Fill your CRM from social conversations
Buska's engine already watches 30+ platforms for your configured signals and scores every mention for buying intent. This skill puts an AI agent on top: it pulls those scored signals over MCP, removes duplicates, qualifies each one against your ICP, and hands you a clean shortlist. Accepted leads then travel to your CRM through the kanban, its Google Sheets sync, and webhooks.
get_signalsscore_lead1The plumbing you plug together
A Buska account with signals configured, the MCP server (buska.io/mcp) connected to your agent, and the kanban-to-Google-Sheets sync enabled in the app for the column you want to export. Webhooks are available if your CRM should be fed directly.
2The agent pulls scored signals
Instead of scraping feeds, the agent calls get_signals to fetch the qualified buying signals Buska's engine has already detected for your keywords. It filters by intent (ACTIVE for people looking to buy, PAIN for people describing the problem), a minimum AI score, and a since date so it only sees fresh signals. Each result comes with the post, the author, the score, and the reasoning behind it.
3It dedupes and double-checks
The agent drops anything you have already processed: same post URL or same author as a previous run. For borderline results it calls score_lead to re-check buying intent and ICP fit before letting a lead through. That keeps low-quality mentions out of your pipeline.
4It writes a CRM-ready summary
For every lead that passes, the agent produces a compact record: author, platform, link to the post, a one-line summary of what they said, the score, and why it qualifies. That is exactly what you need to decide in seconds, and exactly what a CRM row looks like.
5From shortlist to CRM row
A short, deduplicated list of qualified prospects with evidence, not a raw feed. Move the ones you approve into a Buska kanban column: the Google Sheets sync mirrors that column into a shared sheet, and webhooks and CRM integrations carry leads onward from there. You stay the approval step between social noise and your CRM.
The prompt to paste into your agent
Copy it into Claude, ChatGPT or any MCP client connected to Buska.
Connect to my Buska MCP server and refill my pipeline.
1. Call get_signals with intent=ACTIVE, minScore=7, since=2026-08-25, limit=50.
2. Call get_signals again with intent=PAIN and minScore=8.
3. Drop duplicates: any post URL or author already seen in my previous runs.
4. For signals scoring exactly 7, call score_lead to confirm buying intent and ICP fit; keep only confirmed ones.
5. Return a table with: author, platform, post URL, one-line summary, score, and why it qualifies.
6. Finish with the top 5 I should move into my Buska kanban so the Google Sheets sync picks them up.Related documentation
Frequently asked questions
Which CRMs can receive these leads?
Any CRM that reads from Google Sheets or accepts webhooks. In Buska, a kanban column syncs to a shared Google Sheet in real time, and webhooks fire for every new lead, so HubSpot, Pipedrive or a custom stack can pick them up. Native integrations cover 15+ tools.
Is this a real-time push into my CRM?
No, and that is by design. Your agent pulls signals over MCP when you run it, and leads reach your CRM through the kanban's Google Sheets sync and webhooks after you approve them. Nothing enters your CRM without passing through you.
Does the Sheets sync overwrite my CRM data?
No. The sync only writes new rows to the shared sheet when a lead enters the kanban column, and removes the row if you move the lead out. Nothing writes into your CRM directly unless you wire a webhook to it, and that stays under your control.