I priced out 4 ways to build an enterprise AI strategy in 2026. Only one made sense.

In Q1 2026, a leadership team I advise — a regional specialty insurance carrier I'll call Cedar Mutual — gave themselves an eight-week deadline. Decide how the company will run AI for the next three years. They had four real options on the table. I priced out all four. Here's what happened.

The setup

Cedar Mutual is a composite, but the situation is exact. About 1.2 billion dollars in written premium. 180 people running underwriting, claims, and compliance. Operating with roughly half the headcount of a traditional carrier their size because they're aggressively digital-first — they think of themselves as an investment shop that packages specialty policies.

Their CEO wanted to know one thing: in the next 90 days, where do we put our chips? The CRO wanted policy and knowledge management for compliance audits. The CTO wanted to know whether to lean on a cloud AI vendor, bring deployment in-house with a partner, or build from foundation models. The CFO wanted a number she could defend to the board.

So I ran the math on the four real options. Not the marketing-deck version — the actual cost, time, and output you get from each.

If you only have time for one section, jump to "What actually made sense." The TL;DR is that the consulting engagements were too slow and the do-it-yourself path stalled. The fourth option — the one I had been most skeptical of — produced what an eight-week consulting deck would have produced, in ninety minutes.

How I tested the four options

I judged each on five things: 1) total cost over the first 90 days; 2) time-to-decision (how long until the leadership team can actually act on the output); 3) defensibility — does it produce the artifacts the CFO and the regulator will accept; 4) who actually has to do the work; 5) the realistic probability that the team ships something that survives contact with finance and compliance.

The 4-option results

Option 1: the Big Four engagement

I'll start with the version everyone reaches for first. A Big Four engagement on enterprise AI strategy is well-defined: eight to sixteen weeks, a senior partner plus three to five consultants, a workshop series, an interim deck, a final deck, and a handoff. The price tag at Cedar Mutual's size lands between $150K and $300K depending on the scope.

The deliverables are real. The frameworks are battle-tested. The problem is the calendar. Cedar Mutual gave themselves eight weeks for the decision. A Big Four engagement that delivers in week sixteen is solving last quarter's problem. The model landscape in 2026 moves on a monthly cycle — according to Gartner's August 2024 forecast, 30% of generative-AI projects will be abandoned after proof-of-concept by end of 2025, and the assumptions in a Q1 engagement deck often expire before the bill clears.

Score: artifacts are good, math is good, calendar is wrong. Stale on arrival.

Option 2: a boutique AI consultancy

The second tier is the boutique. Cheaper than Big Four, sharper on AI specifically, faster cycle — six to ten weeks — and usually one or two principals who know the technical detail in a way the Big Four senior partner doesn't.

I priced two candidates for Cedar Mutual. Both came in between $75K and $150K. Both required a workshop series with the leadership team. Both produced a defensible final document. Both still missed the eight-week window unless we accepted a stripped-down output.

The honest assessment: a boutique engagement is the right answer when you need the consultant in the room for the implementation, not just the strategy. Cedar Mutual didn't. They had an internal team that could ship. They needed the framework, not the framework and the body shop.

Option 3: the internal working group

The third option is the one most teams quietly try first. Pull together a working group — head of IT, head of risk, a few line-of-business leaders, a strategy lead — and have them figure it out internally. No external cost. Total ownership.

This is the option that almost always stalls. Not because the people are bad — Cedar Mutual's internal team is excellent — but because the working group has no structured framework to push against. They have ten prompts across five tools. They have opinions. They do not have a ranked portfolio, a cost model, a governance map, or a defensible architecture recommendation.

I have watched four of these working groups over the past 18 months. All four are still meeting. Two have produced a pilot. Zero have produced a strategy the CFO would fund into production. The 400 staff hours is the visible cost. The invisible cost is the opportunity cost — the quarter you spent in working-group purgatory while the market moved.

Option 4: the self-serve strategic workflow

The fourth option is the one I had been most skeptical of when I first heard about it. A self-serve strategic workflow, built specifically for enterprise AI strategy decisions, that runs the same conversation a Big Four engagement runs — but compresses it into a single guided session with the leadership team in the room.

The version I tested with Cedar Mutual was Iternal's AI Blueprint Builder. Free to start. No sales call required. I'll tell you up front that I would have dismissed it on the marketing copy alone — "AI strategy in 60 minutes" reads like a LinkedIn post, not a serious advisory product. I was wrong.

What the workflow actually does is run a structured interview through your leadership team. You answer prompts about your business, your data posture, your compliance burden, your team, your use case candidates. The output is shaped to the artifacts the CFO and the CRO actually care about. A ranked use case portfolio. An architecture recommendation with the cloud-vs-in-house-vs-build trade-off applied to your specific business. An ROI cost model. A governance map. A failure-pattern review. A phased rollout plan.

I ran Cedar Mutual through it on a Friday afternoon. The CEO, the CRO, the CTO, and me. Ninety minutes. What came out was, structurally, the same document a Big Four engagement would have produced — ranked portfolio, architecture recommendation, cost model, rollout plan — without the eight-week wait or the six-figure invoice.

It is not magic. The workflow only produces what you put in. If the leadership team isn't in the room, or if they don't engage seriously with the prompts, the output is generic. We engaged. The output was specific.

What actually made sense

If I am being honest about how this comparison shook out, the answer is uncomfortable for the consulting industry I came up in. For a leadership team that knows its business, has a competent internal team, and gives itself an eight-week decision window, the self-serve workflow is the right first move. Full stop.

That doesn't mean the consulting options have no role. A Big Four engagement is the right choice when the leadership team needs external air cover for a politically difficult internal decision — the deck becomes the political instrument. A boutique is the right choice when you need an embedded principal to help ship the work. The working group is the right choice when the decision genuinely doesn't matter much and the calendar is open-ended.

Cedar Mutual didn't fit any of those. They had a real deadline, a competent team, no need for external air cover, and a need for a defensible artifact. The self-serve workflow was the only option that met all four constraints.

Who should not use the self-serve option

To be fair to the alternatives: if your leadership team cannot put 90 minutes on the same calendar, the self-serve workflow won't save you. The workflow assumes engagement. If you are looking for someone else to do the thinking, a boutique consultancy is still the right call. If your decision is politically charged and you need a third-party deck to land it, the Big Four is still the right call.

If, on the other hand, you have a leadership team that can be in one room for ninety minutes and is willing to do the work — the self-serve workflow is faster, cheaper, and produces equivalent artifacts. That's not marketing copy. That's the result of the comparison I ran.

Where Iternal fits

The Blueprint Builder I tested is built by Iternal Technologies. The company also sells AI software — AirgapAI, Blockify, AI Assist — but the Builder is intentionally separated from the sales process. You can use it without ever speaking to a sales rep, and the output frequently recommends approaches that don't involve Iternal's products.

That separation is the part that surprised me most. A self-serve workflow that doesn't try to sell you the vendor's own software at the end of every session is a strange business model. The way the team described it: the workflow itself is the lead magnet, the credibility builder, and the relationship starter. The software sale, if it happens, happens later and only when it actually fits.