer, Wrong Reasons: AI Forecasts Costco - Hassan El Mghari & Omar El Mghari, Together AI
ReportEvoUndo captures a pre-state witness before applying a mutation, which provides the information to restore the original state and later verify that recovery actually worked.
AI Engineer · Your Coding Agent Can't Undo Its Own Mistakes - Tanmay Sah & Dolly Sah
Creator perspective · MoneyOnce the model has read the pointers, it structurally cannot generate numbers — unless it dereferences a pointer, that number or fact cannot appear in Kepler's output. If a model cannot de-resolve a pointer, it cannot print that number out. If you grab something out of a hash map that doesn't exist, you don't get anything back.
AI Engineer · AI Can't Make Up a Number It Can't Look Up - Vinoo Ganesh, Kepler
Creator perspective · MoneyThe source argues that reaching customers who aren't searching requires a different medium: short-form video (TikTok, Reels, Shorts) serving as a discovery channel at the top of the funnel where strangers find you before searching your name.
Ahrefs · 7 Ways to Get Website Traffic, Explained in 10 Minutes
Report · MoneyIf you only do OCR on a figure and want to know a value like Intel's revenue in Q2 2025, you lose all information and will never really be able to match the page. Instead, create an embedding on the visual representation — let the model look at the page screenshot and create the embedding based on that — to get rid of the lossy OCR-based pipeline.
AI Engineer · Better Search Took This AI Agent From 36% to 98% Accuracy - Aamir Shakir, Mixedbread
Principle · MoneyRequests per second (demand) is the bottleneck for GPU latency, and time-to-first-token is what to optimize for. Throughput matters less than time-to-first-token: 200 tokens per second is enough for interactivity. As soon as time-to-first-token exceeds five seconds, user complaints blow up — that's the empirical limit. The aim is sub-three-second time-to-first-token.
AI Engineer · How Revolut Cut AI Coding Costs 90% With Self-Hosted Open Models - David Abitbol, Revolut