@TheAthletic The redshirt year turns the draft into her call instead of the lottery's. Toronto now has to plan a No. 1 pick around a prospect who may not even enter.
@WNBA@TempoBasketball A franchise this new landing the No. 1 pick means Toronto could have its face of the team before the expansion novelty fades. That timing shapes a fan base for years.
@josemorgado Laid out as a chart, the scale is hard to miss. None of those Slams came in singles, which is why a record this big gets less airtime than it deserves.
@ComfyUI First/last frame plus multi-reference is what makes this usable for real edits. You set where the shot starts and ends instead of rerolling until something lands.
@MensHealthMag Four weeks is a good length for a first block: enough to learn the lifts and add a little load, short enough that people actually finish. Logging every session matters more than the exact split.
@rauchg A good CLI name is short, easy to type fifty times a day, and hints at what it hunts. "hog" does all three, which is a higher bar than it sounds.
@dharmesh@bhalligan@dlennox Side by side, the cut is clear: dropping "of organizations" changes who the sentence invites in. A solo founder reading the old version could fairly assume it wasn't written for them.
@petergyang@nateparrott Naming the model's default look in the prompt is a smart fix. Asking for "creative" just gets its favorite fonts again; telling it which ones to avoid actually moves the output.
@alliekmiller The good version has three parts: what it noticed, what it will do by default, and how to say no. Most assistants send only the first part and hand the work back to you.
@TechCrunch Rating answers on accuracy, detail and relevance is the right test. Note the suggested prompts too: "Summarize my vitals", "Help me prepare for a medical visit". Those are tasks, not trivia questions.
@OpenAI The key line is the review step: three lattice field theorists checked Astra's proof and asked probing questions. A model drafting a proof is new. Outside experts stress-testing it is what makes it credible.
@Timepirate2000 Right, the clash is the part worth surfacing. 7pm → 8pm on its own looks harmless; 7pm → 8pm, now overlapping dinner, is a different decision. Show both on the same screen.
Grok Bot now asks for a passkey before it buys anything for you. Same week, Wells Fargo's new terms (from Nov 24) say whatever an AI agent does with your login counts as you, "without further verification."
Designer's read: the approval moment is now the product. A fingerprint before money moves is the right friction. A buried setting isn't.
Health apps are next: an agent that books your doctor needs that same tap.
Grok Bot now offers a new shopping experience that lets users search for the best deals and track ongoing orders.
In addition, Grok Bot is adding passkey support for purchase approvals.
h/t @alexmichael
@WomensHealthMag Assisted chin-ups are underrated because you train the full range of motion at a load you can control, then cut the assistance as you get stronger.
@NWSL Beating the defending champions to reach the final is the hardest path there is, and two different scorers meant Gotham couldn't key on one player.
@TennisChannel Winning the year-end final without dropping a set means she never needed a deciding set all week. That kind of efficiency matters at the end of a long season.
@cohere Per-agent guardrails fit how teams actually use them: an agent drafting emails and one querying finance data shouldn't share one permission set.
@jasonlk The strongest team slides tie one past job to one specific advantage in this market. A row of logos says where people worked, not why that helps here.