Get the number and lock in plans mid-interaction
Set up the idea of hanging out in the future during the interaction, then get her phone number and lock in the plans in the middle of the interaction, not after it ends.
40,424 results · Most relevant
Set up the idea of hanging out in the future during the interaction, then get her phone number and lock in the plans in the middle of the interaction, not after it ends.
Charlie Guo recommends embedding Codex into your product using AppServer. This lets your users use their ChatGPT subscription and token budget to power your product. He says the AppServer is a useful way to integrate into your own platforms if you want to build agents on the cloud, manage different agent providers, or build a unified layer in a large organization.
The speaker advised defining requirements — what a good report looks like for your agent — before categorizing failures, noting that many teams instead ship an agent and react to complaints without ever defining the bar up front.
When facing existential threats from AI-native startups and large incumbents, a company must pivot to being AI-native rather than gradually layering in AI features. Ignore industry experts who claim customers fear new technology; validate directly with customers instead. Push through internal and external resistance, neglect other burning fires, and go all-in on AI.
Consequential creator advice requires independent verification before Today can promote it.
To benefit from AI agents, build an extremely opinionated product that enforces a specific system and best practices. This creates a unique data advantage and a moat that generic AI models like ChatGPT or Claude cannot easily replicate. The software should drive the outcome for the user without manual intervention.
Derrick Reimer recommends hiring a developer with at least five years of experience, a full-stack generalist skill set, and a bias toward action and shipping. Avoid specialists and candidates who have only worked at large companies.
confident
Derrick Reimer advises hiring a developer who thrives working autonomously, is flexible about codebase conventions, willing to let the founder make final calls, and open to using modern tools like AI coding assistants. Avoid candidates who are too opinionated or fight over small conventions.
confident
Erik Allebest recommends relentlessly focusing on user experience above all else, especially in a market without patent protection. He states that product, community, and content are the only durable moats, and that taking care of the community and product builds a great business.
Intent is a red herring; especially early in a relationship, do not judge someone's intent. Focus on whether the agreement was met, as why someone breaks an agreement is less important than that they broke it.
Taking full responsibility for a relationship is toxic. Instead, take 100% responsibility for your side: be honest, clear, and communicative, and bring your side of the deal. Then, only check if the agreed-upon thing happened or if there is no deal.
Hormozi advises to continue allocating resources towards the things that are getting the highest return, rather than pursuing new, uncertain opportunities. He frames this as prioritizing actions that are certain to produce a return over those that might. He also warns that proposing new initiatives is often a sign of emotional discomfort with the current situation, which leads to distraction from what actually works.
In the buy-versus-build tradeoff, the question to ask is what you want your limited set of high-quality engineers working on. It is not support cases, and it is not things that are already solved. Rebuilding the wheel on already-solved problems like content extraction does not make sense. Having internal teams iterate on existing solutions and deal with the support burden of those is not the best use of talent.
Keep humans in the loop for protected calls even as the system gets lighter — never go fully human-free. Building a review system that goes fully human-free loses the learning signal from human corrections. When a review system gets lighter over time, verify it has not gone human-free on protected calls.
When a human corrects the agent, use that correction as feedback to improve its next review. A human correction like the encrypted-column case is exactly the signal that teaches the rules to get smarter — the next time a similar column comes through, that context can stop a needless escalation. Design the review system so it learns from human corrections rather than operating statically; without the feedback, every rule change would be a guess.
Elena Verna believes that if we truly believe in this AI shift, ICs should get paid more than managers, because an IC can drive more outcome and impact than a manager in the old system. Otherwise, everybody rationally becomes a manager and there are managers everywhere. Money and status for ICs has to be equal to managers, because most people go into management because they still want to get paid. If you want to attract HI-ICs, expect to pay them leadership-level salaries — but also expect leadership-level outcomes from them.
In the age of AI, the builder or designer's first roles are to sense the world with their body, frame the questions before generating, and bring their understanding of the situation to AI by explaining what matters and what does not.
hedged
Go deep into details, moving from sentences and images to something you can test and into actual experiences rather than staying at the prompt level, because as soon as you make something concrete you discover how much you have not yet decided — one decision reveals another question.
After building, observe and listen to whether people can use it, what they feel, and what surprises them; then decide with courage, because without a decision you cannot move forward; and repeat the cycle of sensing, framing, generating, testing, understanding, and deciding, because output brings you back to the world with new knowledge.
Kevin Hou advises builders to consider what primitives are in their product and how they might scale with the model's intelligence, and to prioritize features that let the product scale with the next model release.
hedged
Commercial context: The creator describes this as their own product.
Having an expert human in the loop is very important when building AI agents in specialized domains. A subject matter expert can check the assumptions, inputs, and outputs, and make the right corrections. Even when most checks are hard-coded, a human check with professional judgment is still needed. You can also codify the domain expert's checks directly into the model to make it more autonomous.
Choose sources → to get a library of your own and a daily brief.