Buy 14 t of corn from Supplier 3, not 7.
- Never Supplier B on Mondays
- Arrives before you run out

Buy 14 t of corn from Supplier 3, not 7.
Every call comes from three layers: a forecast, an optimizer and a translator. Here's what each one does, with simple examples.
Buy 14 t of corn from Supplier 3, not 7.
We connect to the systems you already use. Your ERP, your transport system, the planning spreadsheet someone updates every Monday. From them we build one map of your operation: plants, lines, warehouses, products, suppliers, routes and customers, and how they connect.
Then we add what isn't written down anywhere. Today those rules live in your planner's head, or on a sticky note under the monitor. On the map, they count in every decision.
Every company gets its own map.
| PLANT | PRODUCT | SUPPLIER | ROUTE |
|---|---|---|---|
| Plant 1 | Yellow corn | Supplier A | Highway 45 |
| Plant 2 | Yellow corn | Supplier B | Mountain road |
| Plant 2 | Soybean meal | Supplier 3 | Highway 45 |
| Plant 3 | Sorghum | Supplier 7 | Coast road |
We bring in what's happening outside your company. Corn and other input prices, diesel, freight rates, the exchange rate, the weather and demand.
Each signal connects only to what it actually moves. A signal that touches nothing you make or ship gets ignored.
Your data goes in at the bottom and a plain sentence comes out the top. The same data gives the same call, and you can always see which data and which rules it came from.
Your team sets the rules and Intelital follows them. Some can never be broken: a plant can't make more than it can make, pack more than its lines can pack, or make a product it isn't set up for. Any option that breaks one is out.
Others come from your planners as they use Intelital. Say it once and it becomes a rule, with who said it, when, and what it applies to. Your team can remove it whenever they want. When that planner leaves, the rule stays.
No Supplier B for Monday deliveries.
Each record keeps what the situation looked like at that moment, what Intelital said and with which model version, what your person or agent did and why, and what happened afterward.
A person approves any call that's too uncertain. We measure every decision in dollars.
| WHEN | INTELITAL SAID | WHAT HAPPENED | WHY | OUTCOME |
|---|---|---|---|---|
| Mon 07:02 | Buy 14 t of corn from Supplier 3 | Purchasing agent · accepted | Cheaper this week, arrives in time | +$12,400 |
| Tue 09:10 | Send 2 trucks, not 3 | Maria · changed it to 3 | A big customer just called | Delivered on time |
| Wed 11:20 | Run line 3 on Thursday | Planner · said no | Silo B was running low | MISSED $2,100 OVERTIME |
It sends what it's about to do. It gets back the best move, the odds, what's at stake, what it costs to be wrong, and whether to go ahead or wait for a person. What it did and how it turned out go into the same record as a person's decision.
Agents will be able to call Intelital over MCP, so any agent framework can use it as a tool without custom work.
PURCHASING AGENTShould I order 18 tons of corn from Supplier 7?
Order 14 t from Supplier 3.
AGENT → INTELITALPOST /v1/decisions{ "decision_type": "purchase", "proposed_action": { "item": "yellow-corn", "quantity_tons": 18, "supplier": "SUP-07" }} INTELITAL → AGENT 200 OK{ "recommendation": { "quantity_tons": 14, "supplier": "SUP-03" }, "odds": 0.82, "expected_impact_usd": 12400, "cost_if_wrong_usd": 3100, "rules_met": true, "needs_approval": false, "reasoning": "Cheaper on this route this week and still arrives before you run out."} AGENT → INTELITAL WHAT HAPPENEDPOST /v1/decisions/dec_8f3a21/outcome{ "action_taken": "accepted" }
