Perplexity’s daily traffic surged 7x in a year. Mistral shipped three model families in just over six months. OpenAI hit 100 million users in less time than it takes most boards to hold their first strategy session.
With AI, there is no slow lane towards transformation. But, while some companies quickly pull ahead, the way many organizations still make strategic decisions—how they approve, prioritize and fund—is far too slow for the AI era. Rear-view legacy decision-making structures designed around historical data and quarterly cycles haven’t kept up.
Fixated on outdated KPIs, and caught in a feedback loop where analyst reports echo headlines and headlines reinforce stale consensus, many executives can’t make decisions fast enough—including when it comes to their AI strategies.
What’s now needed is a shift in how decisions are made as well as new AI infrastructure to support those decisions. For leaders, it’s long past time to reconsider traditional approval cycles and to adopt AI-first operating models driven by live signals rather than lagging indicators. Senior leadership must realize that while ‘speed’ can feel risky, it’s only by making speed a repeatable, insight-driven system that they avoid the greater risk of falling behind.
The rear-view trap
As the tempo of global business rises daily, many enterprises are stuck using yesterday’s playbooks — Gartner decks, competitor press releases, year-end ROI tables—to make strategic decisions, including choices about how and when to adopt AI. The reality is that these static indicators often describe past conditions that have already been reshaped by new AI technologies, evolving user needs and emerging competitors. They’re often representative of a mindset that leads to a ‘rear-view trap’ where capital is misallocated and opportunity is missed.
The risk of this backward-looking mindset is lengthy ROI vetting based on yesterday’s paperwork. The paralysis that this leads to is evidenced in the numbers. Over 88% of AI projects stall before reaching production. This year, the average organization scrapped 46% of AI proof-of-concepts and the number of firms abandoning their AI pilots is up 17% from 2024. Again and again, traditional decision-making structures can’t keep pace and many firms are left trapped in AI pilot purgatory.
Google famously showed the cost of old approval rhythms when it delayed responding to ChatGPT and let OpenAI seize the early lead. Today’s leaders must learn from mistakes and rebuild decision structures for speed, not certainty. It’s time to empower teams to act faster by replacing multi-month approvals with funding tied to opportunity, not consensus.
From slide decks to signal-based infrastructure
If the answer to the rear-view trap isn’t another report, what does speed look like in practice? To match the velocity of AI, organizations must not only change the approach to decision-making at the top, but also introduce new signal-based infrastructure that makes faster decisions possible.
This new approach means replacing quarterly planning cycles with AI that supports real-time signal loops. Enterprises and mid-size companies alike need dashboards that can track emerging trends, systems that can flag behavioral shifts, and operating models where product and engineering teams act on what’s happening now instead of last quarter.
Going forward, firms must treat signals—from usage anomalies and GitHub activity to emergent funding rounds—as decision triggers. These signals are out there, and the companies acting on them win. In healthcare, retail and finance, teams are using real-time signals to redirect roadmaps, accelerate launches and capture emerging demand. Spikes in usage, new tool adoption and test market results are triggering action long before traditional planning cycles catch up.
Firms can start by building an AI-first signal dashboard that clarifies which shifts are inevitable, when they’re likely to impact financial performance and how much value is at stake by moving first. From there, the focus should shift to a few core outcome metrics—like acquisition efficiency, engagement, or customer lifetime value—ensuring every new initiative moves the needle on at least one of them.
Organizations should start small, outside of rigid structures, so that teams can move quickly and take smart risks. And rather than trying to optimize internal workflows, focus early efforts on areas where the company already creates real customer value—because that’s where signals surface first, and where speed matters most.
Speed that builds, safely
The only firms that will match the speed of the market are those that are releasing, learning and adapting in tight loops. For these firms, speed becomes an insight-driven system. Each release eliminates unknowns and feeds back real-user telemetry: not what customers say they might do, but what they actually do with the product in front of them.
With every cycle, their advantage compounds. The data shapes the next iteration, and the next, stacking proprietary insight that no competitor can replicate or rush. In this model, failure is actionable data, not a setback. Productivity is measured by how quickly you can experiment, learn and improve.
Crucially, this kind of velocity doesn’t come at the cost of quality, so long as safety is built in from the start. That means constraining each iteration to a minimum viable agent during development, limiting data inputs or model scope where necessary, and enforcing automated quality and compliance gates once in production.
In other words, in the AI world, speed isn’t reckless—it’s how fast you learn, adapt and build advantage. Speed is the strategy and the system, and the safest move left.
