The challenge
This company was already leaning into AI. Product and marketing were shipping disciplined AI-assisted work, engineering had full license coverage, and the CEO had a sharp read on where growth would strain the business first: implementation and support. The question was where AI would expand capacity fastest, and how to roll it out without unsettling teams that had recently been through change.
Our approach
We interviewed every member of the executive team and assessed every function with an AI surface, using the same five-dimension, behavior-anchored rubric we apply across engagements. At the CEO's request, individual contributors were engaged only once internal messaging was ready, so discovery stayed leadership-level and work with individual contributors moved into delivery. A preliminary readout midway aligned the CEO with the findings before the final report. Where evidence couldn't support a score, the report marked it N/A.
What we found
One diagnosis ran through every function.
Data & Systems Readiness scored 2/5 everywhere it could be scored. Implementation specialists spent about three hours preparing for each customer meeting, manually joining five systems. Support could need six systems to answer a routine question, and there was no knowledge base. Edge cases went to a Slack channel where developers and product managers answered on the fly. Asked how much know-how lived in people's heads, the COO said:
“Lots … it's risky.”
Engineering adoption was a distribution.
Every engineer was licensed and nobody was at zero, yet one engineer had generated roughly 100,000 lines with AI while others sat near 2,000. There was no usage telemetry, so impact was being asserted without measurement. A junior engineer's AI-assisted fix had broken data for all users and taken six to eight weeks to unwind, which explained the caution in the middle of the team.
The value showed up during discovery.
In the COO's interview, we ran a working AI-assisted meeting-prep workflow live. She mapped it onto her implementation team unprompted:
“You just did all your prep with the agent … done.”
Appetite for AI varied between teams.
Implementation asked for relief directly. Support worried about being replaced, because its professional identity came from solving hard questions. That difference shaped the rollout.
What we recommended
We recommended three flagship programs, each tied to a named business outcome:
- An implementation prep agent built on existing licenses. Recovered hours go to proactive customer engagement, which feeds the word-of-mouth referrals and RFP wins the company grows on.
- Support knowledge systematization: one support stack, a queryable knowledge base, and an internal-only AI layer with a human always between AI and customer, framed around a career ladder from support into implementation.
- An engineering floor program built on standardized context, a shared skills library, pre-commit guardrails, usage telemetry, and visible recognition. The CTO's frontier agentic work stays protected as the company's R&D, and the structure carries the rest of the team.
The report also proposed a read-only AI connector to the company's own platform. It would ground internal work in how each institution actually uses the product, and could become a product differentiator for customers adopting AI themselves.
The report named where to hold back. Sales and account management were carved out, since the function ran through one leader already at capacity. Customer-facing support automation was deferred until supervision skills are proven. HR and finance were deferred to a later wave. The single tool purchase we recommended came with a kill criterion.