A lean team, a large portfolio, in under a year.
The measurable side of AI Solutions Engineering — what’s in production, what it has replaced or avoided, how fast it got here, and who’s using it. Every figure below is sourced.
A $10M federal opportunity we couldn’t have competed for before
A USACE solicitation for Scott Air Force Base required proving every quoted price against MillerKnoll’s GSA schedule — roughly 3,115 pages of published-page evidence across 108 workstations, in the exact order the government demands, due before the holiday. That manual burden is why federal bids like this often can’t be pursued directly. AISE built the tooling that assembled the evidence in order, flagged mismatches rather than guessing, and turned hundreds of hours of work into a response submitted on deadline — putting MillerKnoll in the running for a ~$10M opportunity that wasn’t reachable by hand. Federal is the focus today; state and local government work is the next scope expansion.
The whole portfolio, two ways
Where every agent sits in the governed lifecycle, and which parts of the business they cover.
By delivery stage
POC → Development → Test → Production, under managed promotion.
By business function
Agents built per area — breadth across the enterprise.
Built in a short time — on purpose
Key moments across the portfolio, in order. Each one names the project it belongs to.
Our first agent goes live
IT Service Desk technicians start getting answers from vetted knowledge across every brand, instead of searching for them. How we built it became the playbook for everything after.
We choose to build, not buy
A vendor tool for pricing custom furniture requests would have cost about $200K a year. The team showed in days that we could build it ourselves, based on our own pricing rules.
The team forms
A small team is set up to take business problems all the way to working, supported AI tools.
In the pricing team’s hands
The first version goes to the people who price custom requests. They test it on their real work and keep the rules behind it up to date.
Nine agents in production
Agents are now in daily use across IT, Customer Care, HR, Finance and project intake.
Live for every custom pricing request
Every request from dealers now gets an automatic first answer. It went live on a Saturday with no lost requests, on the platform that was about 30% cheaper to run.
Live for manufacturing developers
Developers can ask plain-language questions about a 2,191-page technical guide and get a precise, sourced answer.
AI spend and usage in one place
Leaders can see what AI costs and who is using it across every vendor, and ask questions about it directly.
The first month’s results
Its answers matched our pricing team’s decisions 91.7% of the time. Dealers get quotes back in about 26 hours instead of 72, and a new Solutions Desk books time with dealers to talk requests through.
A lean team, a large portfolio, in under a year.
AI Solutions Engineering is a genuinely small team — three engineers, who had two summer interns ship flagship work alongside them. In under a year it has stood up 30+ agents across IT, HR, Customer Care, Manufacturing, Specials/Pricing, Finance and more — with 13 already carrying real work in production and the rest moving through a governed POC → Dev → UAT → Production pipeline.
Real teams, using it now
Measured from the agents’ own conversation records and AutoBid’s production telemetry.
We point the AI at our own work, too
Reading, checking and organizing at a scale a small team couldn’t do by hand. Each one links to the story behind it.