We think support tickets should close themselves.

The problem Adam kept running into.

Before founding Resolvemark, Adam Ross spent four years in customer operations roles at B2B SaaS companies in San Francisco — first as a support ops manager, then building internal tooling for a 40-person CS team. The problem was structurally the same at every company: support queues where 60-70% of volume was the same 12 request types. Refund requests. Billing confusion. Subscription changes. Password issues. Identical answers, 40 times a day, burning out good support people on low-complexity work.

The tools available in 2022 made the problem worse, not better. Chatbots deflected by showing customers help articles — they measured "deflection rate" without asking whether the customer's actual problem got solved. That conflation of deflection with resolution is still endemic in the industry, and it was the specific thing Adam wanted to fix.

In late 2022, Adam started building an agent that could actually take action: look up an order, verify a policy, process a refund, close the ticket. Not a conversation simulator. The first version handled refunds and subscription changes on a single Zendesk instance. Confidence scoring was imprecise and escalation rules were hand-coded. But the core direction proved correct in the first week of testing — tickets closed themselves.

Three years on, Resolvemark handles 30+ action types, connects to 9 platforms, and sees 91% average first-contact resolution across its customer base. We're still five people. The goal is still the same: the ticket closes itself, or we hand off to a human with full context — never a dead end.