# Predictleads

> Canonical: https://www.yalc.ai/mcps/predictleads/

Real time signals on 100M+ companies, surfaced inside Claude Code as native tool calls.

## Categories

- intelligence
- prospecting

**Website:** https://predictleads.com/?ref=yalc

**Repo:** https://docs.predictleads.com/mcp

## Glance

- Free tier: 100 reqs/mo
- Coverage: 100M companies
- Auth: API key + MCP URL

**Install Command:** claude mcp add predictleads --transport http "https://mcp.predictleads.com/sse" --headers "X-API-Key:YOUR_PREDICTLEADS_KEY"

**Install Notes:** Sign up at predictleads.com (100 free API requests per month). Generate an API key from your dashboard. Predictleads documents an MCP integration; the exact server URL may differ from the placeholder above. Check the MCP integration section of the Predictleads docs for the current URL and headers.

**Yalc Fit Score:** 9

**Yalc Verdict:** The signal layer for event triggered Yalc workflows. When intent (hiring, funding, news, technographics) drives outbound timing, this MCP fires the trigger.

## Plain Description

Predictleads provides structured signal data on 100 million plus companies: news events across 29 categories, job openings, technographics (1.2 billion technology detections all time), key customers, financing rounds, and similar company lookups. The vendor positions explicitly for AI agents and exposes the same dataset via REST, MCP, webhooks, and flat files.

For Yalc workflows, Predictleads is the canonical signal source. Where Crustdata gives you the database (who exists), Predictleads gives you the trigger (what just changed). When a Yalc agent says "alert me when ICP accounts post a hiring signal" or "trigger outbound when a target company raises Series B", Predictleads supplies the event.

## Yalc Framework

**Workflow Step:** listen

### Workflow Narrative

The Predictleads MCP sits at the **listen** node in Yalc's GTM topology. Most static prospecting (firmographics, ICP filtering) happens upstream in Crustdata. Predictleads kicks in when the question shifts from "who matches" to "who just changed".

The MCP exposes the signal endpoints as native Claude tools. Yalc workflows can subscribe to Predictleads webhooks for real time push, or query the MCP on demand for batch intelligence digests.

**Workflow Position:** The signal listener. Yalc subscribes to Predictleads watchers on target accounts, classifies the event via Claude, decides whether to trigger an outbound action. Downstream is the standard Yalc send pipeline.

### Prompt Patterns

- Yalc, set up a watcher on the "ICP" Notion list via Predictleads. When any company posts a "hiring head of marketing" or "raised Series B" event, alert me on Slack with the source URL.
→ Yalc creates a webhook subscription, routes events through Claude classification, posts via Slack MCP.

- Yalc, every Monday pull Predictleads news events for our 200 target accounts from the last 7 days. Cluster by event type, summarize, write to Notion as a weekly intel digest.
→ Yalc batches MCP calls, summarizes via Claude, writes structured digest to Notion.

- Yalc, for every job opening event Predictleads detects on a target account, draft a personalized outbound to the hiring manager and queue it in Instantly. Hold for my approval.
→ Yalc joins job event to enrichment to draft to send queue, respecting human approval gate.


### Chaining

**Upstream:** ICP list (Crustdata or Notion) → Predictleads MCP (signal subscription)

**Downstream:** Predictleads event → Claude (intent classification) → Notion / Slack / Instantly


### Anti Patterns

- Don't subscribe to all 29 event types globally. You'll drown in noise. Filter by event type AND target account list.
- Don't act on every signal. Some events (a routine press release) aren't outreach worthy. Use Claude to score signal strength before triggering anything.
- Don't poll the MCP on a tight loop when webhooks are available. You'll burn the 100 reqs/mo free tier in a day. Webhooks scale better.


**Compatibility:** Predictleads documents an MCP integration that works inside Claude Code, Cursor, and other MCP clients. The free tier (100 requests per month) is enough to pilot a real Yalc workflow. Webhooks are first class for production volume.

## Pros

- 100 API requests per month free. Easy to start.
- 29 event categories cover most B2B trigger patterns (hiring, funding, tech, news).
- 100M+ company coverage at the firmographic layer plus a signal layer most enterprise databases don't ship.
- Real time webhooks for signal events. No polling.
- Multiple delivery formats (REST, MCP, webhooks, flat files). Pick what fits your workflow.

## Cons

- Pricing above the free tier is custom. Procurement friction.
- Signal noise without filtering. Most events aren't outreach worthy.
- Some event categories (press releases) are higher volume / lower signal than others (financing).
- Less people level data than Crustdata. Predictleads is company centric.

## Who For

- GTM teams running event triggered outbound (signal based prospecting)
- Sales engineers who want intent data without enterprise sales suite pricing
- Data teams building internal company intelligence pipelines

## Related

### Tool

**Rule:** Pricing tiers, dataset breakdown, and use case context.


### Mcp

**Rule:** Crustdata for database queries, Predictleads for signal events. They compose well.



## Alternatives

- **Crustdata MCP**
  - Rule: Use both. Static database queries belong in Crustdata, live signals in Predictleads.
  - Url: /mcps/crustdata/
- **6sense / G2 Intent**
  - Rule: Switch when buyer intent (research signals on review sites) matters more than firmographic events.
  - Url: #

## Faq

- Q: How does the Predictleads MCP compare to Crustdata's MCP?
  - A: They compose. Crustdata is a database (who exists, where, what they do). Predictleads is a signal layer (what just changed). Most Yalc workflows install both and chain them.
- Q: What kinds of signals does the MCP surface?
  - A: 29 event categories. High signal include financing rounds, executive hires, expansion moves, technology adoption, hiring patterns, layoff news, product launches. Lower signal include routine press releases.
- Q: Can I use Predictleads via webhooks instead of MCP polling?
  - A: Yes. Webhooks are first class. Subscribe to event types and target accounts, Predictleads POSTs when events match. Yalc workflows route POSTs through Claude for classification before any action.
- Q: How current are the signals?
  - A: Most events surface within hours of detection. Some (job openings, news) are near real time. Some (financing rounds) lag because the source itself lags.
- Q: Does Predictleads have person level data?
  - A: It's company centric. For person level, pair with Crustdata. The strength is company events and technographics.
- Q: Is the free tier really enough to ship something?
  - A: Yes for piloting. 100 requests per month with webhook delivery scales further than 100 polling calls. For low frequency monitoring on a small target list, the free tier covers a real workflow.

**Reviewer:** Othmane Khadri