Keep on to blur preview images; turn off to show them clearly
My top takeaways from executive coach @RNiederman: 1. The biggest skill gap in new leaders is knowing when to coach vs. when to tell people what to do. When you constantly provide answers, you train your team to bring you every problem instead of building their own problem-solving skills. The people you hire are experts in their domain—ask curious questions to help them reach their own solutions, which makes them more motivated and capable. Save direct advice for urgent situations or when someone genuinely lacks the necessary skills. 2. Use these four questions to coach someone to figure out the answer or themselves: When someone brings you a problem, use GROW: Goal, Reality, Options, and Way forward. Ask about their desired goal (what does success look like?), their current reality (where are you stuck?), possible options for a path forward (what could you do next?), and a concrete way forward (what will you actually do next?). These questions help people discover solutions they already have the context to find. You don’t need to follow this exact order; just use whichever type fits the moment. 3. In conflict, aim for mutual understanding, not proving you’re right. When you enter a difficult conversation trying to convince someone they’re wrong, they become defensive and armor up. Instead, focus on helping the other person understand your experience so they can empathize and see clearly what’s happening. This shift from convincing to connecting creates space for genuine dialogue where both people can be heard and find solutions together. 4. Burnout happens when you spend too much time outside your natural strengths, not just from working too hard. For two weeks, write down the five things each day that energized you most and the five that drained you most. Look for patterns. People burn out not just from working hard but from spending too much time doing things that deplete them—even if they’re good at those things. 5. Aim to spend 80% of your time in work that uses your natural gifts. Everyone will always have 20% of tasks they don’t enjoy, but this ratio creates sustainable energy. Design your role so that 80% plays to your strengths—this might mean delegating differently, hiring to complement your weaknesses, or even moving horizontally in your career rather than always climbing upward. It’s your responsibility to navigate your career toward your gifts, not your manager’s job to make your work interesting. 6. Co-founder relationships need scheduled maintenance time, like marriages. Sixty-five percent of startups fail because of co-founder conflict, not business problems. Set up regular check-ins—weekly touch-bases, monthly lunches, quarterly in-person reviews—to ask: How is this working for you? Are we aligned on vision and strategy? What am I doing that frustrates you? What’s gone unsaid? Without dedicated time, you’ll stay too busy to address problems before they become crises. 7. Use this four-step framework for difficult conversations: Observations, Feelings, Needs, Requests. Start with factual observations anyone could verify (not interpretations). Share your feelings without blame (I felt anxious, confused, disconnected—not “I feel like you. . .”). Name your underlying human needs (clarity, collaboration, connection). Make a small, achievable request the other person can actually fulfill. Stay on your side of the net—talk about your experience, not what you assume about them. This lets you be bold without triggering defensiveness. 8. Always ask yourself how you’re contributing to any conflict, even when the other person seems completely at fault. Leaders in conflict typically fall into three unhelpful patterns: playing victim (this is happening to me), blaming others (this is all their fault), or playing hero (I’ll just do it myself). Instead, ask yourself: How am I complicit in creating conditions I claim I don’t want? Taking responsibility for your part—while maintaining humility and curiosity—opens the door to actual resolution instead of endless frustration. 9. Ask yourself, “Would I enthusiastically rehire this person for the same role?” This binary question immediately reveals whether someone is right for their position. If the answer is no, you need to take action—whether that’s coaching, moving them to a different role, or helping them transition out. Your immediate gut reaction to this question tells you the truth you may be avoiding. If you wouldn’t enthusiastically rehire them, no amount of difficult conversations will likely fix the fundamental mismatch. 10. Active listening takes less than a minute to do but creates powerful change. Level 1 listening is internal listening: listening to your own thoughts. Level 2 is focused listening: you can repeat back what someone said. Level 3 is global listening: you notice body language, tone, and emotions beneath the words and can reflect back insights the person isn’t fully aware of. When you reflect back both what someone said and what you sensed they were feeling, they feel truly seen.
Deeply researched product, growth, and career advice
The @ilyasut episode 0:00:00 – Explaining model jaggedness 0:09:39 - Emotions and value functions 0:18:49 – What are we scaling? 0:25:13 – Why humans generalize better than models 0:35:45 – Straight-shotting superintelligence 0:46:47 – SSI’s model will learn from deployment 0:55:07 – Alignment 1:18:13 – “We are squarely an age of research company” 1:29:23 – Self-play and multi-agent 1:32:42 – Research taste Look up Dwarkesh Podcast on YouTube, Apple Podcasts, or Spotify. Enjoy!
Host of @dwarkeshpodcast https://t.co/3SXlu7fy6N https://t.co/4DPAxODFYi https://t.co/hQfIWdM1Un
The @ilyasut episode 0:00:00 – Explaining model jaggedness 0:09:39 - Emotions and value functions 0:18:49 – What are we scaling? 0:25:13 – Why humans generalize better than models 0:35:45 – Straight-shotting superintelligence 0:46:47 – SSI’s model will learn from deployment 0:55:07 – Alignment 1:18:13 – “We are squarely an age of research company” 1:29:23 – Self-play and multi-agent 1:32:42 – Research taste Look up Dwarkesh Podcast on YouTube, Apple Podcasts, or Spotify. Enjoy!
Host of @dwarkeshpodcast https://t.co/3SXlu7fy6N https://t.co/4DPAxODFYi https://t.co/hQfIWdM1Un
RT @bfl_ml: FLUX.2 is here - our most capable image generation & editing model to date. Multi-reference. 4MP. Production-ready. Open weigh…
Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI builders
Final log output of the NMMO3 SOTA run. Trained on a single 6x4090 tinybox. Unoptimized new model. We should be able to make this 3x faster by next release.
I build sane open-source RL tools. MIT PhD, creator of Neural MMO and founder of PufferAI. DM for business: non-LLM sim engineering, RL R&D, infra & support.
RT @KapilReddy: DSPy Pune Meetup! We have some real-world production experience talks lined up. More talks / lightning talks to be added…
Asst professor @MIT EECS & CSAIL (@nlp_mit). Author of https://t.co/VgyLxl0oa1 and https://t.co/ZZaSzaRaZ7 (@DSPyOSS). Prev: CS PhD @StanfordNLP. Research @Databricks.