Your business runs on exceptions. Your software should too.

Every ag retailer has processes that exist nowhere in their ERP. The "if it's over $10K, call the branch manager" rule. The "always email the grower when their prepay drops below $5K" workflow. The "flag any PO from this vendor if the price is more than 5% above last order" check. These are the rules that actually run your business — and they live in people's heads, not in your software.

You've probably tried to solve this with custom development. Maybe you paid a consultant $30K to build a workflow that worked for six months until someone changed the approval hierarchy and nobody updated the code. Maybe you cobbled together Zapier, email rules, and shared spreadsheets into a Rube Goldberg machine that technically works but nobody fully understands.

That's not a workflow system. That's technical debt wearing a disguise.

Workflows should be as flexible as your business

The ag supply chain doesn't follow a playbook. Seasons shift, vendors change terms, growers switch buying patterns, new regulations drop mid-year. A workflow that was perfect in January might be wrong by March. If changing a workflow requires a developer, a sprint, and a deploy cycle, your workflows will always lag your reality.

In Horizen, workflows are built by the people who need them. Not with code — with plain language. Tell the AI agent: "When a sales order over $10K is created, require branch manager approval. If not approved within 24 hours, escalate to the regional director." The agent builds it, activates it, and monitors it. Want to change the threshold to $15K next quarter? Tell the agent. Done.

Custom tables: track what your ERP can't

Here's what nobody talks about in ERP demos: every business has data that doesn't fit the system's schema. Maybe you track equipment rentals alongside product sales. Maybe you manage applicator certifications and need to know which applicators are licensed for restricted-use products in which counties. Maybe you track grower field maps, soil test results, or crop plans that tie back to orders but aren't "orders" themselves.

In most systems, this data ends up in spreadsheets. Disconnected from the transactions it relates to, invisible to the rest of the team, and impossible to report on.

Horizen's custom entities let you create tables that live inside the system as first-class objects. Define the fields you need, relate them to existing entities — customers, products, orders, locations — and they become part of the data model. They show up in views, they're searchable, they're reportable, and the AI agents can read and write to them.

If your team is tracking it in a spreadsheet, it matters enough to be in the system. Custom entities make that possible without a single line of code.

The agent can manage them too

This is where it gets interesting. Custom entities aren't just passive data stores — the AI agents can interact with them. "Update the applicator certification for John Miller — he's now licensed for restricted-use in Boone County through December 2027." The agent updates the record, links it to the contact, and flags any open orders that require a licensed applicator in that county.

"Show me all growers who haven't submitted soil tests for this season but ordered more than $50K last year." The agent queries across your custom soil test table, your customer records, and your order history, and gives you a list with contact info — ready for your sales team to follow up.

This is the power of custom entities inside a unified system. The data isn't siloed in a spreadsheet. It's connected to everything else and queryable by the same AI agents that manage your orders, inventory, and finances.

Workflows + custom entities = your operating system

When you combine custom workflows with custom entities, Horizen stops being software you use and starts being the operating system your business runs on. Approval chains, notification rules, escalation paths, data capture — all configurable, all connected, all managed through the agent layer.

A workflow fires when a custom entity changes: "When an applicator certification expires, notify the branch manager and hold any pending orders requiring that certification." A custom entity updates when a workflow completes: "When a grower signs a new prepay agreement, create a record in the prepay tracker and set the review date for 90 days out."

Your business logic lives in the system, not in someone's head. And when that person leaves — which they will, eventually — the business keeps running.