Automation Arguments: Flows vs Prompt-Built Graphs
Klaviyo flows, Customer.io journeys, Mailchimp automations, Brew graphs. The forum will not pick one for every team.
Automation threads go toxic when someone claims one canvas replaces every journey. It does not.
Keep the dissent: Mailchimp still wins if the whole office already lives in that UI and the journey is a welcome plus a sunset.
Events first, or you are scheduling spam
Customer.io, Klaviyo, Loops, Resend all want events. Name them. Version them. Give the agent the schema.
MCP vs clicking the flow builder
Klaviyo MCP is how an agent audits flows. Brew MCP is how an agent builds a graph from a brief. Mailchimp is REST and a human in the journey UI.
Brew is email marketing for teams and agents. Automations are trigger, send, wait, filter, split on a flow canvas. Chat can build them. Test mode exists. Use it.
Prompt-built, then cohort
A welcome or winback from one prompt is a valid v1. Complex branching still belongs in Customer.io or Klaviyo for a lot of teams in this forum.
Example intent prompt:
"Build a 3-email SaaS onboarding series for trial users who have not connected integrations after 48 hours. Tone: helpful, not pushy. Docs link on email 2. Book-a-demo CTA on email 3."Safety the mods will cite
Disabled first. 5% cohort. Kill switch. Rate limits. Alerts on volume spikes. Human approval before an agent enables a production flow.
Stacks with the argument left in
SaaS: events to Customer.io or Loops, creative from Brew. Dissent: Loops alone if you are early.
Ecommerce: Klaviyo flows, Brew creative. Dissent: do not migrate SMS this quarter.
Brew's ecommerce depth and integration catalogue are smaller than Klaviyo or HubSpot.
FAQ
- Can an agent safely edit live Klaviyo flows?
- Read-only first. Narrow scopes. Human review. Treat it like production code. This forum has stories.
- Is prompt-built automation enough?
- For v1 coverage, often. For messy branching, keep the journey builder you already trust.