Capacity vs. Maven AGI, Two Ways to Kill the Support Ticket

Every company on earth runs a version of the same broken machine. A customer has a question. They fill out a form, or wait in a chat queue, or press 4 to be told the menu has recently changed. A ticket is born. Someone, eventually, closes it. Multiply that by a few billion and you have the customer support industry - a cost center that nobody loves and everybody funds. Two companies have decided to take it apart. They just picked opposite tools.

On one side is Capacity, run out of St. Louis by David Karandish. On the other is Maven AGI, a three-year-old Boston company backed by Dell and Cisco. Both sell what the industry now calls CX automation - AI that answers customer questions and takes action across chat, voice, email and SMS. Both talk about a future where the support ticket, as a unit of work, mostly disappears. And both have real customers paying real money to get there. What makes the matchup worth watching is that they represent two genuinely different theories of how you win a market like this.

The Roll-UpCapacity bought the neighborhood

Karandish is not a first-timer. In 2006 he and his business partner Chris Sims started the parent company of Answers.com. In 2014 they sold it to a private equity firm for north of $900 million. That is the kind of exit that lets you choose your next move carefully, and the move he chose is instructive. Founded in 2017, Capacity did not try to out-engineer everyone from a standing start. It went shopping.

Over roughly three years, Capacity acquired and merged 14 companies into a single support-automation platform. Knowledge management here, a chatbot there, workflow tooling, voice, quality assurance - each bolt-on filling a gap in the stack. The result is a platform the company says now serves more than 20,000 customers and 1.5 million users, and has facilitated over 36 billion automated interactions. The client list runs to names like 3M, Disney, Nike and LinkedIn.

The logic of a roll-up is distribution first, elegance later. If you already have the customers and the channels, you can layer new AI on top of a relationship that exists rather than fighting to create one. The risk is the opposite of the reward: fourteen codebases, fourteen data models and fourteen ways of doing things do not merge themselves into something clean. Roll-ups win on reach and lose on coherence, and the whole bet is that you can bolt a modern AI brain onto a stack you assembled by hand.

Distribution is the hard part. Karandish already knows what it feels like to own a market - so he bought his way back into one.

The RebuildMaven AGI started from zero, on purpose

Maven AGI made the other call. Founded in 2023 by Jonathan Corbin, Sami Shalabi and Eugene Mann, the company builds a single enterprise agent from scratch. Corbin's résumé is the tell: he ran customer success and strategy globally at HubSpot, with earlier stops at Adobe, Marketo and Sprinklr. He spent two decades inside the exact problem the company now automates. Shalabi built personalization systems that powered billions of interactions at Google; Mann came from applied machine learning at Stripe and Google. This is a team that lived the pain and then went to build the fix cleanly, without inheriting anyone's legacy plumbing.

The pitch is depth over breadth. Maven's platform claims to answer up to 93% of customer queries autonomously, resolve issues around 10x faster than traditional methods, and cut response times by as much as 60% once an enterprise tunes and deploys it. The security page reads like an alphabet of certifications - SOC 2, ISO 27001, HIPAA, PCI DSS, GDPR - which is the price of admission for the large enterprises Maven is chasing. Customers named on the site include TripAdvisor, ClickUp, Clio and Hard Rock Digital.

The money followed the story. In June 2025 Maven raised a $50 million Series B led by Dell Technologies Capital, with Cisco Investments, SE Ventures, Lux Capital, M13 and E14 involved, bringing total funding to $78 million. Some of those investors are also customers and channel partners - Dell and Cisco don't write checks that size for a demo. The distribution followed too: in July 2025 Maven landed in the AWS Marketplace AI Agents category, and by February 2026 it had joined the IBM watsonx Orchestrate Agent Catalog and made The Agentic List 2026 in the CX Agents category.

Head to HeadThe same target, two blueprints

Line the two up and the contrast is almost too neat. One is a serial founder who already won; the other is a first-time CEO who lived the problem. One grew by absorbing companies; the other grew by refusing to. One leads on reach; the other leads on a number - 93% - that it repeats like a mantra. Here is the split, plainly.

Neither model is obviously right, which is the interesting part. In older software categories, distribution usually beat technology - the company that reached the buyer won, even with a worse product. But AI is unusually unforgiving of legacy. An agent that reasons over clean, well-structured enterprise data behaves very differently from one bolted onto fourteen stitched-together systems. If the technology gap is large enough, reach stops being a moat and starts being ballast. If it isn't, Capacity's 20,000 relationships are a head start Maven has to earn one logo at a time.

The StakesWhat actually gets decided here

Strip away the branding and both companies are making the same claim about the future of work: that the majority of routine customer questions should never touch a human. That is not a small claim. Support has been a labor line item for forty years, sized by headcount and measured in average handle time. If an agent can genuinely close most of that volume on its own, support stops being a cost center you staff and becomes a software category you buy - which is exactly why investors and platform giants are circling.

The honest caveat is that the biggest numbers here come from the companies themselves. "Up to 93% autonomous" is a vendor figure, not an audited benchmark, and "20,000 customers" spans everything from a small business chatbot to a Fortune 500 deployment. The direction is real; the precision is marketing. Anyone evaluating either platform should ask for their own numbers on their own data, because the gap between a good demo and a good deployment is where most CX automation projects live or die.

The support ticket has two assassins. One is trying to out-distribute the problem. The other is trying to out-engineer it. The bet the whole market is now watching

For a buyer, the practical read is less about who wins and more about what you need. If you want broad coverage across many channels and a vendor that has already consolidated a dozen point solutions, Capacity's breadth is the argument. If you want a clean, enterprise-grade agent built for high-autonomy resolution and you have the data hygiene to feed it, Maven's depth is the argument. The uncomfortable truth is that the eventual winner probably needs both - deep enough to resolve, wide enough to reach - and right now neither has fully closed the gap on the other's strength.

That is what makes this a live story rather than a settled one. Two teams looked at the same tired help desk and drew opposite blueprints. Over the next couple of years, the market will quietly grade both. Not with a headline, but with renewals.