How Paradigm sped up a key strategic project for Crossbeam by 586%
By Kyle Tianshi
Crossbeam, an ecosystem revenue company that works with large datasets of company information to provide insights for businesses, had a key strategic project that was bottlenecked by manual data collection and unreliable AI integrations. With Paradigm, they’ve been able to fully automate workflows and increase the speed and accuracy of that project by 586%.
Challenges
One of Crossbeam’s flagship products, Partnerbase, is the world’s largest database of partnerships between B2B companies. Initially, Crossbeam relied on paid researchers and interns to manually gather company profiles and maintain a list of their relationships. While feasible at a small scale, this approach quickly became tedious and expensive. Identifying new company relationships was much more difficult once major, high-visibility companies were documented, and keeping existing relationships up-to-date required an increasing amount of manpower. Crossbeam eventually transitioned into crowdsourcing data for Partnerbase, which caused growth of the database to flatline.
At the same time, the team wanted to modernize services like their Ecosystem Grader, a tool that analyzes a company’s ecosystem size and provides insights into its growth opportunities on demand. For both Partnerbase and the Ecosystem Grader, they needed a way to rapidly source information from the internet at high volume without compromising accuracy.
Enter Paradigm
When Matt Nicosia, head of AI operations at Crossbeam, set out to reduce this heavy manual workload with an AI integration, he first experimented with more traditional data broker tools and even tried building an in-house AI research tool. These solutions proved impossible to scale and produced poor quality data. Then, he discovered Paradigm.
“We compared Paradigm directly to a couple of those [other solutions] and Paradigm was just more accurate,” he said.
Matt found Paradigm’s spreadsheet interface familiar and accessible. Once he got started, it “was not hard for me to just use it [Paradigm] and get it up and running the way I wanted it to,” he said. With just a company name or domain, Paradigm could quickly retrieve detailed company information, including industry, size, location, investors, and partners. Matt easily adapted Paradigm for Crossbeam’s specific needs. For each column, he could customize the prompt, specify where data was sourced from, control the types of values returned and the formatting, and use other columns in the sheet as context.
For example, to gather data on company partnerships, Matt wrote the following column prompt:
Prompt
List all the partners/integrations of this company.
These can be in the integration/partner page you already found (if available), or in public announcements/articles. Do not use anything from partnerbase.com as a source here. Do not include partnerbase.com as a partner.
This should be the company, not a product (ie, google.com not play.google.com).
If no partners are found, return nothing, not even an explanation.
Formatting
Provide a comma separated list of all partners and integrations for this company. Provide the partner/integration company parsed domain only. Include parentheses and single quotes around each domain. Include a comma even after the final entry. Keep blank if nothing found.
Example: ('crossbeam.com'),('partnerbase.com'),('linkedin.com'),
Paradigm returned clean, consistently formatted lists of each company’s partners in seconds, complete with the agent’s thought process and sourcing.
Partnerships generated by Paradigm with Matt’s column prompt.
The AI team at Crossbeam quickly built out a number of workflows with Paradigm.
“We at Crossbeam use it [Paradigm] to do research on demand... If we want to understand whether a company is a customer of another company, or if we’re trying to understand whether there are certain partnerships that exist between other companies… we use it for that because it’s really highly flexible.”
To speed up their Partnerbase workflow, they set up a Paradigm sheet that takes in lists of companies requiring research and scrapes the web for detailed data on each company. One of the enriched columns is a list of partnerships for each company. In a separate sheet, each partnership is categorized as either a Technology Partner or a Channel Partner, enriched for additional details, then sent into Partnerbase. The impact was immediate: Paradigm began delivering thousands of precise data points at a scale unattainable by manual work.
Partnerbase workflow with Paradigm.
Separately, Crossbeam also automated their Ecosystem Grader with Paradigm. When someone enters a company for analysis into their system, the request enters Paradigm via an input webhook, Paradigm does research on estimating company size, industry, revenue, growth rate, customer count, closed won rate, and a partner ecosystem score on a scale of 1 to 100. The outputs are returned directly back to the customer through an output webhook.
Ecosystem Grader workflow with Paradigm.
Results
Partnerbase’s monthly growth rate surged by 586% after integrating with Paradigm. In just the past seven months, they added 60,000 new companies and over 100,000 new partnerships to the database, increasing its size by 67% at about half the cost of manual efforts. Their Ecosystem Intelligence platform now operates fully autonomously, delivering high-quality, data-driven insights whenever a customer requests them.
Ecosystem Grader workflow with Paradigm.
“It would not have been possible to have a team of humans doing this at scale,” Matt said.
Paradigm was the catalyst for Crossbeam to achieve results that would have been unattainable for a human team. Now, they’re continuing to use Paradigm to power more research, automate growth workflows, and save thousands more hours of valuable time.