--- title: "PredictLeads Integration" description: "PredictLeads is company intelligence sourced directly from company websites, covering 120M+ companies. Agents discover companies, job openings, technology detections, news, financing events, SEC filings, products and connections, and follow companies for ongoing tracking." url: https://flowrunner.ai/integrations/predict-leads-app date_modified: 2026-09-04T19:45:02-07:00 --- # PredictLeads [Sales Intelligence](https://flowrunner.ai/integrations/category/crm-sales) PredictLeads is company intelligence sourced directly from company websites, covering 120M+ companies. Agents discover companies, job openings, technology detections, news, financing events, SEC filings, products and connections, and follow companies for ongoing tracking. [Verified](https://flowrunner.ai/integrations/verified "What does verified mean?") · 32 actions · Custom auth · available [PredictLeads website](https://predictleads.com/) · [Platform Documentation](https://docs.predictleads.com/) · Capability data verified 2026-08-27 1. A scheduled sweep runs across the accounts the team is watching 2. Get Company Technology Detections returns each detection with when it was first and last seen 3. The agent separates newly first seen adoptions from detections whose last seen date has lapsed 4. Get Company Subpages pulls the cleaned page text behind the detection to see whether the page still exists at all 5. Confirmed new adopters are written into the CRM as a dated signal with its evidence attached 6. The lapsed detections post to the account owner with both dates and the supporting page 7. The owner decides whether a lapsed detection is a real removal before any message claims it is ## What This Integration Enables PredictLeads takes one position and builds everything on it: the company's own website is the primary source. Job openings come from the careers page rather than from a job board. Technology detections come from what the site actually loads. Products, subpages, and repositories come from what the company publishes about itself. That provenance is the product, and it is why this connector reads differently from a general enrichment platform. You are not buying a profile. You are buying a dated observation of a public thing. The dates are the part that matters most in a flow. Nearly every dataset carries a first seen and a last seen timestamp, and nearly every operation can filter on both. Filtering detections by first seen after a date finds new adopters. Filtering by last seen before a date finds companies that appear to have dropped something. That single pair of fields turns a static database into a change feed, and a change feed is what a workflow can actually act on. It also introduces the honest limitation this page is built around. A last seen date records when the observation stopped, not why. That distinction is small on a dashboard and large in an email, which is where the [human-in-the-loop](https://flowrunner.ai/concepts/human-in-the-loop) gate below lives. ### Without FlowRunner **Signals bought as a snapshot**: A vendor file says what is true today and nothing about when it became true **Hiring read from job boards**: Postings arrive through aggregators, at aggregator freshness, with the source obscured **Research pasted from browser tabs**: Someone reads the careers page and the blog by hand and summarizes them into a note ### With FlowRunner **Signals carry their own dates**: Every detection reports when it was first and last seen, which is what makes it a change **Hiring read from the company**: Openings come from the company's own site, which is where a role appears first **Page text arrives cleaned**: Subpages come back as markdown a model can read without anyone scraping anything ## Use Case Scenarios ### A target list built from someone else's install base The team wants companies running a specific tool. The agent calls Discover Companies Using Technology, then narrows the result with company size and location filters, and passes each match through the company read to get name, description, and structured location. Because a misspelled technology name is accepted and silently matches nothing rather than being refused, the flow resolves the technology through the catalog first rather than passing a string it guessed. The finished list writes into [HubSpot](https://flowrunner.ai/integrations/hubspot) with the detection's first seen date on each record, so a rep can see how long the company has been running it before deciding how to open. ### Hiring as a demand signal, read from the source On a schedule, the agent calls Discover Job Openings across the whole dataset, filtered by occupation code, seniority, and location rather than by company. A cluster of new operations roles at a company already in the CRM raises a routing event: the agent enriches the account with its recent news and financing events, drafts a short brief, and posts it to the account owner in [Slack](https://flowrunner.ai/integrations/slack). Where the account is not yet in the CRM, the brief goes to the list [Apollo](https://flowrunner.ai/integrations/apollo) works from instead. The openings came off the company's own careers page, which is where a role usually appears before it appears anywhere else. ### Portfolio monitoring without querying every holding An investor operations flow calls Discover Portfolio Company Connections, which returns connections involving portfolio companies across the whole dataset rather than requiring a query per company. Financing events, categorized news, and any relevant filings are pulled alongside, and the digest lands in [Google Sheets](https://flowrunner.ai/integrations/google-sheets) and as a weekly summary in the team channel. Because the account has a monthly request allowance and exhausting it makes every subsequent call fail rather than slowing down, the flow reads the subscription status before the sweep and stops early if the remaining allowance will not cover it. ## Human-in-Loop Highlight The most commercially attractive query in this dataset is also the least certain: find companies whose detection of a competitor's product has stopped appearing. That is the rip and replace list, and every revenue team wants it. But a last seen date records when PredictLeads last observed the technology on the company's site, and observation can stop for reasons that have nothing to do with a purchasing decision. The page moved. The site was rebuilt. The tag was loaded differently. The detection lapsed and the contract did not. So the agent is allowed to build the list and is not allowed to name what it means. It gathers the evidence, both dates from the detection, the extended detection detail, and the cleaned text of the subpages that carried it, and it posts the case to the account owner: "Four accounts show the competitor's detection last seen more than ninety days ago. On three of them the careers and product pages have also changed structure in the same window, which would explain the lapse without any change of vendor. Treat these