Provar

Here is a fact about Salesforce that is both mundane and, if you run software on top of it, quietly terrifying: it updates itself roughly three times a year, whether you are ready or not. Layouts move. APIs shift. Automations you wrote eighteen months ago and forgot about behave differently on a Tuesday. Salesforce calls these "seasonal releases," which makes them sound pleasant, like a farmers' market. In practice they are a standing appointment with risk, and somebody has to make sure the appointment does not end in an outage.

Provar is a company whose entire reason to exist is that appointment. Founded in London in 2014, it builds test automation software purpose-built for the Salesforce ecosystem - not general-purpose testing tools that happen to work on Salesforce, but tools that know what a Salesforce page layout is, what a Lead Conversion does, and what an Approval Process should look like when it is working. This is a narrower thing to build than "testing software," and narrowness turns out to be the point.

The founders - Geraint Waters, Flavia Waters and Paul Noffke - did not arrive at this from nowhere. They had spent years doing end-to-end testing for highly regulated banking projects, the kind where a failed test is not an inconvenience but a compliance event. Waters started his career as an IT consultant at PwC in 1999, then passed through UBS and Barclays, institutions that take the phrase "did you test that" extremely seriously. When they looked at the fast-growing Salesforce world, they saw the same hard problem they had already solved for banks, sitting largely unsolved. So they solved it again, for a different customer.

The hardest part of test automation is not writing the first test. It is maintaining the thousandth one, a year later, after the platform underneath it has moved twice.

The pitch, stripped of adjectives, is two numbers. Customers say they build tests roughly 90% faster and maintain them roughly 80% less often. That second number is the one that matters, because the dirty secret of test automation is that most of the cost is not creation but upkeep. A test suite is a liability that pretends to be an asset until the day it saves you. Provar's design - low-code so manual testers can build, pro-code so developers can extend, and "metadata-aware" so the tests understand the structure of your Salesforce org - is aimed squarely at keeping that liability small.

"Metadata-aware" is worth pausing on, because it is the technically interesting part. A brittle test clicks on a button because the button is at coordinate X. A metadata-aware test clicks the button because it knows what the button is. When Salesforce moves the button in a seasonal release, the brittle test fails silently and the metadata-aware one shrugs. Multiply that across a few thousand tests and you have either a team that spends every release week firefighting, or a team that does not. Provar sells the difference.

The product line has grown up around that core. Provar Automation is the flagship - the thing that runs tests across UI, APIs and data layers, plugs into CI/CD tools like Jenkins and Git, and now generates test steps with AI at, the company says, 98% fewer clicks. Provar Manager is the more corporate sibling: a native Salesforce app for planning, tracking and reporting on quality, so that "did we test that" has an answer a manager can actually look up. Together they are pitched as a "quality hub" - one shared view of whether the software works.

Everyone is racing to ship AI agents on Salesforce. Almost nobody is talking about how you test them. Provar got there first.

Which brings us to the current chapter, and the one most likely to age well or badly quickly: Agentforce. Salesforce's big bet is on autonomous AI agents doing real work inside the CRM, and the entire industry is sprinting to build them. Provar's contribution is the unglamorous question nobody in the demo asks: how do you test an AI agent? An agent is non-deterministic by design; it does not click button X every time. Provar positioned itself as first to market with automated, end-to-end Agentforce testing, and by its Summer '26 release had extended that to Salesforce Government Cloud - which is either a small integration footnote or, if you believe agents are the future of enterprise software, an early claim on a large piece of ground.

The money reflects a company that grew steadily rather than explosively. Provar raised a $17 million Series A led by Kennet Partners in 2021, with additional later-stage funding reported since, for roughly $21 million total. That is not a unicorn's balance sheet, and Provar does not pretend it is one. It is the profile of a focused company that owns a category - Salesforce test automation - while broader tools spread themselves thin across every platform. In 2024 founder Geraint Waters handed the CEO title to Tony Sumpster and took the role of Chief Evangelist, the kind of move that reads as either a founder stepping back or a founder finding the job he actually likes. Probably both.

The most honest thing you can say about Provar is that it makes a product almost nobody wants to think about. No one demos their test suite at the all-hands. Testing is a cost until it is a rescue. But Salesforce is not going to stop changing, AI agents are not going to get more predictable, and the number of enterprises betting their operations on both keeps rising. Provar is a bet that the invisible, unloved work of making sure the thing still works will only get more valuable. So far, that has been a reasonable bet.