Real teams, real deadlines, real outcomes.
Every project started as someone’s actual problem. These are the stories — in the teams’ own words where we have them — of what happened when AISE pointed AI (or a deterministic flow, or both) at it.
Every story we can tell
Automating the SKU pipeline by staying close to the people who run it
Every customer catalog request means looking up each SKU by hand, reconciling its attributes against product rules, and re-keying the result into PIM: 33 SKUs for one customer, about 1,500 for another. Shelby Corbitt is building the pipeline that does this on Claude Managed Agents, with a person approving every record, and building it alongside the team that does the work today.
Turning a 2,191-page guide into something you can actually ask a question
SmartAssembly developers needed an assistant grounded in a 2,191-page admin guide. AISE parsed it into 1,491 structured commands with a self-validating retrieval layer.
Reverse-engineering a file format nobody had documented yet
HR Studio was AISE’s first build on Copilot Studio’s new skills-based experience — new enough that Microsoft hadn’t published how it serializes to disk. So AISE charted the format empirically.
AI tested the AI, and caught a real design flaw before it shipped
An orchestrator that both coordinates a workflow and decides outcomes is doing two jobs at once — a subtle flaw that looks fine in isolation. Evals surfaced it before Production.
The doctrine that shaped every build since
TracE — AISE’s first agent, built and run by non-developers — processed thousands of IT articles never written for retrieval. The hard lessons became the house doctrine.
From a $200K/yr buy decision to a bid engine the business owns
A VP was ready to license an external pricing tool at $200K/yr. Instead, AISE and the Specials team built one together — a multi-agent engine grounded in MillerKnoll’s own documented rules that does days of pricing research in moments. It went live on August 22, 2026, and in its first 30 days it agreed with the quoters’ actual decisions 91.7% of the time.
Thirty days of hypercare: how we sit with the business after go-live
Going live is where the real work starts. For AutoBid’s first month, AISE, the Specials team and the Product Portal team tested together every day, shipped more than 40 fixes through DEV, TEST and PROD, and told the business about each one in plain language. The engine stayed up the whole time.
Letting the data design the docs
Rather than guess what the next-gen IT Support Agent should cover, Anna pointed AI at 90 days of real tickets — and let the evidence write the content roadmap.
One copilot, built once and shipped to two brands
Herman Miller Retail and Design Within Reach needed the same thing — Customer Care reps with instant, on-brand answers. AISE built a single copilot and deployed it as DJ and Oscar, differing only by brand skin.
Scoring cyber maturity from the work actually underway
Instead of a hand-assembled annual snapshot, the Cyber Maturity Index scores MillerKnoll’s cybersecurity posture against NIST CSF 2.0 from the live project portfolio — and projects where the score is heading.
The intern hub that schedules its own coffee chats
Everything a summer intern usually pieces together across PDFs and inboxes — policy, resources, who to email, how to network — in one Teams conversation, plus a feature that auto-schedules the networking coffees that make an internship stick.
An agent as a process step: award letters in under 30 seconds
Most agents are conversational. This one is the opposite — a fully autonomous agent with no chat and no user, turning a nomination in Airtable into a values-aligned, CEO-voiced award paragraph in under 30 seconds.
Reading every survey comment — with the PII stripped out
Open-text comments are the richest and least-read part of an engagement survey. MK Comment Signal turns thousands of them into ranked themes backed by real verbatim evidence, with names, emails, and IDs redacted automatically.
Asking a spreadsheet questions in plain language
Finance needed answers to capital-appropriation questions — remaining budget, spend, ownership — without opening an Excel file and scrolling. Capital Planner answers them in plain language, grounded in the authoritative CAR data.
Turning client-visit surveys into an executive read
A West Michigan client visit is a high-stakes sales moment, and every visit is surveyed — but the feedback sat in spreadsheets no one had time to read. Survey Insights Analyzer turns it into named themes, dimension trends, and NPS, grounded only in the survey files.
A research agent whose POC proved the data was ready
Rosie is a deterministic research orchestrator for Product Innovation. Its most valuable output wasn’t the agent — it was the discovery that MillerKnoll’s end-user research corpus was ripe for a knowledge graph, which moved the work to the Data Science team.
Turning a team’s inbox into its documentation
Many service teams run on a shared inbox, and the know-how lives in replies rather than documents. AISE built a reusable capability that reads an approved shared mailbox, strips out personal data, finds the recurring questions, and drafts the how-to articles and SOPs the team never had time to write. Eight HR mailboxes in, it has produced 224 draft articles.
Making AI spend visible, then letting leaders ask it questions
AI spend at MillerKnoll was spread across six vendors, each with its own portal and billing model. The AI Analytics Dashboard puts all of it on one screen that refreshes daily, and its Data Assistant answers questions about what’s on that screen.
Enablement as a product: every build makes the next one easier
AISE doesn’t only ship agents. Each build leaves behind a standard, a training module, a reusable skill or a capability another team can run, so the next build, ours or anyone’s, starts further ahead.
Not an agent. Not a skill. A practice.
The case study behind all the case studies: pointing AI at a volume of content a lean team could never read by hand — and getting structure back, not a summary.