Customer story
A PE-backed aerospace and defense supplier put agents to work. No migration.
A multi-site Tier 1 supplier ran on an on-prem ERP, spreadsheets, and email, with a decade of judgment living in one person's head. In months, they had real-time apps and agents across operations, finance, and estimating, running on the systems they already trusted.
8 hrs -> 15 min
Monthly operating review prep and analysis
~20 hrs / week
Estimating and quoting work now automated
1-2 weeks -> real-time
Delivery and production reporting lag
The situation
Real systems, held together by spreadsheets and email.
ERP for finance and supply chain, an MES on the floor, and a PLM for engineering, each a system of record that did not talk to the others. The connective tissue was Excel, Outlook, and SharePoint, and the know-how for pricing and planning lived in a handful of spreadsheets and one person's head.
There was no appetite to rip out systems the business already trusted. Most AI on the market would have started with a migration. For this operation, that was a non-starter.
What we built
Agents inside the stack they already ran.
Real-time apps and agents
Automated roughly 20 hours per week of manual work into live apps and agents across finance, operations, and estimating, wired to the ERP source of truth.
A platform that scales across sites
A single platform for a multi-site operation to roll agents out site by site, without ripping out the systems each site already trusted.
A team trained to run it
Trained the internal team toward a Center of Excellence, train-the-trainer, so the customer builds and owns new use cases over time.
Compliant by architecture
Delivered on a FedRAMP-capable architecture (GCP), so the pilot could survive IT and security review in a regulated defense environment.
Before and after
Automate the analysis, sharpen the insight, act faster.
The same people, the same systems, but the manual chasing turned into real-time evidence they could act on.
Operations
Before
Static lean charts and boards. A one-to-two-week lag before production and delivery results could be analyzed for downtime and quality.
After
Real-time apps at each value stream showing work orders and delivery metrics, plus schedule-attainment dashboards connected directly to the ERP source of truth.
Finance
Before
The monthly operating rhythm meant manual analysis of financial statements against operations metrics. Profitability was only granular by customer.
After
SKU- and program-level profitability available through apps and agents. MOR prep and commentary dropped from 8 hours to 15 minutes.
Estimating and sales
Before
Quoting was manual, with a decade of tribal knowledge in OneNote and little systematic estimating.
After
Long-term-agreement variance is visible instantly (roughly 20 hours per week automated), and 10+ years of quoting is aggregated into an estimating insights app.
How the engagement worked
Land and expand. Earn trust before scaling.
Start at a single site with urgent, high-payoff problems, prove the value, then hand the platform to a team that can run it. The methodology behind this is our proof of value.
Forward deployed engineer
Sit at a single site, baseline it, get IT and operations ready, and build the first high-payoff use cases alongside the people who own the work.
Consultant and coach
Roll out and scale the initial use cases, coach and train the team, and lay out a roadmap of the deeper, cross-functional problems to tackle next.
Support
The site owns its own delivery model and trains the next users. We move to office hours and a quarterly leadership report-out.
They kept their stack. The work got faster.
No migration, no new system to learn, and a full audit trail of every action the agents took. The legacy ERP, the spreadsheets, and the email all stayed exactly where they were, and the manual work turned into real-time evidence the team could act on.
Bring your hardest workflow.
We’ll work it through with you.
Every conversation starts with your problem, not our pitch. Book 30 minutes and we’ll explore whether myai fits.