Verdigris left Salesforce for Lightfield to build an AI-native GTM

Verdigris builds electrical ground truth for AI data centers. Its sensors read power signals at high frequency so operators can catch equipment degradation, recover stranded capacity, and see inside critical infrastructure. The work includes implementation, deployment, quoting, invoicing, fulfillment, and ongoing customer operations.

Why Verdigris’s co-founder decided to switch from Salesforce

Verdigris had a decade invested in Salesforce. This included nearly 10,000 contacts, and a custom Implementation object with 211 fields. It required constant manual upkeep.

Two years ago, Thomas Chung started asking a pointed question: now that AI could do real work, was the Verdigris team taking full advantage of what this new capability afforded?

The first option was to extend Salesforce with its command-line tools and Agentforce. Thomas found that these tools could make one person more productive, but they did little to change how the broader team worked.

The second option was to build a shared interface on top of Salesforce. Thomas imagined an agent in Slack with enough access to run the system for the company. That approach came closer, but it would have required Verdigris to build core capabilities itself. It also raised questions about attribution and still depended on the same Salesforce data underneath.

Both approaches left the team maintaining infrastructure instead of changing how it worked.

"It didn't represent a path to unblocking the team to move quickly, which was what motivated the move,” Thomas said.

So Verdigris moved to Lightfield — a CRM the whole team can administer and query in plain language, not just the technical few. The migration carried everything: a decade of contacts and opportunities, the 211-field implementation object, and the workflows built around them, rebuilt in Lightfield.

Skills put Lightfield to work

Verdigris's processes now run as Skills and automations in Lightfield — natural-language instructions that teach the agent how the company works, pointed at work that had never been systematized.

The center of it is a sales playbook that lives in Lightfield Knowledge and informs every query and workflow the CRM runs. It includes stage-by-stage guidance, based on MEDDPICC sales methodology, on what AEs are supposed to do and warning signs to watch out for, all informed by an analysis of 2,330 of Verdigris's own closed deals.

A daily hygiene check scores every open deal on MEDDPICC and opens tasks for gaps that emerge. A weekly pipeline report rolls the same scoring up for leadership. The moment a deal closes, automations spin up the implementation and order records and route them to the right people.

"You don't go into Salesforce's admin page and describe what you need in human terms," Thomas said. "But you can in Lightfield. You can say, I need the following ten things to happen when we close a deal, enforce that on a state change, make sure the right people do it, and notify me in Slack if it doesn't. And it will just do it."

Customer health is the next piece under construction. The team is building a skill that ranks every account by the state of the relationship, including how many relationships exist, how often the team touches them, and how Verdigris's own service delivery is going. It replaces something the team tried and abandoned in Salesforce because it was too difficult. "We can look at the number of relationships, the touch points, the frequency, our own service delivery, and have it all visible in one place, coming out with a stack-rank list," Thomas said. "That's the job of several people being condensed into an automation.”

The whole company works from the same customer data

With Lightfield, all customer interactions are queryable by every person at the company. This has brought every team closer to customers - not just sales - and increased how responsive the whole company can be to customer feedback.

For example, before Lightfield, customer requests used to reach engineering via subjective feedback from the sales team. Different AEs shared different perspectives on what customers needed, leading to mixed signals on where to prioritize product development. Now, sales and engineering work together from a single source of truth, querying customers’ actual feedback on calls to inform their roadmap.

"The conversation from a customer to a salesperson to feedback to the engineering workflow really gets collapsed when you have a system like Lightfield," Thomas said. "More than just an individual going faster, the sales team going faster, we're also seeing the product organization move faster as well."

By operating on a shared, self-updating record, Verdigris now has the tools to move as fast as they need to capture the enormous market opportunity in front of them. Reps sell with the founders' judgment and 2,330 deals of evidence behind every stage. Processes that used to require a specialist are described in plain language and enforced by the system. And every function works from the same customer truth, so what one team learns, the whole company keeps.

As Thomas shared, "it becomes this tool that really shortens the learning loop for a company."