Four to eight weeks · Scored use cases · One roadmap

a plan you can fund and defend

For teams that are done experimenting and ready for a plan they can fund, staff, and defend.

Book a consultationSee the opportunity map →
A surreal outdoor strategy whiteboard on a mossy island, diagramming how people, business, and AI intersect
4–8 wksTypical engagement, start to handoff
78%of organizations now use AI somewhere (Stanford HAI, 2025)
1Roadmap your team can execute without us

Definition

What AI consultingactually is

A structured engagement where an outside team assesses how your organization works, identifies where AI can create measurable value, and delivers a prioritized implementation plan — workflow mapping, scored use cases, build-versus-buy recommendations, governance guidance, and a phased roadmap your team can execute with or without us.

A strategy you can't execute independently isn't a strategy, it's a dependency.

4–8 weeksEvery use case scoredDocumentation you own
78%of organizations use AI somewhere (Stanford HAI, 2025)
4–8 wksKickoff to enablement and handoff
4Scoring criteria: impact, effort, risk, readiness
1Budget line, and a defensible answer for it

The problem

Why most AI initiativesstall before they ship

Adoption stopped being the hard part. Nearly everyone has started. Far fewer have finished anything.

As of Stanford HAI's 2025 AI Index, 78% of organizations report using AI in at least one business function.

The failure pattern is consistent, and it's almost never technical.

Painterly topographic map fragment with a single gold path threading between contour lines
Failure pattern 01

The decision has no owner

AI lives in operations, IT, and marketing simultaneously and belongs to none of them. Three teams run three pilots, none of them talks to the others, and the org ends up with three tools, three vendors, and no strategy. The technology was never the bottleneck. The org chart was.

Failure pattern 02

The use case was chosen by enthusiasm

Someone saw a demo. The demo was genuinely impressive. Six months later there's a chatbot nobody uses, because the workflow it was built on wasn't the one costing you money. Excitement is a bad prioritization function — it selects for the most visible problem instead of the most expensive one.

Failure pattern 03

Nobody planned for the unglamorous parts

Data access. Permissions. Who reviews the output before it reaches a donor, a student, or a patient. What happens when it's wrong. Staff training. Policy. These are the things that kill implementations after the pilot succeeds, and the things a demo never shows you.

The shape of it

Most organizationsdon't have an AI problem

1

Budget line · a hundred plausible ideas

They have a prioritization problem. A hundred plausible use cases, a dozen loud vendors, one budget line, and no reliable way to tell which idea is worth the year. We help you find that answer — and then help you make the case for it to the people who sign off.

We start with your operations, not the technology. Before we name a single tool, we map how work actually moves through your organization — where it slows, where it duplicates, where people are doing something a machine should be doing and where they're doing something a machine absolutely should not.

That order matters. It's the difference between a roadmap built around your constraints and a roadmap built around a vendor's product line.

The AI opportunity map

Every use casescored the same way

Impact, effort, risk, and readiness — applied consistently, so the recommendation survives the question “why this and not that?” Hover a quadrant to see how we treat what lands there.

Low effort → high effortPayoff ↑

Do these first

The work that clears in a quarter and pays for the roadmap. Usually internal, usually invisible to the public, usually the least exciting item on anyone's list — which is exactly why nobody had championed it before the scoring made the case.

Deliverables

What you leavethe engagement with

Written to be forwarded — to a board, a dean, or a leadership team that wasn't in the room.

  • An AI Opportunity Map — every workflow assessed, every use case scored on impact, effort, risk, and readiness
  • A prioritized roadmap — phased and sequenced, with named owners, time horizons, and explicit dependencies
  • Tooling and vendor recommendations — build versus buy versus wait, with cost and lock-in implications
  • Governance and policy starting points — data handling, human review checkpoints, acceptable use, disclosure
  • An executive summary — the version you bring to your board, your dean, or your leadership team

Process

Four to eight weeks,in four moves

Discovery takes the longest because it depends on your people's calendars, not our capacity.

Abstract painterly strategy map with nodes and a prioritized path in the oak palette
01
Talk to the people doing the work

Discovery & workflow mapping

We interview the people doing the work, not just the people describing it. Process mapping across your highest-volume workflows, an inventory of the systems and data you already have, and an honest read on what your team has the capacity to absorb this year.

Stakeholder interviews Systems & data inventory Capacity read
02
Make the recommendation defensible

Opportunity identification & scoring

Every candidate use case gets scored the same way: impact, effort, risk, and readiness. Scoring is the part most strategy work skips, and it's the part that makes the recommendation defensible when someone on your board asks why this initiative and not that one.

Use-case inventory Consistent scoring rubric Shortlist with rationale
03
90 days, next year, leave alone

Roadmap & recommendations

A sequenced plan: what to do in the next 90 days, what to stage for next year, and what to leave alone entirely. Tooling recommendations with real cost and vendor-lock implications. Governance and policy starting points, sized to your risk exposure rather than a generic template.

Phased roadmap Tooling recommendations Governance starting points
04
Documentation, not dependency

Enablement & handoff

Documentation your team can act on without us in the room. Working sessions with the people who'll own execution. And a clean line into a build phase if you want one, with no obligation to take it.

Working sessions Written documentation Optional build phase

Qualification

Is this the rightengagement for you?

This is a fit if…

  • You've run one or more pilots that worked in a demo and never scaled past it.
  • Leadership is asking for an AI strategy and no single person owns producing one.
  • You handle sensitive data — student records, donor information, patient data, public records — and need guardrails settled before tools get selected.
  • You have budget for one meaningful initiative this year and you need to be right about which one.
  • You've been quoted a large implementation and you're not confident the underlying problem was ever diagnosed.

This probably isn't a fit if…

  • You already know exactly what you want built and just need it built. Start at AI Prototype & Build instead.
  • You want a specific tool deployed next week without assessing whether it's the right tool. We'd take your money and you'd be unhappy in a quarter.
  • You're looking for a report to justify a decision that's already been made. We'll tell you what we find, which is occasionally inconvenient.

Related services

Where this sitsin our AI services

Consulting and strategy is the widest of our AI engagements. If you already know which part of the problem you're in, one of these will get you there faster.

AI Readiness Audit

A focused assessment of whether your data, systems, and team can support AI at all. Start here if the honest answer to “are we ready” is “we don't know.”

Start here

AI Strategy Sprint

One week, one room, one prioritized roadmap. Start here if you already have candidate ideas and need them ranked and sequenced fast.

Start here

AI Prototype & Build

A working, lightweight version of the thing — an assistant, an agent, an internal tool. Start here if the decision is made and you need proof it works.

Start here

Credibility

The kind of organizationswe think alongside

We've spent five years building digital work for universities, nonprofits, healthcare organizations, and public-interest institutions — the environments where a decision has to survive a procurement process, a privacy review, and a board meeting before it survives contact with users. Oak Theory is a women-of-color-owned studio of eighteen, and that context shapes how we ask questions.

Those constraints aren't an obstacle to good AI strategy. They're the reason it needs to be good. See our higher education work and our work with nonprofit organizations.

Questions

The things teamsask us first

Abstract painterly strategy map with nodes and a prioritized path in the oak palette

Get the planyou can actually fund

Tell us where your organization is stuck and we'll tell you whether a strategy engagement, a one-week sprint, or a readiness audit gets you there fastest.

One budget line. Be right about which idea.

Book a consultation