How a Food Distributor Cut Order Processing Time by 70% with AI Agents
A regional food distributor automated order intake, inventory queries, and supplier coordination using Oido AI agents — reducing processing time from hours to minutes.
The distributor's team of 12 spent most of each day manually processing incoming orders via email, phone, and WhatsApp. Each order required checking inventory across three warehouses, confirming pricing, and coordinating with logistics. Errors from manual data entry caused 6% of orders to require rework, and the team was consistently 48 hours behind on order confirmation.
We deployed three Oido agents that work together: one watches the shared email inbox and WhatsApp Business channel for incoming orders, extracts structured order data, and confirms stock availability via an MCP server connected to the distributor's ERP. A second agent handles supplier coordination — checking lead times and placing replenishment orders when stock drops below thresholds. The third agent sends order confirmations back to customers over their original channel, with estimated delivery windows pulled from the logistics system.
The Problem
The distributor operated on thin margins typical of food distribution — speed and accuracy were everything. But their order intake process was entirely manual:
- Email orders — PDF attachments and free-text emails that a coordinator read and typed into the ERP
- Phone orders — Calls taken during business hours, written on paper, then entered hours later
- WhatsApp orders — Long-standing customers messaging orders directly to sales reps, who forwarded them to operations
Every handoff introduced delay and risk of error. The 48-hour confirmation lag meant customers couldn't plan their receiving schedules, and the 6% error rate triggered frequent emergency deliveries that ate into margins.
The Solution
We designed a three-agent system that mirrors the human workflow but runs continuously, across channels, with no handoff gaps.
Agent 1: Order Intake
This agent monitors the shared inbox and WhatsApp Business number. When an order arrives:
- It extracts structured data from email text, PDF attachments, or WhatsApp messages
- It queries the ERP via a custom MCP server to verify stock availability across all three warehouses
- If stock is sufficient, it reserves inventory and generates a provisional order
- If stock is short, it alerts the supplier coordination agent
Agent 2: Supplier Coordination
Running on a schedule, this agent checks inventory levels against reorder thresholds. When an item drops below its threshold, it:
- Queries supplier lead times from the ERP
- Generates a purchase order
- Sends it via the supplier's preferred channel (email or Telegram)
- Logs the expected delivery date back into the ERP
Agent 3: Customer Confirmation
This agent takes the structured order from Agent 1, pulls estimated delivery windows from the logistics schedule, and sends a confirmation back to the customer over their original channel — all without a human touching it.
The Technology
The stack was chosen for reliability and minimal disruption to existing systems:
- Oido Studio — hosted the agents, managed routing, and provided the unified interface
- n8n — handled the deterministic ERP sync (orders must be written exactly once, with transaction integrity)
- Custom MCP server — wrapped the legacy ERP's REST API so the AI agents could read inventory and write orders
- Postgres MCP server — let agents query the replicated inventory database directly for real-time stock checks
- WhatsApp Business API + Telegram — provided the channel interfaces customers already used
The Results
The system went live in staged rollouts over 6 weeks. Within 3 months of full deployment:
| Metric | Before | After |
|---|---|---|
| Daily order processing time | 3.5 hours | 45 minutes |
| Order rework rate | 6% | 0.4% |
| Customer confirmation time | 48 hours | < 5 minutes |
| Customer satisfaction | 82% | 94% |
| Errors per 100 orders | 4.2 | 0.3 |
The cost savings from reduced rework and reallocated headcount delivered full ROI in under 4 months. The distributor has since expanded the system to handle invoicing and route optimization.