RELAYDESK / BUSINESS PROCESS AUTOMATION
Every enquiry.
A clear next step.
A working enquiry-management prototype that turns an incoming message into a routed request, an editable reply and a decision someone can review.
BEMRIX concept project · Rules-based automation · Browser-local demo
01 / THE PROBLEM
The message arrives.
Then the manual work begins.
Small service teams often handle the same steps for every new enquiry: read the message, decide who should own it, ask for missing information and remember what happened next.
RelayDesk demonstrates a repeatable first-response workflow. Routine preparation happens immediately; a person reviews the route, adapts the draft and approves the next step.
No invented time savings or conversion results. This is a functional portfolio prototype, not a deployed client system.
02 / THE WORKFLOW
Four steps. One visible process.
Capture
Add a fictional enquiry with a name and message. The workspace accepts up to 40 requests.
Route
Match service keywords, flag possible urgency and explain the original rule results. Mixed or unknown requests go to general review.
Prepare
Start with a service-aware reply template. Change the owner and edit the wording before approval.
Review
Approve the draft locally. Edits require a fresh approval, and the activity log records the transition.
03 / BUILT TO BE TRIED
A useful demo.
On any screen.
Desktop pairs the inbox with the review workspace. On a phone, enquiry cards scroll horizontally and the detail panel follows below.
- Create and classify new enquiries
- Inspect routing reasons and override the owner
- Edit, save and approve reply drafts
- Filter by review status and export JSON
- Resume browser-saved work or reset to samples
04 / HONEST AUTOMATION
Clear rules.
Human judgment.
This implementation uses deterministic keyword rules and templates. It does not use an AI model. Ambiguous requests stay in general review, and urgency is a keyword signal that a person must verify.
Everything stays in this browser’s local storage. No emails are sent, no CRM is connected and the local activity history is not a secure audit trail. Use fictional details when trying it.
A production engagement would add authenticated team access, a database, agreed routing policies, delivery integrations and monitored retries. AI-assisted interpretation could be evaluated against real enquiry examples before being introduced.
Built with Next.js, React and TypeScript. Interface screenshots captured from the working demo.
Let’s make it happen