AI Transformation at Milk By Mom
Milk By Mom freeze-dries breast milk. New moms ship their frozen milk to the lab and receive it back as a shelf-stable powder they can scoop and reconstitute, with the convenience of formula and the nutritional value of their own milk. I have worked with them since their early days, first as their Salesforce architect and now as their AI transformation consultant on an ongoing contract. The mandate is simple: make the business easier to scale.
From Salesforce to a custom ERP
I designed and ran Milk By Mom’s Salesforce instance through the company’s early growth. As order volume and lab operations matured, the business moved onto a custom-built ERP, and when that migration finished my contract changed shape with it. A custom system gives you exactly what you build and nothing else: every new report, integration, or workflow is developer time. My work now is reducing how often the company needs custom development just to answer a question or adjust a process.
Data the team can query in plain language
The first piece was an AWS layer with an RDS database that holds the company’s operational data in a shape Amazon Bedrock can query. Business questions get answered by asking them, and the alternative was opening a ticket for another hardcoded report in the ERP. Each report that no longer needs to be built is developer time the company gets back.
The lab floor
Lab workers scan a QR code at their workstation and their time is tracked against it. The scan costs a worker about a second, and it turns shop-floor activity into structured operational data: who worked which station, when, and for how long. Time tracking that used to be a manual afterthought is now something the business can actually analyze.
Current project: collection insights
The work in flight is on the collection side, structuring the data around incoming milk so the company can see patterns in how and where milk arrives and plan its growth from evidence rather than instinct.
The shape of the engagement
The contract has no defined end date. Milk By Mom gets senior systems work in fractional form, and the backlog is always the same question: what would make scaling easier next?