ETTE AI Champions Program: a case study covering program approach, real-world examples, outcomes, and the team behind a 4-month internal AI adoption initiative.
AI Champions: the adoption model ETTE tested on itself first
Between March and June 2026, fifteen ETTE team members ran the AI Champions Program across ticket work, reporting, security analysis, and client communication, entirely in ETTE-governed tools. This page documents how it ran, what changed, and what the model looks like for clients now.
The AI Champions operating model
ETTE tested the model on its own service work first, then converted the lessons into a client-ready adoption path.
-
ETTE's own workflowsTickets, QBRs, security analysis, compliance docs
-
AI Champions cohortPilot first, full-team rollout, mentor pairing
-
Shared skills libraryReusable prompts, documentation flows, review habits
-
Governance lessonsTool access, AI usage policy, data boundaries, quality control
-
Client adoption programThe same model adapted to client workflows
A phased rollout built on everyday service work
The program ran on the team's day-to-day service work, anchored by regular all-team sessions. Team members brought back what worked and what didn't, demonstrated their own work products, and shared learning resources like curated videos and live demonstrations. A small pilot group went first, and the full team joined once the pilot proved out. Useful discoveries became shared skills the whole team now runs by default.
Mentor pairing
Everyone who joined at the full-team kickoff was paired with a peer from the pilot group rather than a manager. Questions about a skill or an odd tool behavior went to someone who had hit the same wall a month earlier.
Skills as shared IP
When someone built a useful pattern, like a QBR template or a ticket documentation flow, it was reviewed and promoted to an organizational skill in the shared library. Everyone on the team gets these skills by default.
Session timeline
What the work actually looked like
Every example below comes from session notes or was described directly by the person who did the work. All of it ran inside ETTE's organizational accounts, in tools ETTE manages and governs, under the AI usage policy introduced at kickoff. The focus was ETTE's own servicing and the deliverables clients receive.
Our Operations Manager redesigned ETTE's quarterly business review format from scratch. The old format relied on screenshots pasted into slides. Using Claude's Cowork mode and a refined skill, he rebuilt the QBR workflow so that raw system data goes in and a formatted, professional client report comes out. The same approach was applied to a client's board-level security presentation, co-produced with a Tier 3 security engineer.
ETTE manages wireless infrastructure at five government agency sites under a multi-year contract. Our Program Co-Lead needed a comprehensive five-year analysis to support a contract renewal, a task spanning roughly 300 monthly reports. Claude's SharePoint connector ingested the documents and the contract, and the team produced a draft that would have required thousands of dollars of manual analysis time.
A client's board needed a security briefing on very short notice, and the underlying material was dense: screenshots from email security tools, Microsoft Secure Score exports, and reports from other security platforms. A Tier 3 security engineer consolidated all of it, worked it through Claude to surface the risks a non-technical audience needed to see, and iterated the result into a board-ready presentation. It identified achievable Secure Score improvements and gave the board a clear read on where they stood, on a deadline manual work couldn't have met.
A Tier 3 systems engineer ran a monthly invoicing script that required entering credentials and completing two-factor authentication manually for each Microsoft-only client. The new Graph API authentication wasn't working as expected. Claude walked through creating an app registration with the right permissions and a stored secret. The script now runs without any manual intervention.
A Tier 2 infrastructure engineer's first LightSail deployment took two days of manual SSL configuration, PHP setup, and Amazon-specific quirks. The next one, with Claude reading the documentation live and walking through each step, was done in a couple of hours. He also started a Claude Code audit of a long-running legacy codebase that no one on the team fully understood.
Multiple customers have now commented: 'Wow, you guys really know about this stuff — that's really impressive what you're doing.' That's a differentiator for us. We want to make sure we're pulling away from the pack of other MSPs.
— Program Lead, Closeout Session · June 30, 2026What changed by the time it closed
Program leadership observed a clear improvement in ticket documentation across the team. Notes became more descriptive and more useful to colleagues, and client communications got more professional.
