Recommendation · MoneyTo resume agent work, resume from a work log — the talk itself resumes from a work log. On his personal projects the speaker enforces this with a git hook that blocks any commit that doesn't update the work log, so the memory keeps itself.
AI Engineer · Scale the Judgment, Not the Model — Andrew Orobator, Reddit
Recommendation · MoneyThe agent that wrote the code will always tell you that it works, so make it hand back a recording of the thing actually running — producing that recording forces it to make the thing work, and you can't fake a passing run.
AI Engineer · Scale the Judgment, Not the Model — Andrew Orobator, Reddit
Recommendation · MoneyWhen an agent fails, don't just fix the bug — run the postmortem and fold the lesson back into the skill so the failure can't happen twice. This turns failure into a constraint the system carries forward.
AI Engineer · Scale the Judgment, Not the Model — Andrew Orobator, Reddit
Recommendation · BodyDr. Søberg endorses sleeping in a cold room at 19 degrees Celsius to activate and grow brown fat, even while sleeping under a duvet with clothes on.
Andrew Huberman · Essentials: Cold & Heat Exposure for Metabolic Health | Dr. Susanna Søberg
Recommendation · BodySleeping in a 19 degrees Celsius room activates and grows brown fat, and you won't notice it much while sleeping; the cold activation likely comes from cold on the face, which has many cold receptors.
hedged
Andrew Huberman · Essentials: Cold & Heat Exposure for Metabolic Health | Dr. Susanna Søberg
Method · MoneyWhen a failure surfaces: use observability tools that give automatic data on model failures; give users a simple thumbs up/thumbs down to report failures and collect as much data as possible where customers indicate failure; reproduce the issue like classic QA on software and try to understand where the model is failing, whether it is a tool call, and what part of the eval suite failed; then review it and add it to the eval pipeline. Keep this feedback loop going to access new frontiers of things being missed.
Method, not a verified protocol.
AI Engineer · If Your AI Scores 100%, Your Eval Is Wrong - Ramakrishnan Lokanathan, eslams.com
Recommendation · BodyDr. Marius Bakken recommends doing strength work year-round rather than going on and off, because you otherwise get a relative load you are not used to; doing it on the same day as the X session; finishing strength sessions with strides or 45-15 repeats to lower muscle tone; and taking a longer cool down afterwards to help bring tone down.
Garage Strength · The Genius Who Solved Endurance Training w/ Dr. Marius Bakken
Recommendation · BodyBakken recommends going up to altitude for one to four weeks, coming down for about a month, then going up again, using altitude as a periodization tool.
Garage Strength · The Genius Who Solved Endurance Training w/ Dr. Marius Bakken
Recommendation · BodyFor marathon prep, do periodization: from about 6 to 12 weeks (or 4 to 10 weeks) before the marathon, do more marathon-specific work with longer, more durable sessions. One model: run part of the training very easy and then go straight into intervals at the end. Another: 'Italian type' sessions where you go up and down in speed for a long period of time. A third: simply do longer and longer easy running, which can work for quite a few runners. Limit marathon-specific work to a limited period of the marathon-specific weeks only, not year-round. Before the marathon-specific weeks, you can do the math described in the book perfectly fine and build a very good base. During the tapering period, try to bring control into the system and not overdo things. Bakken agrees the marathon periodization.
The pre-marathon window is given as a range ('about 6 to 12 weeks, or 4 to 10 weeks')
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 two models: run part of the training very easy and then go straight into intervals at the end, or simply do longer and longer easy running, which can work for quite a few runners.
Garage Strength · The Genius Who Solved Endurance Training w/ Dr. Marius Bakken
Recommendation · BodyTreat hill sessions as hard X sessions, not threshold sessions. Doing hill sessions as threshold sessions will slow you down instead of getting close to race pace without negative effect. One option is to split a hill session — half in the hills and half on flat terrain — to get both benefits; within the Norwegian system, hill/X sessions are limited to about once a week, so combining flat and hill work is a way to get both stimuli.
