35,848 results · Most relevant
Coach Kyle advises that in environments like the dog park, gym class, work, social circle events, or house parties, you don't need to display intent on approach. Instead, open indirectly by commenting on the situation or asking a question about the shared environment, e.g. 'Oh, your dog doesn't play? I saw him being really shy over there.'
Coach Kyle provides specific scripts for transitioning from the indirect topic to being about her: asking 'What ethnicity are you, by the way?', saying 'Wait, by chance are you [blank] ethnicity?', complimenting something you notice like 'Wow, I actually really like your tattoo' or 'Actually, can I just say you have really interesting style', or 'Actually, I really like [blank accessory]'.
Coach Kyle describes the closing sequence (step five): suggest meeting up or keeping in touch, get her phone number, figure out when she's free, and invite her out. Scripts include 'We should hang out sometime, let's keep in touch, cool, what's your phone number?', asking 'When are you usually free? Are you free on the weekends or what?', and proposing a concrete timeframe like 'I'll probably be back on, like, Tuesday, Wednesday. So maybe we can do something next week.'
Coach Kyle advises displaying intent (step four) only after you've made it about her a couple of times, done a name introduction, gotten closer in body language/positioning, and she's asked questions. The easiest way is the same as a direct approach: 'Hey, excuse me, I was actually walking by and I just want to say, I thought you looked really cute.' He also gives an alternative: 'Yeah, I actually saw you come into the park earlier and I was like, oh wow, she's actually pretty cute. I should at least go up and say hi, right?'
Take what she said and give personal input on that topic. Even without a shared experience, talk about your own personal experience on the topic. In the demo, Coach Kyle shared that he was the opposite: he got his first tattoo from his artist and didn't want to risk switching, so he got his whole body done by the same artist. Respond with a lengthy, enthusiastic reply while planted with your body to carry the conversation forward.
endorsed
When you give a compliment, don't deliver it while walking away or signal the interaction is ending through your body language. A big reason compliments don't turn into interactions is that your body language and positioning indicate you're not committed.
Alex summarizes the overall plan: structure the webinar, add elements to the conversion process, expand delivery, and blow the door off the front end.
Alex doesn't want the client in the DMs at all — one person can just point people over instead — and suggests hooking up AI to respond to Instagram DMs since it's already linked up.
Hormozi gives a script for receiving disclosures: respond in a reinforcing way — 'Thanks for sharing that. Your secret's safe with me. It's your news. Whatever you want me to do with it, I'll do with it.' He frames the moment of disclosure as a reinforcing event where you should signal their secret is safe, and applies it to scenarios like someone coming out of the closet.
Hormozi recommends that when sharing something material, calculate the trust decision independently each time: weigh what you gain versus what you risk. If the incentive against you is too big, don't share it. He notes you don't need to calculate for every single disclosure, only material ones.
To graduate from the current stage, the practice needs to build out the sales motion discussed, including the VSL and sales ads material. The full funnel is: ads, pitch one, schedule, pitch two, with VSL in between; a V2 test is scan plus VSL, then appointment as the next meeting, with a VSL before the appointment and close.
Hormozi recommends using meta ads as lead gen for the workshop — you'll have to sift through the crap, but you also have the ocean to go after. He recommends charging $99 for the workshop to filter out riffraff and having a card on file, which makes the close easier.
The host diagnoses that not having a defined acquisition process limits a business's ability to scale. He recommends training underutilized salespeople daily to replicate exactly what the owner was doing, and notes that getting the two sales guys productive positions the contestant to win the 90-day scaling competition.
The host suggests seeding multi-event pricing by asking how many events the client runs per year, then offering a better rate for committing to all of them. Anchor high on price, then offer 10% off for committing to all three events, noting the anchored price is already above the client's current price.
On top of traditional SEO — which remains a massive component of GEO — add offsite elements: affiliate mentions and reviews on third-party websites. These new flavors are separate from SEO but part of the same pie.
To mimic Claude Tag, run one agent per Slack channel, with each agent having separate memory and access only to messages within that channel.
Anthropic advised that if your agent needs higher memory or a GPU, you can hook up your Modal instances to run your agent loop inside Modal instead of the provided sandboxes, which today are all the same (no per-agent compute customization). If you care about a different scale of compute, use a different provider for that.
Anthropic advised using the session tab — the session event stream — to understand checkpoints and status of an agent running for hours, noting there is an events API under the session for monitoring long-running agents. They also suggested creating a custom tool called something like 'update status for human' so Claude calls it on intervals whenever it feels it's made progress and gives a status update.
Create a custom tool called 'update status for human' so Claude calls it on intervals whenever it feels it's made progress and gives a status update.
Anthropic advised that at some point you still need to run the thing locally to make sure it works well. They suggested using deployments to auto-deploy something you can test with a custom deploy agent, and if you have different environments for each branch, you can run your application in sandboxes and have Claude self-validate its changes. They also described having Claude act as a QA agent that plays around in the browser, validates changes, and possibly returns a video.
Choose sources → to get a library of your own and a daily brief.