Advisors

Advisors are automated AI workflows that continuously monitor your data and deliver actionable insights directly to your team via Slack. Think of them as always-on analysts that scan your accounts, opportunities, and custom data — then surface the things that matter most.

What Can Advisors Do?

• Spot expansion opportunities — Find accounts with high product usage that are ready to upsell
• Flag at-risk deals — Identify stale opportunities missing recent activity
• Monitor customer health — Track support ticket trends, NPS scores, and engagement metrics
• Custom analysis — Build any data-driven workflow tailored to your business

How Advisors Work

Advisors run on a schedule you define (daily, hourly, or custom). Each run follows this process:

1. Fetch data from your connected data source (Snowflake, Salesforce, PostgreSQL, etc.)
2. Shape the data by joining related tables and aggregating metrics
3. Apply rules to filter for the rows that matter
4. Generate insights using AI to analyze each matching record
5. Deliver results as Slack DMs to the right people on your team

Insights land in the recipient's Slack DMs automatically — no need to check a dashboard.

Creating an Advisor

Navigate to Dashboard → Advisors → New Advisor to launch the creation wizard. You'll need an Owner or Admin role to create advisors.

Step 1: Type

Choose a starting point:

• Start from scratch — Build a custom advisor from the ground up
• Use a template — Pick a pre-built template (e.g., expansion alerts, churn risk) and customize it

Step 2: Data Sources

Select which tables from your connected database you want the advisor to analyze. These are the tables you connected via the Connection Wizard — for example, your accounts table, opportunities, telemetry data, or support tickets.

Step 3: Data Shaping

This is where you define how your data comes together. Data Shaping lets you:

• Pick a primary entity — Choose the main table (e.g., accounts) and a key column (e.g., account ID). This determines the grain of your analysis: one row per account, one row per opportunity, etc.
• Join related tables — Bring in data from other tables. For example, join your accounts table with telemetry data or activity logs.
• Select and aggregate columns — Choose which columns to include. For one-to-many relationships (like multiple telemetry readings per account), pick an aggregation: most recent value, average, sum, count, and more.
• Preview your data — Click Run Preview to see exactly what the advisor will analyze, with live data from your database.
• Set a notification delivery column — Map a column containing email addresses (like OWNER_EMAIL) to control who receives each insight via Slack DM.

For a detailed walkthrough, see Data Shaping.

Step 4: Rules

Define conditions that filter your shaped data down to the records worth analyzing. Rules use a visual builder with support for:

• Multiple rule groups with AND/OR logic
• Nested conditions for complex filtering
• 15+ operators — equals, contains, greater than, is empty, and more

For example: "Show me accounts where latest_usage is greater than 80% AND current_tier equals Standard."

Only rows matching your rules get sent to the AI for analysis.

Step 5: Prompt

Write the instructions the AI will follow when evaluating each matching record. Your prompt template can reference the shaped data columns, and you can customize:

• The prompt itself — Tell the AI what to look for, what tone to use, and what to recommend
• Model settings — Choose the AI model and adjust parameters

A good prompt is specific. Instead of "analyze this account," try: "Given this account's usage trend and current tier, recommend whether they're a good candidate for an enterprise upgrade. Include specific data points in your reasoning."

Step 6: Review

Review your complete configuration:

• Primary entity and key column
• Number of joined tables and selected columns
• Rule conditions
• Prompt template
• Schedule (set when and how often the advisor runs)

Click Create Advisor to save. Your advisor will run on its next scheduled time, or you can click Run Now to test it immediately.

Running Advisors

Scheduled Runs

Set a cron schedule for your advisor (e.g., daily at 8 AM, every 6 hours, weekdays only). The system automatically evaluates the schedule and kicks off runs when they're due.

Manual Runs

Click Run Now on any advisor to trigger an immediate run. A progress modal shows real-time status updates as your data is fetched, analyzed, and processed.

What Happens During a Run

1. Data is fetched from your external database using your Data Shaping configuration
2. Rules filter the results to matching rows
3. Matching rows are submitted to the AI in a cost-efficient batch
4. The AI analyzes each row and generates insights
5. Insights are delivered to Slack (typically within a few minutes of batch completion)

Viewing Insights

Generated insights appear in two places:

• Slack DMs — Each insight is delivered as a direct message to the person specified by the notification delivery column (e.g., the account owner's email)
• Dashboard → Insights — All insights are visible in the web dashboard with filtering, search, and status tracking

Each insight includes:

• A title and summary of the finding
• A recommendation for what to do next
• A priority score
• The underlying data that triggered the insight

Insight Delivery via Slack

Advisors deliver insights to Slack using email matching:

1. The advisor reads email addresses from your notification delivery column (e.g., OWNER_EMAIL)
2. It looks up each email in your Slack workspace to find the matching user
3. The insight is sent as a Slack DM from the Pingd bot

Important: The email address in your data must match the email of an actual member in your Slack workspace. If there's no match, the insight will still be created in the dashboard but won't be delivered via Slack.

Tips for Effective Advisors

• Start simple — Begin with one table and a few rules, then expand
• Use preview — Always preview your data shaping before running the full advisor
• Be specific in prompts — The more context you give the AI, the better the insights
• Set appropriate schedules — Daily is great for most use cases; hourly for time-sensitive alerts
• Check your email mappings — Make sure the notification delivery column contains emails that match your Slack workspace members