How can we generate weekly exec-ready reports from Snowflake/BigQuery/Databricks and post them to Slack without an analyst doing it manually?

Most teams handle this with a weekly ping that looks like: “Hey, can someone pull last week’s ARR, churn, and ticket volume from Snowflake and drop a summary in #exec-updates?” Then an analyst scrambles through Snowflake/BigQuery/Databricks, pastes charts into a doc, rewrites it in exec-speak, and finally posts it to Slack. The data is good—but the process is pure busywork.

Quick Answer: You can generate weekly, exec-ready reports directly from Snowflake, BigQuery, or Databricks and auto-post them into Slack by wiring a scheduled Gumloop Data Analysis Agent to your warehouse and Slack workspace. The agent runs your queries, summarizes the results in clear executive language, formats charts/tables, and posts a clean brief into the channel you choose—no analyst in the loop.

Why This Matters

If leadership needs the same metrics every Monday and those reports depend on an analyst, you’ve built a human cron job. It doesn’t scale, it’s error-prone, and it delays decisions when the one person who “knows the SQL” is busy or out. A reliable, automated report pipeline from Snowflake/BigQuery/Databricks into Slack keeps execs informed and frees analysts to focus on actual analysis instead of copy-paste and chart formatting.

Key Benefits:

• Analysts stop doing manual pulls: Replace “can you run that query again?” with a scheduled agent that hits Snowflake/BigQuery/Databricks, runs pre-defined logic, and ships a finished report to Slack.
• Execs get consistent, readable updates: The same metrics, same structure, and clear narrative every week—no hunting through dashboards or raw tables.
• You keep governance and control: Use Gumloop’s model controls, audit logs, and role-based access to ensure data access is safe and compliant while automation runs in the background.

Core Concepts & Key Points

Concept	Definition	Why it's important
Data Analysis Agent	A Gumloop agent configured to connect to Snowflake/BigQuery/Databricks, run SQL or parameterized queries, and interpret the results.	It’s the “analyst brain” that can turn raw tables into trends, deltas, and English summaries.
Scheduled Tasks for Agents	A Gumloop feature that runs agents on a recurring schedule (e.g., every Monday at 8am) without manual triggers.	This replaces the human calendar reminder; reports show up in Slack reliably, even when no one remembers to ask.
Slack Posting Workflow	A Gumloop Workflow that takes the agent’s output and posts a structured message (with text, tables, links) to specific Slack channels or threads.	Execs see finished artifacts where they already work—Slack, not yet another dashboard to log into.

How It Works (Step-by-Step)

At a high level, you’re doing three things: (1) telling Gumloop what to pull from Snowflake/BigQuery/Databricks, (2) teaching a Data Analysis Agent how to summarize it, and (3) scheduling that agent to post into Slack every week.

Here’s how that looks in practice.

1. Connect Snowflake/BigQuery/Databricks and Slack to Gumloop

   • In Gumloop, add your warehouse as a data source (Snowflake, BigQuery, or Databricks) using scoped, service-style credentials.
   • Connect Slack so Gumloop can post messages into specific channels (e.g., #exec-updates, #gtm-leadership).
   • Use role-based access control to ensure only appropriate agents and workflows can access the warehouse connection.

2. Build a Data Analysis Agent with your core queries

   • Create a Data Analysis Agent in Gumloop.
   • Define the “job” clearly in the agent instructions, e.g.:
     • “Every week, pull MRR, new logos, churned logos, active users, and NPS from Snowflake. Compare to the previous week and previous month. Call out significant changes (>5%) and provide 2–3 bullet insights.”
   • Attach tool calls that let the agent run SQL (or call stored procedures/views) against Snowflake/BigQuery/Databricks.
   • Point the agent at pre-defined, version-controlled queries (e.g., views like weekly_kpis, product_usage_summary) so you’re not embedding fragile SQL directly in prompts.
   • Enforce model choices and restrictions: pick the LLM(s) you want this agent to use and apply any spend or access policies via Gumloop’s model controls.

3. Wrap it in a Workflow and schedule it to post in Slack

   • On Gumloop’s canvas, create a visual Workflow (formerly “Flow”) that:
     1. Triggers on a schedule (e.g., Mondays at 7:30am PT).
     2. Calls the Data Analysis Agent to fetch and analyze warehouse data.
     3. Formats a clear, exec-facing report:
        • Headline metrics (e.g., “MRR: $X (+Y% WoW, +Z% MoM)”)
        • Tables for key KPIs (ARR, churn, NRR, signups, NPS, ticket volume)
        • 3–5 bullet insights + risks/opportunities
        • Links to deeper dashboards in Looker/Mode/Tableau if you have them.
     4. Posts the output directly into Slack via the Slack integration.
   • Choose the Slack destination:
     • #exec-updates for leadership
     • #product-metrics for product teams
     • #support-health for ops/support
   • Turn on Scheduled Tasks for Agents so this runs weekly without anyone tagging the agent manually.

