zeroleftQ: How many runs in the following workflow
How don't know how to choose the tier. I need to know the number of runs for an example. I assume that I can use BOYK if I need more so I can start with tier 1. Workflow: pick values from a table on supabase and post them on mattermost. I handle this with n8n but it is becoming expensive. How do you compare here to n8n and how do you handle triggers, change in supabase table values (new rows or updates to rows) via webhooks don't work on n8n, what about yours?
SeanP_AgenticFlowAI
May 16, 2025A: Hey Zeroleft,
Let's break this down for your Supabase -> Mattermost flow!
Credit Usage Example (Tier 1 = 4 fixed credits/step):
Workflow:
Fetch from Supabase (API Call/MCP): 4 credits
Post to Mattermost (API Call/MCP): 4 credits
Total per Run: ~8 AgenticFlow credits (fixed). If you add an LLM step using our built-in models, that adds variable credits.
Tier 1 (7,500 credits): Roughly 10,000 / 8 = 937 runs/month for this simple flow, before considering BYOK or any built-in LLM variable costs.
BYOK Impact: Yes, using BYOK for any LLM steps significantly stretches your AgenticFlow credits, as you only pay us the fixed step cost for that LLM action.
AgenticFlow vs. n8n & Triggers (Supabase Table Changes):
Comparison:
Cost: LTD can be more predictable than n8n cloud or self-hosting costs.
AI-Native: AgenticFlow is built for easier AI integration if your flow needs it.
Connectivity: n8n has many direct nodes; AgenticFlow uses MCPs (2500+) and robust API Call nodes.
Triggers for Supabase Changes:
Webhooks (Ideal): If Supabase can send a webhook when a table changes (new/updated row), AgenticFlow can receive this via its API Endpoint trigger. This is instant and efficient.
Polling (If no webhooks): If Supabase doesn't offer change webhooks, you'd schedule an AgenticFlow workflow (currently via external cron + API call, native scheduler on roadmap: https://agenticflow.featurebase.app/p/scheduler-node-to-trigger-workflow-on-interval). This workflow would query Supabase (via API Call node or a Supabase MCP if available: https://agenticflow.ai/mcp/supabase) to check for changes since the last run.
Reliability: Our API endpoint for webhooks is designed to be robust. If you had issues with n8n's webhook handling for Supabase, ours might offer a more stable experience, but it always depends on Supabase sending the webhook reliably too.
Tier Choice:
Tier 1 is a good starting point if the volume of Supabase changes is moderate and you plan to use BYOK for any significant AI processing. If you have many daily changes or want lower fixed costs per step (3 credits on T3/4), a higher tier becomes more economical. The webhook method is always more credit-efficient than frequent polling.
Hope this helps you decide!
ā Sean