Ingest
Website forms, ad lead sources or imported data in an agreed format.
KEEP / Source, timestamp and reference IDTECHNOLOGY / CONNECTED WITH PURPOSE
Tools connected.
Teams in control
of how it runs.
An automation flow must make clear where data comes from, how it is processed and who steps in when something goes wrong. That is where ZvZ's technical design starts, both in ZvZ AI Hub and in client implementations.
Follow one data flow01 / ARCHITECTURE
From one signal.
To one next action.
The diagram below shows one way to design lead intake. Platforms and steps are adapted to your workflow, access rights and data.
Validate
Check required fields, normalize values and detect whether the record was already processed.
ROUTE / Valid or needs reviewHand off
Create or update the record in the target system, assign an owner and send the right notification.
CONFIRM / Status in target systemMonitor
Log outcomes, track errors and reconcile source data with the team's handling status.
REVIEW / Sync, intake and outcomes02 / DESIGNED FOR OPERATIONS
When it works.
And when it fails.
Cases to agree on and test before a flow goes live.
Missing or duplicate data
Records missing required fields move to a review state. Duplicate rules decide whether to update an existing record or create a new one.
Resubmissions of the same request are tested to avoid duplicate tasks or repeated notifications.
Target system not responding
The flow distinguishes success, pending and failure. Retry logic, retry count and alert recipients are defined per connection.
When the outcome is unknown, data is reconciled before resending.
Changes need an approver
AI analysis or automated rules operate within agreed limits. Budget caps and execution conditions are set upfront.
Suggestions beyond that scope go to a human owner.
03 / CHOOSING TOOLS
Fit the systems
you already have.
n8n, Make, CRMs and ad tools can each handle different parts of a solution. The choice depends on required connections, data volume, operators and running costs.
For ZvZ AI Hub we use Claude Code to build and test these integrations today, and we plan to use the Claude API for lead scoring and follow-up drafting.
Before finalizing scope, we check versions, plans, API permissions and limits of each platform. Unverified connections are flagged in the plan.
Make vs n8n (article in Vietnamese) ↗Things to agree on
- Where source data lives and which system receives results.
- Who owns tool accounts and who grants integration access.
- Which data fields are used, for what purpose and for how long.
- How credentials are managed, revoked and handed over.
- Platform costs, usage limits and who handles changes.
04 / HANDOVER
Understand the system.
Know the next step.
Diagrams & rules
Data sources, processing steps, target systems, branching conditions and points needing human input.
Test results
Cases tested, reconciliation results and conditions that still depend on platforms or data.
Operating guide
How to check status, spot errors, handle common cases and reach the owner.
FROM DIAGRAM TO YOUR WORKFLOW
Are your tools
talking to each other?
Tell us which platforms you use and which steps are still manual.