One thing about rapid AI development that has stood out to me over the years is how little I care, or better put, pay attention to AI capabilities outside my workflow or interest areas.I raged and raged against the dying of the light when I read ghost stories about AI advancements in creative writing and video generation.How dare they come for the thing I love and hold sacred?It felt like a poisoned arrow had torn clean through my chest when Scorsese joined German AI startup Black Forest Labs as an advisor.“The machines are coming with a vengeance!” I cried.Then, with tears still streaking down my face and in the same breath, I all but squealed to a colleague when I realized how much better my AI model of choice had gotten at retaining context and making our shared workflow easier.I am a selfish man. And if I could pick and choose, I’d focus AI research solely on areas I deem necessary and ethical, and keep it far away from art, sentiment, and magic.Far-fetched as it may seem, enterprises share that same selfishness. Only in their case, the bottom line is what they want to keep holy.The benchmarks are great. So are the constant advancements across models. But for most businesses, since the dawn of our capitalist reality, the question has always been very simple: how do we use this new thing to take on our problem space in a way that is efficient, cost-effective, and pleasing to the bottom line?That is why the next big AI bet may not be better models.It may be helping businesses actually use the models that already exist.For Enterprise AI Adoption Today, “Impressive” is Not the Same Thing as “Useful”TechCrunch reported this week that Anthropic and Blackstone are backing Ode, a $1.5 billion enterprise AI services company built around AI implementation, not model-building.https://techcrunch.com/2026/07/15/anthropic-blackstone-bet-the-next-trillion-dollar-ai-business-is-implementation-not-models/?embedable=trueThat tells us something.For a while, the AI conversation was dominated by the model race. Bigger models. Better models. Longer context windows. Stronger reasoning. Cleaner images. More realistic video. Faster voice. Higher benchmark scores.All of that matters, of course. But it does not matter equally to everyone.A media company may care deeply about improvements in video generation. A legal team may care about better reasoning and citation handling. A customer support platform may care about speed, cost, and reliability at scale. A finance team may care less about creative output and more about whether AI can reduce errors, flag anomalies, or speed up reporting.The “let’s throw the most powerful model at everything” phase was never going to last forever. At some point, every impressive demo has to answer to procurement, operations, compliance, and finance.And those people are not sentimental.They want to know what changed. Did costs go down? Did revenue go up? Did the team move faster? Did risk reduce? Did the workflow improve enough to justify the spend?This is why implementation is becoming such a serious opportunity. The model might be the engine, but someone still has to build the vehicle, connect it to the road, teach people how to drive it, maintain it, and prove it gets them somewhere worth going.That is where startups can win.How Useful Is Your AI Product, Really?This is exactly the kind of question HackerNoon’s Proof of Usefulness Hackathon is built around. In an AI market full of demos, wrappers, and “powered by” language, usefulness is still the test that matters most. Does your product solve a real problem for real people? Can users understand the value without needing to be dazzled first?If you’re building something meaningful and want to sharpen your positioning while competing for over $150,000 in cash prizes and software credits, this is a solid place to start.You can get started here: https://www.proofofusefulness.com/submitOf course, implementation does not only happen inside large enterprises.Every day, startups take broad technologies, specific problems, and messy user needs and try to turn them into something people can actually use.Meet BettorEdge, Crown TV Zambia, Fireframe Studios: Startups of the WeekBettorEdgeBettorEdge is a social sports betting marketplace built around peer-to-peer wagering, letting users bet against real people instead of the house. The platform combines public and private bets, competitions, performance tracking, live odds, and Discord-based betting tools, giving sports fans a more community-driven way to wager without traditional sportsbook-set lines or house edge. For users already talking sports in group chats and Discord servers, BettorEdge turns that existing behavior into a more structured betting experience.Crown TV ZambiaCrown TV Zambia is a Lusaka-based television network focused on credible news, local productions, and community-centered storytelling. Built by a team of local and international journalists, the station produces programming across news, entertainment, public affairs, and lifestyle, while also giving Zambian voices a platform across broadcast, online, and social channels.Fireframe StudiosFireframe is a Helsinki-based film studio building immersive-first genre films with virtual production at the center of the process. Its integrated pipeline, Hypnos, is designed to make premium formats, spatial sound, virtual production, and format-aware blocking part of the work from day one, rather than something bolted on after the edit.Want to be the next Startup featured in this newsletter? Share Your Story Today!That’s all for this week.Until next time, keep building useful things!