Case studies / 01
Digital Media Organization
From AI experiments to an operating capability.
The organization had AI opportunities across multiple functions, but the value did not come from choosing one large AI project. It came from identifying recurring workflows where information had to be assembled, interpreted, communicated, or turned into action - and redesigning those workflows around AI.
At a glance · $35M organization
Outcome figures reported by the client organization.
The business problem
The business problem
The organization ran a set of interconnected functions - finance, planning, leadership, board communication, fundraising, events, digital product, and audience-facing experiences. Different teams, different rhythms, one shared characteristic.
Across most of them, the same shape of work kept recurring. Someone gathered information from several sources. Someone interpreted it. Someone summarized it, reformatted it, and prepared a communication. Then, on a predictable cycle, someone did it again.
The opening question was deliberately not "how can we use AI" or "what tool should we buy." It was narrower and more useful:
Where does knowledge work repeatedly consume time without creating equivalent strategic value?
The AI operating layer, and the pattern common to every workflow
Human judgment sits in the same position in every workflow. That was the design rule, not an afterthought.
Stage one
Discovery
Discovery examined workflows across financial planning and analysis, event planning, board communications and reporting, funder outreach, and interactive web experiences. For each one, the same questions applied.
- What inputs does this workflow require, and where do they come from?
- Which tasks are genuinely repetitive, and which require judgment?
- Who consumes the output, and what do they do with it?
- Where do delays actually occur?
- Where is AI appropriate, and where does human ownership need to remain?
The workflows came first. Technology came second. That ordering mattered more than any individual finding.
Stage two
Assessment
Every candidate workflow was weighed against the same criteria, which made the comparison honest rather than enthusiastic.
- Business value - would improving this materially affect cost, speed, quality, decision-making, fundraising, or stakeholder experience?
- Repeatability - does it happen often enough to justify changing it?
- Information - are the required inputs available and reliable enough?
- Judgment - which parts should AI assist with, and which stay human-led?
- Risk - what happens if the system produces a poor answer?
- Adoption - can the redesigned workflow realistically become normal practice?
The governing principle was that not every workflow that can use AI deserves AI investment. Several candidates were set aside on exactly those grounds.
Stage three
Plan & Build
Rather than one system, the work produced AI-assisted workflows in five operating areas.
Financial planning and analysis
AI-assisted workflows supported assembling financial information, comparing budget against actuals, interpreting variance, developing scenarios, and preparing recurring management summaries. AI assisted the analysis and the preparation. Financial ownership stayed where it was.
Event planning
AI-assisted workflows supported research, requirements gathering, coordination, communications, and run-of-show preparation - the recurring planning tasks that consume disproportionate time relative to their difficulty.
Board communications and reporting
Organizational information was synthesized into clearer executive narrative, then reviewed by leadership before it went anywhere. The sequence - information, synthesis, narrative, human review, board communication - kept leadership responsible for the final message and the judgment behind it.
Funder research and outreach
AI was applied to parts of the funder research and outreach preparation workflow: assessing relevance, synthesizing research, and preparing outreach for human review before any engagement. Preparation was assisted. The relationship was not automated.
Interactive web experiences
AI was also applied to audience-facing interactive features, which meant the work touched both internal operations and external digital experience rather than stopping at the back office.
Stage four
Outcomes
AI moved from isolated experimentation into repeatable workflows across finance, planning, leadership communication, fundraising, and digital product development.
Rather than treating AI as a separate initiative, the organization began using it as an operating capability - one that could reduce repetitive knowledge work while preserving human ownership of important decisions.
The result was not one AI product. It was a portfolio of targeted AI workflows connected to how the organization actually operated.
What changed
Before, AI was a set of experiments owned by whoever was curious. After, it was a set of workflows owned by the functions that depended on them - with a consistent rule about where human judgment sat in each one.
What this demonstrates
The best AI strategy may be a portfolio, not a platform.
Organizations rarely have one AI opportunity. They have dozens. The strategic work is determining which workflows matter enough to change, what should be automated, and where human judgment still creates the value. Transformation happened when AI became part of how the organization worked, not a separate technology initiative.
Where is the constraint in your business?
The details differ, but the starting question is usually the same: where could AI materially change the economics, and what needs to be true before you invest?