Capacity vs Microsoft in the Fight to Kill the Support Ticket

The support ticket is one of the most durable objects in business software. For twenty years it barely changed: a customer or an employee hits a wall, fills a form, waits in a queue, and eventually a human replies. In 2026 that object is finally being pried apart - not by a single product, but by two companies approaching it from opposite ends of the market. Capacity, a St. Louis platform that just reported crossing $100 million in recurring revenue, wants to answer the question before the ticket exists. Microsoft, which needs no introduction, decided the same thing and shipped it inside a license most companies already pay for.

This is the shape of a lot of AI fights right now, and support is the clearest example. A focused specialist that has spent years on one problem, versus a platform giant that can turn your product into a checkbox. Both are chasing the identical outcome. How they get there, and what a buyer should actually weigh, is where it gets interesting.

01 / THE PRIZEWhat "killing the ticket" really means

Support has a metric that has quietly become a strategy: ticket deflection. A deflected ticket is one that never gets created, because a self-service answer or an AI agent resolved the question first. Every deflected ticket is a cost that does not happen and a wait that nobody sits through. For years deflection was a nice-to-have dashboard number. AI turned it into the whole pitch.

The reason is simple. Most support volume is repetitive - password resets, policy questions, "where is my order," "how do I configure this." A system that reads a company's knowledge and answers accurately can take a large slice of that off human plates. The two companies here agree completely on the destination. They disagree on how you build the road.

02 / THE SPECIALISTCapacity's bet on depth

Capacity was founded in 2017 by David Karandish and Chris Sims. The pair were not first-timers - they had built the parent company of Answers and sold it in 2014 for roughly $960 million. When they came back, they picked a single problem and stayed on it. The result is what Capacity calls an AI Knowledge Orchestration Layer: one place that connects a company's knowledge, data and systems, then uses that layer to power AI agents, live agent assist, automated quality checks, conversational intelligence, and outbound campaigns across chat, email and voice.

The strategic claim behind all of that is focus. Capacity does support automation and only support automation. It is not trying to also be your email client, your spreadsheet, or your operating system. When a company whose entire business is one thing goes deep, it can reach corners a broad suite will not bother with - the messy integrations, the odd channels, the quality-assurance layer that reviews the agent's own answers.

Capacity also grew the way a focused company often does - by absorbing adjacent tools and folding them into the same knowledge layer, rather than shipping a pile of disconnected bots. The $100M ARR figure it reported matters less as a trophy than as evidence: enough companies decided support automation was worth a dedicated line item, separate from whatever their platform vendor throws in.

03 / THE GIANTMicrosoft's bet on gravity

Microsoft did not build a support-automation startup. It did something harder to compete with: it made support an agent inside software you already run. In June 2026 the company brought its Service Agent to general availability in Microsoft 365 Copilot, unifying customer-service workflows across Dynamics 365 and Microsoft 365. In one conversation, a service rep gets case data, knowledge sources and the tools to actually take action. Copilot Studio, Microsoft's agent-building platform, lets IT teams design and deploy their own custom agents on the same rails.

The mechanics of the licensing tell you the whole strategy. To reach case data and service workflows, a customer needs a Dynamics 365 Customer Service license. To unlock the fully integrated experience - case context sitting next to Microsoft 365 data and AI actions in one place - you add a Microsoft 365 Copilot license. In other words, the support agent lives exactly where your tickets, your documents and your identity already are. That is not a feature advantage. It is a distribution advantage, and distribution is the thing startups cannot buy.

04 / THE REAL QUESTIONNot "is it good," but "is it worth a second invoice"

Here is the uncomfortable part for every focused AI company, Capacity included. When the platform ships your feature, the buyer's question changes. It stops being "is this product good?" and becomes "is it enough better than the thing I already pay for?" Microsoft's Service Agent does not have to be the best support AI on the market. It has to be good enough for a buyer who already owns it to stop looking.

That is why focus is not a marketing word here - it is the business model. Capacity's argument is that support, done seriously, is a cross-channel operation with real depth: voice campaigns, quality assurance on the AI's own output, integrations that reach outside the Microsoft world, and a team whose entire attention is on this one problem every day. For a company where support is core - a technology or SaaS business, a contact center, anyone whose customers judge them by response quality - that depth can clear the bar of a separate line item. For a company where support is a cost to contain and most data already sits in Microsoft 365, the platform agent may simply be enough.

05 / WHAT TO STEALHow to actually make the call

If you are the person who has to choose, skip the demo theater and ask three questions. First: where does your knowledge live? If almost everything is already in Microsoft 365 and Dynamics, the platform agent starts with a real head start on context and identity. If your knowledge is scattered across tools and channels Microsoft does not touch, a vendor-neutral layer earns its keep. Second: how central is support to how customers judge you? The more it is a product surface rather than a cost center, the more depth pays off. Third: who maintains the knowledge? Every one of these systems is only as good as the content it reads. The best AI agent wired to stale documentation still gives confident, wrong answers.

Most large organizations will not pick one and be done. They will run Microsoft's agent where the ecosystem makes it free and obvious, and reach for a specialist where support is deep enough to justify it. The interesting outcome is not a knockout. It is coexistence - the giant owning the default, the specialist owning the demanding edge.

What makes this worth watching beyond support is that it is a preview. AI is going to run this same play across category after category: a focused company that got there first, and a platform that can absorb the feature into software everyone already runs. Support automation just reached the fork earlier than most. Whether depth can hold its ground against distribution is the question, and Capacity versus Microsoft is the clearest live test we have.

FAQCommon questions