The trouble was the form. Wachter, a technology integrator, had six weeks to make its StaffSafe safety-monitoring service easier for restaurant franchisees to buy. Thunder helped build the Salesforce workflow. Then a customer saw the early version. It asked for too much information, too soon.
- The job: make Salesforce and connected systems work for the business.
- The customers: organizations with complicated sales, support, and data operations.
- The lesson: test the customer’s task before expanding the technology.
Thunder’s account says the team simplified intake and delivered on time, slightly under budget. A franchisee could become a qualified lead online, then use a portal after sales follow-up. The revealing detail is the correction: a working system still needed a less demanding front door.
The work after the purchase
That is a useful introduction to Thunder. The company occupies the space between purchasing enterprise software and getting an organization to use it well. Someone must connect records, settle workflow questions, move data, and decide which requests should reach a person. Those decisions determine what the expensive platform actually does on Monday morning.
Founder and CEO Carter Wigell knows the platform from the inside. He joined Salesforce in early 2001, among its first 100 employees. Thunder, founded in 2020, has a suitably direct mission: help customers love Salesforce, forever. The affection presumably gets easier when the quoting process works.
Its services span customer support, revenue operations, and data and AI. Buyers can bring it a contact center, an awkward quoting process, or fragmented information that makes automation unreliable. The customer roster crosses software, media, business services, and hospitality. A dedicated federal practice, launched in June 2025, extends the approach to government operations.
Thunder positions itself as a specialist alternative to large systems integrators, with reusable implementation assets and experienced practitioners. Its broader competition includes other Salesforce consultancies and the customer’s own internal team. Hiring Thunder makes sense when the missing ingredient is expertise or delivery capacity, rather than another software license.
Give the agent a small job
Consider Ansira. Following two acquisitions in 2024, the company needed to consolidate support systems while keeping everyday inquiries moving. Its general hotline offered a sensible starting point for automation: recurring questions, shared knowledge, and callers who needed help outside business hours. Thunder was already working on the Salesforce migration.
The resulting voice assistant, Sara, uses Amazon Connect and Nova Sonic, retrieves information through OpenSearch, and relies on Lambda for conversational logic. About 200 knowledge articles supply its answers. Crucially, callers can reach a live agent; unresolved after-hours conversations become Salesforce tickets. The design includes a route out of automation.
The AWS case study reports initial call deflection of 12 percent and a 54 percent reduction in hold time. Thunder’s August 2026 collaboration announcement reports 30 percent deflection. These are different published snapshots. They suggest useful progress, while leaving plenty of work for people.
Smartsheet offers another version of the same idea. After initially implementing a service agent itself, it brought in Thunder to extend agentic AI through support processes. The published case reports 47 percent case deflection and 54 minutes saved per agent each day. Configuration, document ingestion, and access to unstructured information sit behind those tidy numbers.
At Yahoo, the target was an advertising-sales chore: responding to requests for proposals. Thunder built an Agentforce solution that creates CRM records, summarizes documents, and suggests responses using relevant business information. Training included teaching trainers. A tool that produces a suggestion still needs people who understand how to use it.
A business built for the next request
Thunder earns its place through consulting, implementation, integration, and continued support. Amplify, its managed-services program, addresses the work that survives launch: configuration changes, reporting problems, data hygiene, and the backlog an internal administrator cannot finish. The commercial logic is straightforward. A customer’s Salesforce environment keeps changing as the customer’s business changes.
Amplify combines primary contacts working US hours with delivery expertise in Manila. Thunder describes flexible capacity and continuity of knowledge as selling points. For buyers, the useful comparison is the capacity and skills available over time, alongside platform licensing, AI consumption, and the effort their own staff must contribute.
“It’s a great capability that a customer can really enjoy using, because it gives them autonomy with zero pressure.”
Alan Clark, Wachter Product Experience Manager, on the StaffSafe workflow
Investors have funded that services business. A $16 million Series A in 2023 was led by Recognize, with Salesforce Ventures participating. Washington Harbour Partners added growth backing in December 2024, targeting data, AI, cloud capabilities, and Amplify. Those investments finance Thunder’s expansion; they are separate from what customers spend on implementations.
The direction has since become clearer. Thunder announced an agentic AI research lab with Salesforce and USC Marshall in October 2025, then an AWS strategic collaboration in August 2026. Both place it close to the platforms and institutions shaping enterprise AI. Customers still need the resulting work to survive contact with daily operations.
Start with the task people avoid
The examples suggest a practical sequence: pick a bounded task, inspect the information it needs, show the workflow to users, and measure what happens. Keep a human handoff. Budget for refinement. Thunder’s own federal guidance warns that agents need tuning after launch and that weak leadership backing can leave pilots stranded.
The conditions matter. An assistant working from unreliable knowledge can answer badly at greater speed. A portal that demands too much can discourage the customer it was meant to help. Wachter’s form makes the point neatly: progress began when somebody looked at a functioning system and asked whether the task felt easy.