Recommendation · MoneyThe source recommends generating cash ethically within the current season using a cash-forward offer, described as the best shot at solving the cash problem in that timeframe. The Money Models book is suggested as a resource giving a couple of different angles to hit the cash-generation problem from. The source also expresses confidence that the offer will work even with the team's capability uncertain.
creator confidence, not a guarantee
Alex Hormozi · Can I Save This Failing Theme Park in 90 Days?
Recommendation · MoneyAlex recommends using Unspeakable video clips as the primary creative for the paid ads; experimenting with a buy four, get six weeks free offer in ads — it might destroy; using 'as low as $200 per week' ($2,000 divided by 10) as another ad angle; and testing multiple front-end offer framings that all mean the same thing — buy four get six, buy one month get whole summer, buy one month get two free — because one of them will convert best. He also says to pair the offer with an 'early bird' angle and to always answer the reason why you're making an offer.
Alex Hormozi · Can I Save This Failing Theme Park in 90 Days?
Recommendation · MoneyWhen driving traffic to an offer, use four flows: email list, general admission attach rates, paid ads direct, and brick-and-mortar.
Alex Hormozi · Can I Save This Failing Theme Park in 90 Days?
Recommendation · MoneyIn the giveaway, when a lead refers the winner, the referrer also wins the prize. This works because parents love when their kids go with their friends' kids. When announcing the winner, call them and let them tell the referrer the good news on a three-way call.
Alex Hormozi · Can I Save This Failing Theme Park in 90 Days?
RecommendationDrawing on the LeBenz story, Karan Singhal's takeaway is that when AI repeatedly tells you to take a symptom seriously, you should — seek testing rather than dismissing it.
The a16z Show · What Happens When Everyone Has a Health Expert in Their Pocket? | OpenAI’s Karan Singhal
Recommendation · MoneyMake sure the agent doesn't try to give investment advice or answer questions like the weather in New York when they're out of scope; datasets should include out-of-scope requests that should be correctly deferred.
AI Engineer · Evals for High-Stakes AI Agents — Doneyli De Jesus & Annabell Schäfer, Langfuse
Recommendation · MoneyBuild a dashboard of business-facing metrics from agent data to reassure stakeholders the agentic solution benefits the business: track thumbs up/down customer feedback in an easy-to-consume dashboard, focus investigation on the thumbs-down minority rather than celebrating the majority, and report on tokenomics for cost-focused stakeholders.
AI Engineer · Evals for High-Stakes AI Agents — Doneyli De Jesus & Annabell Schäfer, Langfuse
Recommendation · MoneyFrom production traces, track aggregate metrics like cost, latency, and tokens, and monitor user frustration signals such as how many users write all-caps messages because they're upset. Compliance guardrails can also serve as quality signals in monitoring.
AI Engineer · Evals for High-Stakes AI Agents — Doneyli De Jesus & Annabell Schäfer, Langfuse
Recommendation · MoneyMonitor for security-risk flags where the agent might have given investment advice to customers, and investigate those specific cases by filtering traces on the categorical score condition.
AI Engineer · Evals for High-Stakes AI Agents — Doneyli De Jesus & Annabell Schäfer, Langfuse
Recommendation · MoneyThe host recommends checking who viewed your LinkedIn profile every week as a warm-lead source, since LinkedIn is the only platform that shows profile viewers and a view signals interest in your content or outreach.
30 Minutes to President's Club | No-Nonsense Sales · #616: How I Use LinkedIn Sales Navigator to Find Warm Leads (Crash Course)
Recommendation · MoneyAsk the buyer if this is their first time presenting a deal of this kind to the board and what pushback or questions usually come up, so you can help them prepare. If it's their first time and they're presenting rather than asking or telling, expect a different (harder) outcome for your deal than you anticipated.
30 Minutes to President's Club | No-Nonsense Sales · #613: Master Parallel Selling Strategies to Speed Up Every Deal | Sam Mckenna | 30MPC Hall of Fame
Recommendation · MoneyIf you have a feature you really care about — like hiring — make sure you find the right signals by tuning the system to that feature, e.g., analyzing earnings calls, transcripts, or podcast interviews with the CEO; then check whether that would have actually improved the prediction.
AI Engineer · Backtest Your Agent Against Yesterday's Web — Ryan Ahern, Exa
Recommendation · MoneyUse the OR combiner so either feature triggering yes counts as predicting acquisition; switch to AND when you need to be more constraining — both features must be true. Test both and compare scores; if AND improves the score, both features need to be true to be predictive.
AI Engineer · Backtest Your Agent Against Yesterday's Web — Ryan Ahern, Exa
Recommendation · MoneyCheck whether adding a second feature — 'has it been reported to be exploring a sale' — improves the score or makes it worse; adding it picked up many acquisitions but also false positives.
AI Engineer · Backtest Your Agent Against Yesterday's Web — Ryan Ahern, Exa
Recommendation · MoneyUse Exa to go out, search the web, turn unstructured information into a structured signal, and then trade against it in a live model.
AI Engineer · Backtest Your Agent Against Yesterday's Web — Ryan Ahern, Exa
Recommendation · MoneySearch a bunch of companies using Exa Snapshot and version search to test whether a feature like 'has this company lost its CEO or CFO or appointed interim leadership' predicts acquisition.
AI Engineer · Backtest Your Agent Against Yesterday's Web — Ryan Ahern, Exa
Recommendation · MoneyUse the additional code-shipping ability AI gives you to steer toward moats: understand what moats exist in your space and which ones will stay around, then go for that. Even that may not be enough in and of itself.
the speaker notes even this may not be sufficient on its own
MicroConf · SaaS Moats in the AI Era: Sell the Outcome, Not the Tool
RecommendationAmeisen says the on-ramp to the interpretability field is mostly self-studying and reading key papers. He points to neuronpedia.org, an open-source interpretability site that hosts models and lets you run interpretability experiments on them, and encourages people to just do some interp and read about how models work. You can also reach out to him on X to learn more about how AI models work, and watch the videos the podcast team produced describing how AI models work, including topics like JSpace.
Lenny's Podcast · Inside the black box of AI: How AI models actually think | Emmanuel Ameisen (Anthropic)
Recommendation · MoneyThe speakers recommend taking screenshots and recording videos and putting them on the pull request, noting the GH CLI supports this now.
AI Engineer · Teach Your Agent to Prove It's Done — Nick Nisi & Zack Proser, WorkOS
Recommendation · MoneyThe speakers recommend that if the model can't see the running canvas, print out the ball's coordinates at every point so bugs become visible to it.
AI Engineer · Teach Your Agent to Prove It's Done — Nick Nisi & Zack Proser, WorkOS