Custom ERP Systems

At the size of our clients, generally 150 people at most, ERP stops meaning much. It's three letters that cover so many things they lose meaning, and the real question is what the company does. Three of ours:

A nationwide event caterer ran sales in a CRM and operations in spreadsheets. Nothing was live. You talked to a client, roughly sketched the cost in a few spreadsheets, checked another spreadsheet for who had what equipment (some things can't be in two cities at once), then held a weekly meeting to plan resources, only to throw the plan away when the client changed their mind or didn't accept the proposal. Every part of the company was involved, and no part of it had the full picture at any point. The quality of the work was fantastic; the redo behind it was the problem. Now a service added to a deal computes every supply quantity, the staffing, per diem, ground transportation, crate shipping by destination, permits and fees into one live cost summary, the warehouse gets its pick list from the deal itself, and every crate and asset is barcoded down to zone, row and level.

A school-meals producer ran on an old Access database and spreadsheets. Finding the cheapest source for each item was manual, collecting vendor prices was slow, and proving meals met USDA meal patterns was tedious. Now bid sheets go out as protected spreadsheets and come back through a staged price import with history, every ingredient carries a lowest-price-vendor flag, nutrition and CN credits roll up from item to portion to meal to week to menu with pass-warn-fail flags, and the production reports, pull lists, shipping labels and allergen matrices that were hand-built each morning come out as one-click reports.

A custom cutting-tool manufacturer ran on a legacy on-premise ERP. Reading the state of the business meant the brain trust analyzing a snapshot of two or three hundred work orders, which took negotiation and archaeology before anyone could get to the decisions. The effort went into making the machine run, not making it run well. Now a sales order line creates the work orders from the tool's engineered build, operators clock in and run operations from a shop-floor screen, statuses roll up from operation to order automatically, outside processing is tied to the exact operation it serves, and the general ledger is in the same system. Since then they've absorbed retirements, departures, new hires and new production processes.

None of these companies would describe themselves as needing an ERP. They manage inventory, decide between employees and temporary labor, decide whether a piece of equipment flies or ships or stays local, track what's hot and what's cold and what gets finished on site. Same tools, same processes, in their own vocabulary. More on the manufacturing side at Manufacturing & Distribution.

The backend, the user interface, and the basics are the same across clients: ticketing, document management, notes, the reporting system, the calendar. Every client then adds their own fields, their own workflows, and dashboards put together from existing components. How deep a component goes depends on how central it is. Every system gets a calendar. If scheduling matters a little, it's a display with some information on it. If scheduling is the business, it drags and drops, updates other calendars, and runs a workflow of its own. Scheduling equipment and people looks the same at a caterer as it does at a company that makes high-speed cutting tools: certain people can do certain things, and there's only so much equipment.

The parts that are the same aren't one piece of code everyone shares. They're built on a platform the client configures: adding fields, moving fields, adding reports, adding a widget to a dashboard or a whole new dashboard or page. All of that stays under the client's control after go-live, without calling us.

AI runs in production today. For the producer, supplier nutrition and formulation PDFs are read and their fields filled automatically, with in-app alerts. For a services client, ticket summaries and 30- and 90-day client reviews write themselves. The platform exposes an MCP endpoint so an assistant can ask ad hoc questions of your data; the scheduled workflows do the heavier work a chat assistant wouldn't. More in MCP for business systems.