Common Mistakes to Avoid

• Letting the agent write its own SQL from scratch:
  This is where subtle bugs creep in. Instead, point the agent to vetted views or stored queries (e.g., weekly_kpi_view) that your data team owns, and have the agent focus on interpretation and narrative, not raw query generation.

• Posting raw tables with no narrative:
  Execs don’t want to scan 50 rows; they want “MRR is up 4.2% WoW, driven mostly by EU expansion; churn spiked in SMB due to pricing changes.” In Gumloop, explicitly instruct the agent to produce: “1–2 sentence summary + top 3 drivers + risks + 3 recommended follow-ups.”

Real-World Example

Imagine this Slack thread late Friday:

“Can we get a Monday morning summary in #exec-updates with MRR, NRR, new logos, churned logos, active users by product, and support ticket volume from the last week? Would love call-outs on anything unusual.”

With Gumloop, you turn that into a reusable, scheduled automation:

1. Data source setup:

   • Connect your Snowflake (or BigQuery/Databricks) instance with read-only access to a curated metrics schema: finance_metrics, product_usage, support_analytics.
   • Your data team exposes stable views:
     • finance_metrics.weekly_summary
     • product_usage.weekly_product_metrics
     • support_analytics.weekly_ticket_summary

2. Data Analysis Agent configuration:

   • Instructions:
     • “Every week, pull last week and prior-week data from the above views. Compute WoW and MoM changes for MRR, NRR, churn rate, average revenue per user, and ticket volume. Highlight any metric that moved more than 5% WoW or 10% MoM. Provide a 1-paragraph summary plus 5 bullet insights (3 positive, 2 risks). Keep language concise and exec-ready.”
   • Authorize it to call the warehouse and access only those views.
   • Lock the agent to your approved models (e.g., GPT-4o, Claude, or internal via proxy) using Gumloop’s model restrictions.

3. Workflow & Slack message:

   • Build a Workflow in Gumloop:

     • Node 1: Trigger – “Every Monday at 7:00am PT.”
     • Node 2: Agent Call – Data Analysis Agent runs queries and generates the report content.
     • Node 3: Formatter – Ensure the output is structured like:
       • Title: “Weekly Business Review – Week of 2026-04-06”
       • Section 1: KPI snapshot (table)
       • Section 2: Highlights & risks (bullets)
       • Section 3: Support health (ticket SLA, CSAT/NPS)
       • Section 4: Links (Looker dashboards, Snowflake worksheets)
     • Node 4: Slack Post – Send the formatted message into #exec-updates, optionally pinging @exec-team.

   • Result every Monday at 7:01am:

     Weekly Business Review – Week of Apr 6

     Headline: MRR up 3.8% WoW (+11.2% MoM), NRR at 121%, churn rate stable. Support volume spiked on billing issues but SLAs held.

     KPI Snapshot

     • MRR: $3.2M (+3.8% WoW, +11.2% MoM)
     • NRR: 121% (+2 pts WoW)
     • New logos: 46 (+12% WoW)
     • Churned logos: 9 (flat WoW)
     • Total tickets: 1,084 (+18% WoW, driven by billing experiments)

     Key Insights

     • EU mid-market drove ~60% of net new MRR.
     • Self-serve upgrades increased after pricing page changes; no meaningful churn impact yet.
     • Billing tickets increased 25% WoW; most related to promo code edge cases.
     • Two large customers show declining usage in the last 3 weeks—worth outreach.
     • Support maintained 96% SLA compliance despite higher volume.

     Deep dive dashboards: [Finance KPIs], [Product Usage], [Support Health].

   No analyst touched a dashboard that morning. If execs reply with follow-up questions in Slack, you can also allow the same Data Analysis Agent to be tagged in-channel—“@Gumloop, break this down by segment?”—and have it run additional queries against Snowflake/BigQuery/Databricks on demand.

Pro Tip: Start with one narrow “Weekly Exec Metrics” Workflow, get sign-off on the exact numbers and narrative style, and only then replicate it for GTM, product, or support. It’s much easier to scale once everyone trusts one canonical automated report.

Summary

You don’t need an analyst manually pulling from Snowflake/BigQuery/Databricks every week to keep leadership informed. By combining Gumloop’s Data Analysis Agent, scheduled tasks, and Slack integration, you can define the metrics once, wire them to vetted queries, and have an exec-ready brief show up in Slack on a schedule. The automation handles the grunt work—querying, calculating deltas, writing the narrative—while your data team stays focused on new insights and better models, not recurring exports.

Next Step

Get Started