as removals, or as unverified?" The owner rules on each. What the agent must never do is write "churned off competitor" into a CRM field, because the moment it is a field it becomes the first sentence of a call, and a rep opening on a false premise costs more than the account was worth. Note also that Follow Company and Unfollow Company govern which companies produce webhook traffic for the whole account, so unfollowing to tidy a list can silently end another team's feed. Agent processes routinely Detects exception requiring judgment Clear match Continues automatically Ambiguous Routes to human via preferred channel Human decides Agent resumes with decision ## Agent Capabilities 32 actions ### Follow 3 - **List Followed Companies** Lists the companies this account follows. Following is what enables PredictLeads webhooks for a company, and it is also how a custom identifier of your own is attached to a company record. - **Follow Company** Starts following a company, which is the prerequisite for receiving any PredictLeads webhook about it. This is the one operation the vendor marks as free of charge. - **Unfollow Company** Stops following a company, which stops any webhook traffic about it. The follow list is account wide, so this ends deliveries for every flow that depends on that company. ### Companies 5 - **Get Company** Reads one company by identifier or by domain, returning name, ticker, description, language, and structured location. Passing a domain means no identifier lookup is needed first. - **Discover Companies** Searches the company graph by location, size, industry, and revenue rather than by a known identifier. Results are ordered by when they were last updated. - **Get Extended Company** Reads the extended profile of one company, which carries the additional attributes PredictLeads holds beyond the basic company record. - **Discover Extended Companies** Searches extended company profiles by location, size, industry, and revenue, ordered by when they were last updated. - **Get Similar Companies** Lists companies PredictLeads considers similar to a given one. For the leading results it also returns the reason each is considered similar, which is what makes the output usable as an explanation rather than only as a list. ### Job Openings 3 - **Get Company Job Openings** Lists a company's job openings, sourced from its own website rather than from job boards. - **Get Job Opening** Reads one job opening by its identifier. - **Discover Job Openings** Searches job openings across every company rather than within one, filtering by occupation code, title, seniority, and location. Hiring signals are what this dataset is usually used for. ### Technologies 5 - **Get Company Technology Detections** Lists the technologies detected on a company's website, each with when it was first and last seen. Those two dates are what turn this into a change signal rather than a snapshot, and they are also why an apparent removal needs confirming rather than announcing. - **Discover Companies Using Technology** Lists the companies detected using a given technology. This is the reverse lookup and the usual way to build a target list from a competitor's or a complement's install base. - **Get Extended Technology Detection** Reads one extended technology detection, which carries the additional detail PredictLeads holds beyond the basic detection record. - **List Technologies** Lists the technologies PredictLeads tracks, optionally narrowed by a fuzzy name search. Resolve a technology here rather than passing a guessed string, because a misspelled value is accepted and silently matches nothing. - **Get Technology** Reads one technology by identifier or by fuzzy name. ### News Events 3 - **Get Company News Events** Lists categorized news events about a company, extracted from a broad set of blogs, press, and news sites. - **Get News Event** Reads one news event by its identifier. - **Discover News Events** Searches news events across every company, filtered by category and by the size and location of the company involved. This is how a flow watches for a kind of event rather than for a particular company. ### Financing 2 - **Get Company Financing Events** Lists a company's financing events, most often funding rounds, extracted from the news dataset and normalized into a round type and an amount. - **Discover Financing Events** Searches financing events across every company by round type and company location, which is the usual way to watch for companies that have just raised. ### SEC Filings 2 - **Get Company SEC Filings** Lists a company's filing records. Only public companies have them. - **Get SEC Filing** Reads one filing by its identifier. ### Connections 2 - **Get Company Connections** Lists categorized relationships between a company and others, with a summary of how the two work together where PredictLeads has one. Partnerships, integrations, and investor relationships all appear here. - **Discover Portfolio Company Connections** Lists connections involving portfolio companies across the whole dataset, which is how an investor monitors its holdings without querying each one in turn. ### Content 2 - **Get Company Subpages** Lists a company website's subpages with their category, such as blog, careers, about, or terms, and the cleaned page text in markdown. The cleaned text is what makes this usable as an input to a model without anyone scraping the site. - **Get Company GitHub Repositories** Lists a company's public code repositories. ### Products 4 - **Get Company Products** Lists the products PredictLeads associates with a company. - **Get Product** Reads one product by its identifier. - **Discover Products** Searches products across every company by category and subcategory. - **Discover Startup Platform Posts** Searches posts on popular startup platforms, which is how a launch is spotted close to when it happens. ### Account 1 - **Get API Subscription** Reports the account's plan, which datasets it may reach, and how much of its monthly request allowance is left. Use it as the connection test and as the check before a large fan out, because exhausting the allowance makes every later call fail outright rather than slow down. ## Frequently Asked Questions ### What can FlowRunner do with PredictLeads? FlowRunner agents can run List Followed Companies, Follow Company, and Unfollow Company in PredictLeads, plus 29 more actions. ### Does connecting PredictLeads to FlowRunner require OAuth? PredictLeads uses a custom authentication method to connect to FlowRunner. ### Can PredictLeads trigger a FlowRunner workflow automatically? PredictLeads doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger. **Work at PredictLeads?** This integration exposes PredictLeads to AI agents on every FlowRunner plan, including through MCP, at no cost to you. [See what FlowRunner offers integration partners](https://flowrunner.ai/integrations/partners), including how to keep this page current. --- Markdown version of https://flowrunner.ai/integrations/predict-leads-app. Site index: https://flowrunner.ai/llms.txt