Diagnosing a customer issue used to mean opening three tools, switching between tabs, and spending 20 to 30 minutes piecing together a picture that a senior engineer might have assembled in minutes from memory. We decided to bottle that expertise and make it available to everyone on the team, instantly.
The problem with institutional knowledge
Every support team has its unwritten knowledge: the engineers who know which error patterns mean what, who remember that one customer’s unusual setup, and who can map a cryptic ID to an account, a user, or a root cause without even blinking.
That knowledge is invaluable. It’s also invisible. It doesn’t live in a doc. It doesn’t transfer automatically to new hires. And it certainly doesn’t scale.
When a customer hits a problem, the clock is ticking. The last thing an engineer should be doing is clicking between disconnected internal tools, searching for context that exists somewhere — just not here, not now.
“What normally takes 20–30 minutes across three separate internal tools took 30 seconds.”
Introducing AppliOracle
AppliOracle is an internal AI assistant built specifically for Applitools support engineers. It connects to our internal data sources and surfaces a complete picture from a single query: account identity, user details, error history, account health, and likely root cause, all at once.
There’s no switching between tools. No hunting for context. You ask a question, and the answer comes back with everything you need to understand and resolve the situation.
Two moments that showed us what it could do
EXAMPLE 1: SPEED & DEPTH
A complete incident picture from a 12-character ID
A support engineer knew that a VMware account was experiencing test failures, but the cause wasn’t clear. They entered a single account ID into AppliOracle.
Within seconds, it returned everything: the customer was a large enterprise technology company, the affected user was in Bengaluru, their browser and OS were identified, and the root cause was a request that had hung for 16 minutes, triggering a cascade of 5 to 6 downstream errors.
AppliOracle also confirmed that the rest of the account was healthy, immediately ruling out a broader outage. No follow-up queries needed. The entire investigation that would normally require three tools and half an hour was done in under 30 seconds.
EXAMPLE 2: BUILT-IN GUARDRAILS
The tool caught its own mistake before it reached the customer
A medical device company operating under FDA regulations encountered a security error. AppliOracle correctly identified that a corporate firewall was intercepting encrypted traffic and began walking the engineer through a resolution path.
Then it stopped itself. Mid-response, it flagged that the customer’s configuration held a related certificate, but not the specific one required for the fix it was describing. It reversed its own instructions and redirected the engineer to the correct path.
No incorrect guidance reached the customer. AppliOracle is designed to question its own assumptions continuously; it does more than just answer questions, as it actively verifies them before committing to a direction.
The results so far
Both examples above came from a single engineer in the early weeks after rollout. The same quality of output is now available to the entire team, which leads to results like:
~98%
+47%
0
faster incident resolution (from 20-30 minutes to under 30 seconds)
more tickets resolved per week since adoption in April
days ramp-up needed (new hires benefit from day one)
Weekly tickets resolved jumped from around 75 to over 110 in the weeks following AppliOracle’s introduction, with a peak of 124 in late April. Customer response quality has improved alongside speed, because engineers are working from complete information rather than partial context.
Why it works
AppliOracle isn’t just a search tool. It reasons across multiple data sources simultaneously, surfaces the most relevant information without requiring the engineer to know exactly what to look for, and, critically, it is honest about uncertainty. It checks its own logic before presenting a conclusion.
This combination of speed and self-correction is what makes it genuinely useful in high-stakes support contexts, especially for regulated industries where a wrong answer has real consequences.
It’s also fully scalable. Every new hire gets the same depth of institutional knowledge on their first day that a senior engineer has built up over years. The tool doesn’t replace experience — it democratizes it.
And it doesn’t exist in isolation. AppliOracle is a direct expression of our broader CS Blueprint, the belief that AI should be embedded into how we work, not bolted on as an afterthought. Faster, more accurate support creates the foundation for everything else: stronger Joint Success Plans built on real account data, Executive Business Reviews backed by complete incident history, and a customer experience that drives both retention and expansion. When our engineers walk into every customer conversation with full context and zero guesswork, we’re not just resolving tickets faster—we’re showing up as the trusted partner our customers expect us to be.
Get in touch with the Applitools Support team for fast and accurate help.
Quick Answers
AppliOracle is an internal AI assistant designed for Applitools support engineers that aggregates account identity, user details, error history, and account health into a single, comprehensive view. It eliminates the need to switch between tools by providing complete, actionable context for every query.
No, AppliOracle does not replace experience; instead, it democratizes it. By providing every team member with the same depth of institutional knowledge that senior engineers have built over years, it enhances the capabilities of the entire team.
AppliOracle ensures accuracy through active verification. It reasons across multiple data sources, is designed to flag and question its own assumptions, and verifies its logic before presenting a conclusion, as demonstrated when it corrected its own instruction path during a security incident.
