AptEdge

A Company Built Around a Boring, Expensive Problem

Here is a fact about enterprise software that almost nobody puts on a slide: the answer to your support ticket usually already exists. Somebody solved it. Maybe last March, maybe in a Slack thread, maybe in a Confluence page that three people have read and none of them can find. The knowledge is real. It is just scattered across so many tools that a support agent can spend twenty minutes hunting for something the company already knows. AptEdge, a Redwood City company founded in 2020, decided that this - not the model, not the magic - was the actual problem worth solving.

The pitch is almost deflating in its practicality. AptEdge does not promise to replace your support team with robots. It builds an AI answer engine, AnswerGPT, that plugs into the places knowledge already lives - Salesforce, Zendesk, ServiceNow, Jira, Confluence, SharePoint, Slack - reads all of it, and hands the agent a concise answer with the receipts attached. The company's own slogan for this is a support-desk pun, "Case Closed," which tells you roughly how seriously it takes itself and roughly how seriously it takes the work.

There is a reason to take the founders seriously even if you find the category crowded. AptEdge was started by Aakrit Prasad and Anthony Kilman, who were part of the team behind AppDynamics - the application-performance company Cisco bought for roughly $4 billion. So these are people who have already built enterprise software that large organizations paid enormous sums to own. And then, having done the glamorous thing, they went and picked the least glamorous corner of the enterprise stack: the technical support desk. That is either a strange choice or a very deliberate one, and the tell is that Cisco - the company that bought their last company - is now an AptEdge customer.

The Hard Part Isn't the AI

If you have watched any AI support demo, you know the failure mode. The bot answers confidently, fluently, and wrongly. In consumer chat that is annoying. In enterprise technical support - where the "answer" might be a configuration change that takes down a production system - a confident wrong answer is not a feature, it is a liability. This is the constraint that shapes everything AptEdge builds, and it is why the company keeps returning to a slightly unsexy acronym: RAG, retrieval-augmented generation.

The idea behind RAG is unromantic and, for support, exactly right. Instead of asking a language model to conjure an answer from its training, you first retrieve the relevant material from the company's own knowledge - the real docs, the resolved tickets, the internal runbooks - and then ask the model to compose an answer grounded in that material, with sources. AptEdge has written at length about this approach as a way to eliminate hallucinations in customer support. The subtext is honest: in this business, being right matters more than being clever, and the companies that forget that will ship something that demos beautifully and dies in production.

The other unglamorous truth AptEdge leans into is that knowledge bases are always out of date. The wiki is stale the day it is published, because keeping it current is somebody's least favorite job. So the platform's Knowledge Creation feature works the other direction - it drafts and updates articles from the tickets your agents actually resolve. Knowledge that maintains itself is a modest promise, but modest promises that come true are how enterprise software gets bought.

What You Can Actually Do With It

Strip away the category language and AptEdge is a set of tools a support organization bolts onto the systems it already runs. An agent working a ticket gets Answer AI - a drafted, grounded response in the console, so they are editing rather than starting from a blank box. A support leader gets Ticket Insights, which clusters cases to show which problems keep recurring and which fixes actually worked. And customers, ideally, never open a ticket at all, because In-Product Self-Service connects what they are doing in the product to the knowledge that resolves it.

The company frames the payoff in the language enterprise buyers care about: fewer escalations, faster resolutions, less agent churn, better CSAT. It also makes a point that is easy to underrate - it says teams can launch in two to four weeks using pre-built connectors, not the six-month integration slog that kills so many enterprise pilots. Time-to-value is itself a feature, and AptEdge treats it like one.

None of this is a story about eliminating people. It is a story about deleting the twenty minutes of searching that sits between a ticket and its answer. That is a smaller ambition than "AI will run your support desk," and it is a considerably more believable one.

The Money and the Market

In June 2023, AptEdge closed an oversubscribed $11 million seed+ round led by Stage 2 Capital, with Unusual Ventures, National Grid Partners, Carya Venture Partners and Counterpart Ventures joining. Eleven million dollars sounds modest until you count what bad support costs a company - churn, escalations, burned-out agents - and the investors clearly did that math. "AnswerGPT will set the bar for how support teams find answers," said Stage 2's Jay Po, which is the kind of thing an investor says, but it is directionally the bet.

The market AptEdge is competing in is not empty - Forethought, Ada, Cresta, Aisera and the native AI features inside Zendesk and Salesforce are all reaching for versions of the same customer. AptEdge's answer is focus: it is built specifically for the technical support desk of B2B software companies, the place where answers are complicated, wrong answers are costly, and grounding in real knowledge is the whole game. The company says its customer base grew more than 300% since 2022 and counts Cisco, Everbridge and Hexagon among its enterprise users. That is not a coronation. It is a foothold, in a fight that is far from settled - and that, on the evidence, is exactly the kind of problem these founders seem to enjoy.