Client survey results improved during the program period. In direct client conversations, feedback had been "very, very happy." Multiple clients specifically commented on ETTE's AI knowledge as a differentiator.
Six or more organizational skills were deployed to all team members by closeout, covering client communication, ticket documentation, QBR generation, onboarding, and performance reporting. Each skill is version-controlled and available by default.
Everyone who joined at the full-team kickoff was paired with a mentor from the pilot group. Pairings were skill-matched where possible: the Tier 2 infrastructure engineer mentored a colleague who shares development responsibilities, and the Tier 2 technician mentored a Tier 3 systems engineer.
Training continued past closeout. Team members, including program leadership and several engineers, are pursuing further structured AI learning and skill development as part of how ETTE keeps its practice current.
AI Champions across roles
Descriptions reflect documented contributions from session notes and program closeout observations.
Designed and ran the program. Built the organizational skills library in GitHub, set up the mentorship pairings, and led all six sessions. Used Claude to compress recurring compliance documentation work that would normally take months into a matter of weeks. His patch-failure troubleshooting and log-aggregation workflows are now standard practice.
Co-led the program and handled client-facing AI strategy. Used Claude to analyze 300 monthly wireless reports across five government agency sites for a five-year analysis that would otherwise have cost thousands of dollars in manual consulting hours. Connected financial and payroll systems through Claude connectors. Trained three team members on Claude basics outside the scheduled sessions.
One of the program's most prolific contributors. Rebuilt ETTE's QBR format, replacing screenshot-heavy slide decks with data-rich client reports. Worked with a Tier 3 engineer on a client security presentation that would have taken two weeks to produce manually. Uses Claude daily for research, threat analysis, and report generation, after starting the program as a regular ChatGPT user.
Turns dense technical data into client-friendly deliverables. Fed a client's security data (screenshots, email security logs, Microsoft Secure Score exports) into Claude to pull out the risks a non-technical reader needed to see, and identified an 8–10% Secure Score improvement achievable without disrupting users. Introduced the team to the Claude Office add-in for consistent PowerPoint formatting. Reviews every AI-assisted deliverable before it goes to a client.
Went deepest on the technical side. Cut an AWS LightSail deployment from two frustrating days to a few hours. Started a Claude Code audit of a long-running legacy client codebase. Built a shared Claude project with a colleague to track a 10-page website update; brief status notes went in, and progress documents, an Excel tracking sheet, and a client-ready email draft came out. Also wired up HelpSpot's API for private ticket documentation.
Cut user guide creation from hours to roughly five minutes. Used the HelpSpot connector to find historical tickets from a vague memory cue instead of a manual search. Drafted sensitive client emails in several tone variations and picked the best fit. Worked on reusable PowerShell scripts for calendar permission assignments. Mentored a Tier 3 engineer who joined at the full-team kickoff.
Used Claude to set up Microsoft Graph API app registration for ETTE's monthly invoicing scripts, which used to require manual credential entry and a two-factor prompt on every run. The scripts now run without any manual step. Recognized at closeout for standout documentation quality during the program.
Runs suspect email headers and URLs through Claude for a rapid first-pass phishing assessment. Also uses it to draft clearer responses to non-technical users, particularly on security tickets where plain language matters.
Joined at the April 24 kickoff. Planned and executed a migration in a single day that would previously have taken several days. Leans heavily on screenshot uploads for visual troubleshooting.
What ETTE now offers
AI Champions cohort engagements for client organizations
ETTE now runs the same phased-cohort approach inside client organizations: tool selection and access, an AI usage policy, mentor pairing, structured sessions, and an organizational skills library built around the client's actual workflows. These engagements are part of ETTE's AI Strategy & Enablement practice, which covers the full range of advisory levels and ongoing support.
The program fits knowledge-worker teams that want practical AI fluency and want to keep the capability in-house after the engagement ends.
Clients in legal, government contracting, financial services, and nonprofit work have already expressed interest. ETTE ran the program on its own team before offering it to anyone else, and its sessions, usage policy, and skills library came out of that experience.