Garage Strength · The Genius Who Solved Endurance Training w/ Dr. Marius Bakken
Anecdote · MoneyMasaba Gupta recounts that between 2012 and 2016 she was really going for the creative in her and wasn't compromising. Around 2018–2019 she decided she didn't want to be a small boutique brand but a retail giant with multiple categories, and asked what gear she needed to shift to do it. She notes that you tend to believe every piece of clothing you create is art, but the art has to find a buyer. To find the bridge between commerce and art, she had to drop her creative ego — which she was holding onto dearly — and she spent from 2019 until now doing so. She says it's only for a moment that you have to drop the creative ego, and that she is back to her creative self again now, with an installation at Nilaya in Bombay two days before the talk whose theme was 'to return to thyself is the gre
Figma · Config India 2026: What makes the House of Masaba feel like Masaba
Recommendation · MoneyThe team endorses using majority voting on held-out evaluations to help mitigate variance from individual LLM judgments.
AI Engineer · An Agent That Fixes Its Own Mistakes - Saloni Gandhi & Ziwei Guo, LinkedIn
Recommendation · MoneyDustin Perrier recommends having a next action from every demo, because post-demo is the graveyard for almost everybody, including him. Send a post-demo email the same day with a little area you can personalize, and put demo prospects into a drip campaign instantly because they may be busy and forget how wonderful your product is.
MicroConf · SaaS Sales Demos: Why Getting a No Is Better Than a Yes
Recommendation · MoneyDustin recommends that when a prospect asks a question you can't fully answer in the demo, you acknowledge it and defer the deep answer to a follow-up conversation rather than going deep in the demo. He gives the phrasing 'Kind of, sort of. We could talk about it afterwards.'
MicroConf · SaaS Sales Demos: Why Getting a No Is Better Than a Yes
Recommendation · MoneyDustin Perrier discourages dropping the agenda slide from your demo — most people drop it thinking it's unnecessary, and they're wrong. Use the agenda to lay out what prospects should expect and to make clear you'll have an ask at the end of the demo, giving them clarity about where you're bringing them. He suggests telling prospects exactly what you'll ask at the end, such as asking whether they're ready to sign up for a trial account and to enter a credit card.
MicroConf · SaaS Sales Demos: Why Getting a No Is Better Than a Yes
Recommendation · MoneyDustin Perrier recommends that every time a prospect asks an objection or question — cost, hosting, data — you add it into your slide deck for next time. He argues this improves your credibility, shows you know what you're talking about, and prevents you from guessing answers. He shares that he has been caught flat-footed by questions and had to guess his way through, and it has never worked out for him.
MicroConf · SaaS Sales Demos: Why Getting a No Is Better Than a Yes
Recommendation · MoneyDustin Perrier recommends crafting a problem story that walks prospects through: this is the problem I experienced, this is why I wrote this software, this is what it solved and what it hasn't solved. Don't use a flat problem statement — it's not emotional and won't draw prospects into your presentation. In the discovery phase, present the problem story and ask 'Do you find that this is also true for you?' Then challenge them: 'Can you give me an example of that?' so they must produce evidence they actually experience the problem. Making prospects prove their problem serves two purposes: proving they deserve a yes, and engaging them in the conversation rather than just presenting.
MicroConf · SaaS Sales Demos: Why Getting a No Is Better Than a Yes
Recommendation · MoneyDustin recommends keeping feature demonstrations to 20-40 second blips in the demo. When a prospect asks about features outside your planned demo, he suggests addressing their concern briefly rather than going deep, and going too deep into technical details derails the presentation.
MicroConf · SaaS Sales Demos: Why Getting a No Is Better Than a Yes
Recommendation · MoneyDustin Perrier recommends making demo registration prominent on every page of your website. He observes that people with weak demos hide registration on a contact page or bury it under the fold in the footer, making it incredibly difficult for prospects to register. He reports that over half of the people his peer review group has reviewed had demo registration buried or hidden on their website.
MicroConf · SaaS Sales Demos: Why Getting a No Is Better Than a Yes