Editor’s note: This piece looks at a question we get from nearly every enterprise prospect: should you start with AI strategy or jump straight to building? The answer depends on where your organization actually stands, not where you’d like to be.
The conversation has shifted from whether enterprises should adopt AI to how they should sequence it. Almost every organization I talk to now has at least one AI pilot running somewhere. Far fewer have a coherent path from that pilot to something that actually changes how the business operates. That gap is where most of my conversations with prospective clients start.
Thanks for reading! Subscribe for free to receive new posts and support my work.
Why This Question Keeps Coming Up Now
Enterprise AI adoption has crossed a threshold that strategy hasn’t caught up with. McKinsey’s 2025 State of AI survey found that 38 percent of respondents work for organizations with more than $1 billion in annual revenue, and the same research shows organizations are now actively experimenting with agentic systems that plan and execute multiple steps in a workflow rather than simply responding to prompts. (Source: McKinsey, The State of AI: Global Survey 2025)
That’s the environment I’m seeing with clients too: budget is no longer the blocker, sequencing is. Teams either bring us a strategy document with no build partner attached to it, or a build request with no strategy behind it. Both stall for the same reason: nobody defined what “done” looks like before work started.
Do You Need Strategy Before You Need Code?
Not always, and I’d be careful with the assumption that strategy has to come first in every case. If your organization already knows exactly which process is broken, has clean data behind it, and just needs the technical work done, jumping straight into a Custom AI Development Company engagement is often the faster, cheaper path. Strategy work on top of that would just be paying consultants to confirm what you already know.
Where strategy earns its cost is the opposite situation: multiple candidate use cases, unclear data readiness, or a leadership team that hasn’t agreed on what “success” means for AI in the business. In that case, engaging enterprise AI consulting services for a readiness assessment, roadmap development, and use-case prioritization prevents a much more expensive mistake later: building the wrong thing well.
What Happens When Companies Skip Straight to Building
The technical work usually goes fine. The adoption doesn’t. I’ve watched this pattern repeat across industries: a well-built model or chatbot ships on schedule, passes every QA gate, and then sits mostly unused six months later because nobody mapped it to an actual workflow owner or decided how it would be measured.
The data backs this up at scale. Nearly two-thirds of organizations in McKinsey’s 2025 survey have not yet begun to scale their AI programs at the enterprise level, with most still stuck in testing or proof-of-concept phases without a clear strategy for large-scale adoption. (Source: Neodata Group, McKinsey State of AI 2025 analysis)
That’s not a technology problem. It’s a sequencing problem, and it’s the exact gap strategy work is meant to close before a single line of code gets written.
How Do You Know Which Track You’re On?
I ask clients three questions early on, and the answers usually make the path obvious:
Can you name the specific process AI would change, in one sentence? If the honest answer is “we want to use AI somewhere in operations,” that’s a strategy conversation, not a build ticket.
Do you already know who owns the outcome once it ships? Predictive analytics, workflow automation, or a domain-trained LLM all need a business owner accountable for adoption, not just an IT sponsor for the build.
Is your data in a state where a development team could start next week? If the answer involves multiple disconnected systems and no data governance conversation has happened yet, that’s usually a readiness assessment problem before it’s a development problem.
Two “no” answers out of three is a reasonable signal to start with strategy. Three “yes” answers means you’re probably ready to scope a build directly.
What Does a Combined Engagement Actually Look Like?
Most of the enterprise work I’ve been part of over the past year didn’t split cleanly into “strategy, then development.” It ran as one engagement with two phases: a short readiness assessment and use-case prioritization sprint, followed directly by a scoped build, often starting with an AI-driven MVP rather than a full platform. That sequencing matters. It means the first thing that ships is small enough to validate the assumption before the larger investment gets made, which is exactly the discipline McKinsey’s research points to as separating the small group of high-value performers from everyone else stuck in pilot mode.
My Advice for the Next Six Months
If you’re weighing where to start, don’t treat strategy and development as competing options. Treat strategy as the thing that tells you which build is worth funding, and treat development as the thing that makes the strategy real. Skipping either half tends to produce the same outcome: a technically sound project that never earns its budget back.
If you’re wondering which track makes sense for your organization’s specific situation, I’m happy to walk through it directly.
Ready to scope your next AI initiative the right way? Hidden Brains pairs AI strategy and consulting with hands-on custom AI development, so the roadmap and the build come from the same team.
FAQ
Should a small or mid-size business start with AI strategy consulting, or go straight to development? It depends on clarity, not company size. If you can already name the exact process AI would change and your data is in reasonably good shape, a scoped development engagement is usually faster and cheaper than adding a strategy phase on top. If you’re choosing between several possible use cases, a short readiness assessment first prevents building the wrong thing.
How long does an AI readiness assessment typically take before development can start? This varies by organization, but a focused readiness assessment and roadmap sprint is generally measured in weeks, not months, especially when it’s scoped to feed directly into a development engagement rather than run as a standalone deliverable.
Can AI strategy and custom development be run as a single engagement? Yes, and in practice that’s often the more efficient path. Running readiness assessment, use case prioritization, and an initial build (frequently an AI-driven MVP) as one continuous engagement avoids the handoff gap that causes many strategy documents to never get implemented.
What’s the biggest risk of skipping the strategy phase entirely? Based on McKinsey’s 2025 data, the risk isn’t technical failure, it’s adoption failure: a well-built solution that ships on time but never gets integrated into a real workflow because no one defined ownership or success metrics before development began.
Sign in to leave a comment.