Relayflow
A concept automation hub that connects a CRM, invoicing, and support tools, and draws every process as a line on one map.
ConceptFictional project. No real client, and nothing here is a measured result.
- Sector
- Business operations
- Service
- Integration & automation
- Year
- 2026
Lines
- NNew customerCRM to invoicing
- IInvoice to paidQuote to payment
- SSupport to accountSupport desk to hub
- LLedger exportHub to accounts, nightly
Connected tools
- CRMWebhook
- InvoicingWebhook
- Support deskSync, 5 min
Needs a look
IStop 4 of 6
Send invoice
Runs when a draft is approved. Sends through invoicing and waits for payment.
Field map
deal.amountinvoice.totalcontact.emailinvoice.todeal.termsinvoice.duecompany.tax_idinvoice.tax
If this stop fails
- Retry5 times, backing off
- Then alertAccounts
Runs by hour
- 09:41INV-2041IPayment inRetry 2 of 5
- 09:42D-3318NWelcome sentOn time
- 09:44T-882SFind accountOn time
- 09:47Q-1187IDraft invoiceOn time
In this concept
Relayflow is a concept project. There is no real client behind it. We made it to show how we would connect the tools a growing business already pays for. Everything below describes design goals, not measured results.
The problem
Most small companies run on three systems that do not talk to each other. Sales lives in the CRM. Money lives in the invoicing tool. Customer problems live in the support desk.
People fill the gaps by hand. Someone copies a signed deal into an invoice. Someone checks a spreadsheet to see if a customer who just opened a ticket has an overdue balance. It works until that person is on holiday.
The usual fix is a pile of one-off automations. Each one does its job, nobody can see them all at once, and when one breaks quietly the first warning is an angry email.
The approach
We would start by drawing what actually happens, not what the process document says. Sit with sales, accounts, and support, follow a few real customers from signed deal to paid invoice, and mark every point where someone retypes something.
Three decisions would shape the rest of the work:
- One record in the middle. Every tool reads from and writes to a single customer record in the hub, so each new tool connects once rather than to every other tool.
- Draw it like a transit map. Each process is a coloured line and each step is a stop. The lines meet at the hub. Anyone in the company can read it, including people who will never open a settings page.
- Failure is a normal state. Every stop has a retry rule and a named owner. A stuck record waits at its stop in plain sight until someone or something moves it on.
What we would build
- Connectors for the CRM, invoicing, and support tools, using their webhooks where they exist and scheduled sync where they do not.
- A hub service that holds the shared customer record and keeps a full history of each run.
- A durable queue, so a run survives a tool going down and picks up where it stopped.
- The flow map: a live view of every line, stop, and record in transit.
- Field mapping that shows which value goes where, with a test run on sample data before anything goes live.
- Alerts that go to the person who owns the stop, with the record and the reason attached.
Goals, not results
The goals we would design against: nobody retypes a value that already exists in another tool, a failed step is visible the moment it fails, and a new team member can explain the whole operation from one screen.
Key screens
Flow map
Every process on one panel. Lines run from their source tool into the relay hub and out to where the work finishes. Records move along the lines as they are processed.
Stop detail
Select any stop to see its field mapping, its retry rule, who owns it, and how it has been running through the day.
Departures
A plain list of what runs next and what is waiting: the time, the record, the line, the next stop, and the status. It is the first place to look when someone asks where an invoice went.