1
Connect your customer database via MCP
Dana needs access to the database where your customer data lives. Go to Customize > MCPs (Browse marketplace), find your database provider, and click Enable.Common setups for customer metrics:
- PostgreSQL for application data (users, accounts, feature flags)
- Redshift or BigQuery for analytics warehouse data (usage events, revenue)
- Snowflake for combined operational + analytics data
2
Create a customer health skill
Go to Customize > Skills, click + (New skill), and name it
customer-health; you invoke it from Slack as @skills:customer-health.Define the metrics Dana should pull for any given account. Keep the skill general — Dana will figure out the right tables and queries from your schema:You can also ask Devin to generate a customer health skill tailored to your specific schema and metrics.3
Run it from Slack
Make sure you’ve connected the Slack integration in Settings > Connections and that Devin is added to the channel where your team discusses customer accounts.Invoke the skill using the
!dana keyword and the skill mention:Dana queries your connected databases, runs the SQL from your skill, and replies in-thread with a formatted summary:Acme Corp — At RiskYou can follow up in the same thread to drill deeper:Risk factors: MAU declining, no expansion in 3 months, 3 recent tickets about login failures.
4
Make it routine
Once the skill works well for ad-hoc lookups, consider automating it:
- Scheduled automations: Use an automation with a weekly Schedule trigger to have Dana post a health summary for your top 10 accounts every Monday morning into a
#customer-healthchannel. - Pre-meeting prep: Before a quarterly business review, ask Dana to pull summaries for all accounts in a specific tier —
@Devin !dana !customer-health Run this for all accounts where plan = 'Enterprise'.

