35,595 results · Most relevant
The host recommends measuring through MER or contribution profit instead of obsessing over Facebook CPA. A smart founder friend told them that people are obsessed with Facebook ROAS but don't understand you're still paying for the ad to be served, like billboard advertising.
endorsed
The speaker advises clipping the hell out of your videos to repurpose them for YouTube Shorts, TikTok, and Instagram Reels, and emphasizes identifying the best sound bites from long videos and making sure they get published. A long interview takes them about a month total to prepare, film, and edit, so maximizing cut-downs matters.
The month-long timeline is the speaker's anecdotal estimate.
For both videos and blogs, you want AI to help with a lot of the preliminary research, but the final product should be written by you; then you can use a tool to clean up grammar or be very explicit that it should make very little edits. When scripting or ideating, create multiple different versions with AI's help and look at them yourself.
Presented as the speaker's recommendation.
The team generates five scripts per model to check for intramodal (intra-model) variability, which Bouchard highly suggests for anyone building their own evaluations; five drafts is the good middle ground in terms of cost.
Designers should jump into this flexible influx period and insert themselves into the process, articulating design intent into the tool so it gets built by people thinking about design psychology.
Max Kanat-Alexander recommends focusing on reducing the cost of effort, reducing the cost of rigor, above caring where you are on the disposable-durable scale. The cheaper you make rigor, the less you have to worry about where you are on the scale. Make rigor cheap so it's easy to change your mind when a disposable project turns out to need to be durable. Since you will never be a hundred percent right on your disposable-vs-durable estimates, build for cheap reversibility. Adopt a set of practices aimed at decreasing the cost of engineering rigor, which are distinct from rigor itself.
To deploy in production, you need not just great parsing accuracy but also tools for evaluation and review and constructing your individual workflow.
Gareth Yoder's leadership points: (1) make your legacy codebase AI queryable by exposing it via an MCP server, which cut research feedback loops from seven days to 15 minutes; (2) treat migration docs as part of your product — they are just as critical as business functionality because all producers will use them to onboard, and while they will eventually get deprecated, during the interim process they are critical; (3) reimagine and don't lift and shift; (4) adopt AI incrementally, continuing to enhance your product as AI tools arrive, because there is never a perfect starting point — the important thing is to get started and continue to enhance and adapt as the technology changes; (5) observability is non-negotiable — you need to know who's producing your data and who's consuming your
Movie Chain is a game where you take the last movie you watched and pick a common denominator — the director, the writer, or the music composer — to bridge to the next movie you want to watch. It's a bit of a Kevin Bacon number type of game. The creator's family example: they watched Fall Guy (because Netflix recommended it), didn't really enjoy it, but used Ryan Gosling as a bridge to Project Hail Mary; from Project Hail Mary they hopped to The Martian because they share a book writer, and from The Martian to Prometheus because Ridley Scott is the common director; Michael Fassbender linked Prometheus to Inglourious Basterds. Reddit came up with their own version of the game in the R Movies subreddit, though the creator swears their family invented their own version first.
Gary Vaynerchuk argues creators should post content on every platform, including Snapchat, Spotlight, and YouTube Shorts, because it only takes about two extra minutes per post and skipping platforms leaves free reach on the table.
Gary Vaynerchuk argues that if you're trying to build something, claiming you don't have time to create content, build brand, and grow your business is incoherent — nothing could be more important than letting the world know why they should buy your thing.
Gary Vaynerchuk says that if he were starting from zero today, he would make content on the internet about things he knows — citing business, wine, and sports as his own examples.
GaryVee says the number one thing everyone can do is teach people how to fix the things they know how to fix at home by themselves, with little tweaks or ways to figure out what's wrong with them. Providing value creates the karma and energy that leads to the sale, and the most committed person to bringing the most value to the audience wins by letting the residual effects of that value drive their actual business. He notes he has watched value-first creators succeed over and over, and that doing good things leads to good results — actual karma being the most real thing in the universe to him.
Businesses should choose neutral AI applications that use whatever model supplier is best for them in the moment, rather than locking into one lab's product; smart businesses should think carefully about the neutrality of who provides their intelligence and who does the routing. Model supplier choice is not just about general intelligence — suppliers differ in which modality (voice, video, slides) they are best at at any given time. Choosing a neutral intelligence provider lets a business save money and de-risk ever having picked the wrong AI partner.
speaker's recommendation, endorsed; the modality claim is described
GaryVee advises creators and businesses to shoot content at stores, parking lots, and festivals, arguing that the real world is the best studio and that a business's actual location is a better set than anything staged.
Implement a learning loop where your AI worker acts on the world (resolves tickets, issues refunds), observes the results, and improves from them — with learning transferring across your customer base so the business compounds. This is the one mode likely to stay ahead of the expanding LLM capability frontier, the one moat the hosts are confident will stay ahead for years. Static capability will be eroded by frontier models.
Rob Walling recommends implementing one or more SaaS moats now, even if AI hasn't yet impacted your business — the impact is coming. Anar Volset adds: estimate how quickly AI will spread through the economy into your specific niche, since the timeline depends on what your customers are looking at. Even bootstrappers and indie hackers should think about moats — the old playbook (find a niche, thread a needle, get SEO) no longer guarantees survival.
endorsed
Al Chen recommends building trust with security compliance teams by not just saying 'trust us, here's the answer' — show the evidence and the work on how you arrived at the answer.
In a low-stakes environment building an internal chat agent, you can start with tracing and some monitors and go to production directly without heavy offline evals. In low-stakes environments, build out test cases over time and add them as you progress. If you are building a customer-facing case, especially in regulated environments, you probably don't want to risk the agent doing something it's not supposed to do. In regulated industries, the foundation is always tracing: know what steps the agent is taking, what inputs and outputs it sees, and what tools are being called.
endorsed
Doneyli recommends using experiments to prototype changes — a new model, a system prompt change, or new tools/data access — before promoting to production. He discourages 'just running on vibes': pushing changes to prod because 'it looks kind of good' is what everyone has been doing, and it's not systematic.
Choose sources → to get a library of your own and a daily brief.