Enterprise Cuts Workflow Time 80% with AI Agents
A corporate operations team automated inter-departmental handoffs, procurement, IT approvals and HR onboarding, 5 days down to under 24 hours.
The company's internal processes relied on email-based request chains. A procurement request would start in Finance, route to a department head for approval, then to Procurement for sourcing, then back to Finance for payment, each hop taking 1-2 days. Employees had no visibility into where their request was in the pipeline, and the operations team spent 60% of their time chasing down stalled requests.
We deployed a multi-channel Oido agent accessible via Slack and Telegram that handles the full lifecycle of internal requests: intake via natural language, routing to the correct approver, follow-up escalation on stalled steps, and status reporting. Each request type maps to a workflow in n8n that enforces business rules and SLA deadlines.
The Problem
The enterprise had grown to 400 employees across 4 offices, but its internal processes hadn't scaled. Every department had its own request process:
- Finance, purchase requests via email form, approved via email chain, eventually entered into the ERP
- IT, access requests via a shared inbox, manually assigned, tracked in a spreadsheet
- HR, onboarding tasks coordinated via email CC chains with frequent dropped handoffs
The common pattern: request submitted → sits in inbox → forwarded to next person → sits again → forwarded again. Average end-to-end time was 5 days, and 15% of requests were abandoned entirely.
The Solution
We deployed a single Oido agent accessible in Slack (default) and Telegram (for mobile staff). The agent handles all request types through a unified conversational interface.
How it works
An employee opens a chat with the agent and describes what they need:
"I need a new laptop for a developer starting next Monday."
The agent identifies the request type (IT equipment procurement), extracts key data (developer role, start date), creates a ticket in the workflow system, routes it to the department head for approval, and sends the employee a confirmation with a tracking link.
Behind the scenes, n8n orchestrates the deterministic steps:
- Create request record in Postgres
- Determine approval chain based on request type and amount
- Send approval request to the first approver via Slack
- If no response in 4 hours, send a reminder
- If no response in 8 hours, escalate to the approver's manager
- Once approved, route to the fulfillment team (IT, Procurement, HR)
- Notify the requester at each status change
Escalation logic
The agent tracks every active request against SLAs. Any request that stalls at a step for longer than its SLA generates a summary report sent to the operations team each morning. Critical delays trigger immediate alerts.
The Results
| Metric | Before | After |
|---|---|---|
| Average resolution time | 5 days | 18 hours |
| Requests processed (Q1) | ~800 | 1,200+ |
| Request abandonment rate | 15% | 2% |
| Employee satisfaction | 58% | 89% |
| Operations team time on chasing | 60% | 12% |
| SLA compliance | 40% | 95% |
The system has since been expanded to handle expense report approvals, contractor onboarding, and inter-office equipment transfers.