"We have three AI initiatives running. None of them made it to production." A retail CTO said that in a conversation.

The same week, a Head of Innovation at a services company put the problem another way: "Everyone here wants AI. Nobody knows which problem to solve first."

And a fintech CEO, straight to the point: "The board is asking for results. The technical team has no map."

Three quotes, three different companies, the same pattern. Budget and interest already exist in all three. What's missing is a diagnostic that shows where AI creates real value and in what order to tackle it.

Why an isolated pilot doesn't turn into results

Most companies that approach Softo have already tried something with AI. A pilot here, an automation there, an internal assistant nobody uses anymore. The pattern repeats: scattered initiatives across different areas, with no owner, no success criteria, and no connection between them.

The mechanism behind this is simple. Each area decides to test AI on whatever is in front of it, with no visibility into what other areas are testing and no shared criteria for what counts as success. Marketing measures engagement, Operations measures time saved, and nobody can add those two numbers up in a board conversation. When the pilot ends, it doesn't reach a decision to invest further or stop; it becomes a slide filed away, because there was no baseline set before it started.

The technical competence is almost always already on the team. What's missing is the layer before that: where AI solves a problem that genuinely hurts, what to prioritize given limited resources, and who will sustain it once the initial enthusiasm fades. Without that layer, AI becomes an expensive opinion: the board sees no aggregate return, and the company accumulates experiments instead of accumulating capability.

How AI Foundations works

AI Foundations is a 4-week engagement, with fixed scope from the start, designed to close that gap before any build decision. The format is deliberately lean: 4 weeks forces the team to prioritize instead of mapping everything, and this is only possible because the diagnostic and output-production process is already structured and actionable.

The difference from a traditional strategy consultancy shows up in the first week, not in the final report. Classic consulting tends to take months of diagnostics before any tangible result, delivers generic recommendations that ignore the client's real stack and context, and disappears after delivery.

AI Foundations delivers actionable artifacts from the first phase onward, anchored in the company's real operation and stack, with Softo's team working side by side with the client's technical team.

The three phases and what each one delivers

The engagement follows a three-phase arc, and each phase exists to eliminate a specific risk that the previous phase would otherwise leave open.

Diagnostic We map where the company stands and where AI has real potential, across multiple fronts of the business, cross-referencing technical maturity, available data, and operational pain. The output of this phase is a maturity scorecard and a list of critical gaps, prioritized by impact, along with a read on why previous initiatives stalled. Without this map, any following decision is a bet: the company could be optimizing a front that doesn't move the number the board is asking about.

Planning With the diagnostic in hand, we define what comes first, in what order, with what expected return. Each priority front gets a strategic thesis, a business case with the return calculation behind it (cost reduction, speed gain, or incremental revenue, depending on the front), a tools and architecture map, and a line on the adoption roadmap with an owner, a deadline, and success criteria defined before a single line of code.

A vague input like "automate customer service" becomes something concrete: "reduce average response time by X%, with Y as owner, tested within Z weeks, measured by metric Q." That is what separates a list of ideas from an executable plan. Without this plan, the company executes fast but executes the wrong thing, and the cost of reversing an architecture decision after it's built is far higher than the cost of getting it wrong on paper.

Governance We put in place the mechanisms for the operation to continue after the engagement ends: who approves the next initiative, who measures results and how often, what level of risk is acceptable for experiments, and how the company avoids depending on a single vendor or a single person to keep everything running.

Without this, even the best AI initiative dissolves within a few months, because nobody has clear authority to decide what comes next.

How this plays out in practice

Examples from two companies that went through AI Foundations recently, in different contexts.

A B2B services company arrived with scattered uses of AI across Marketing, Sales, and Operations, with no clarity on priorities or governance. The diagnostic mapped maturity, bottlenecks, and opportunities across the three areas, and the result was a prioritized agenda with an evolving plan connecting technology, processes, data, and people, instead of a broad theme with no entry point.

A family office with market intelligence had a routine of tracking and curating strategic information that depended entirely on manual effort, with no path to scale. Before automating anything, the team mapped the process and defined the quality criteria the solution would need to preserve, because automating a poorly understood process only multiplies the error faster.

Only then did the team design the AI-driven flow for searching, curating, and synthesizing information about the portfolio.

Different contexts, the same principle underneath: diagnose before deciding, decide before executing.

The next step after AI Foundations

A company that completes AI Foundations knows where AI creates value in its business, what to prioritize first, and how to sustain it once the initial project ends. From there, the conversation shifts in nature: it moves from strategy into execution, typically through an Outcome Pod dedicated to putting those priorities into production.

If your company has already invested in AI and hasn't seen a return, or if you still don't know where to start, that's the conversation worth having first. Talk to a specialist or take the free maturity diagnostic.