Merit Is Compounding Faster Than Pedigree – And AI Is Accelerating Both

Every few months, Silent Eight’s CEO Martin Markiewicz and CMO Vlada Grebenykova sit down for a wide-ranging discussion revolving around where the market is heading. While the pair usually walk away with a notebook full of varying details to fact-check, a recent conversation kept arriving at the same answer from every direction.

Covering three distinct topics – where AI is heading, what the next trillion-dollar company might look like, and how enterprise trust actually gets built – Martin and Vlada’s observations returned to the same idea, said three different ways: merit is compounding faster than it ever has, and pedigree is compounding slower.

The Frontier Is Only A Few Years Ahead 

Ask any technologist what's coming in 2028 and the honest answer is a hedge. Nobody can forecast three years of AI progress with confidence. But a pattern has held across every prior technology wave, just compressed into a shorter timeframe: whatever a small group of people can do on the frontier today becomes normal for everyone within two or three years, not decades. 

The internet took the best part of a decade to go from research labs to households, mobile took most of one, and coding agents went from a novelty to the default way software gets written in under three years.

The behaviour currently defining that frontier is what practitioners have started calling "loop engineering" – rather than prompting an AI step by step, the practitioner sets a precise goal, lets the system check its own progress against that goal, feeds itself corrective feedback, and runs largely unsupervised for hours or days at a time. 

Andrej Karpathy, a founding OpenAI researcher and former Tesla AI director who joined Anthropic's pretraining team in May 2026 [1], put the underlying instinct plainly on the No Priors podcast, stating that to get the most from today's tools, "you have to remove yourself as the bottleneck." [2]

What predicts who reaches that frontier first is not pedigree. It is the willingness to adopt an uncomfortable new discipline before it is proven. 

That discipline is native to coding agents today; Martin’s expectation is that the same loop-based approach will show up across marketing, financial crime compliance, and most other knowledge-work functions within two to three years – not because it is exotic, but because it already works, and diffusion is largely a matter of time. 

Crucially, none of this means people disappear from the process. It means human attention moves to where it is genuinely valuable: defining the objective clearly, building in the mechanisms that let a system check its own progress, and stepping back in at the points that call for real judgement. None of that is a new idea in management — it is how good leaders have always run high-performing teams. AI has simply made it possible to apply at a speed, and on tasks, that used to require constant supervision.

Loop-engineering itself matters less than the discipline behind it: define the outcome, let the system check its own progress against it, and only step back in when real judgment is needed. 

That's not a new management idea – it's how good leaders have always run high-performing teams. AI has simply made it possible to apply at a speed, and on tasks, that used to require constant supervision.

The Next Trillion-Dollar Company Won't Grow Like the Last One

The conversation's second thread concerned who will build the next trillion-dollar company. Martin's view is that it will still be, in structure, a software company – but not in growth shape. 

Historically, a startup's odds of reaching the next stage tend to improve the further it progresses; seed-to-unicorn odds are commonly estimated at somewhere between roughly 1-in-40 [3] to 1-in-100 [4], depending on the study and period measured, with attrition falling sharply once a company clears its first few growth milestones.

What looks different now is the speed at which a company can climb that ladder in the first place. Investment firm Coatue made the point directly in its own research, suggesting that “each decade and architectural shift produces a handful of companies that have the potential to reach trillion-dollar market caps,” while naming Anthropic as one of them [5]. 

The figures support the claim: Anthropic's own disclosures show that its annualised revenue run rate rose from roughly $9B at the end of 2025 to more than $30B by April 2026 – growth the company itself described as “crazy” [6]. 

That is not a company adding customers one deal at a time; it is a materially different growth curve, and it is the shape Martin expects the next trillion-dollar company to take – AI-native from the start, not a legacy software company that bolted AI on later.

Enterprise Trust Still Belongs to Whoever Solves the Problem

The most grounded part of the conversation concerned how enterprise trust actually gets built, with Martin – drawing on experience from personally selling into some of the world's largest financial institutions – suggesting that no deal comes down to relationships.

Martin’s answer for how trust is built, despite being almost annoyingly simple, was merit. Solve a real problem well enough that the customer believes they're working with someone smart and exceptionally efficient, and everything else follows. 

ElevenLabs offers a clean illustration. Nobody adopted its technology due to an investor list, but  because its product became the clear category leader on capability – the company accounted for around 98% of observed spend among its mid-market voice AI customers in early 2026 [7]. 

Two further examples, from categories with little in common with voice AI, make a similar point, as the same test is playing out in coding tools. 

Cursor grew from a standing start to roughly $2B in annualised revenue [8] within about three years – one of the fastest scale-ups any B2B software company has managed – largely before it had built much of an enterprise sales operation; individual developers adopted it because it was simply better, and enterprise deals followed the product's reputation rather than the other way round. It is now used within close to 70% of the Fortune 1000 [8]. 

Wiz tells a similar story in a category where relationships are supposed to matter even more: it reached $100m in annual recurring revenue within 18 months of launch [9] – the fastest any software company had managed at the time – while competing against cybersecurity incumbents with decades of enterprise relationships behind them. It won regardless, and by the time of its acquisition by Google in 2026, it counted more than half the Fortune 100 among its customers [10].

It is a version of the logic widely attributed to Steve Jobs, who is said to have built Apple’s hiring philosophy around it, stating: “If you wanna hire great people and have them stay working for you, you have to be run by ideas, not hierarchy." [11]

Vlada's own experience offers a direct, smaller-scale view of how durable that philosophy can be. Working at Apple in 2013, two years after Jobs' death, she found decisions were still argued on the strength of the idea rather than the seniority of whoever raised it, and the obsession with getting the product right remained the loudest voice in the room.

That, ultimately, is the real test of a merit-based culture: whether it survives the person who built it. The pattern holds more broadly, too. The companies that win enterprise trust fastest are rarely the ones with the best-connected board or the most-polished pitch. They tend to be the ones obsessed enough with the problem that the product does the convincing before the sales team ever has to.

Exceptional products come from exceptional people solving real problems – not from relationships, brand recognition or investor pedigree. Enterprise buyers, especially risk-averse ones like banks and government institutions, can tell the difference between a vendor who's optimizing for the relationship and one who's actually obsessed with their problem.

The Takeaway for Business Leaders

Watching AI compress adoption timelines, growth curves, and trust-building all in the same direction gives marketers less room to hide behind brand and more incentive to make the product's merit the story. 

The companies winning right now aren't winning because they're the loudest – they're winning because, on the frontier, they were simply the best at the one thing that mattered. That's a much harder story to tell than a relationship-driven one, but it's the only one that scales at the speed AI has made possible.