Customer Support's Best Upgrade Isn't a Smarter Chatbot

01 /The Concept

Hark helps companies "understand their customers like never before." The platform plugs into customer service and technical support workflows to make it easier for customers to communicate problems – and faster for support teams to resolve them.

Hark's current customers are mostly manufacturers of physical consumer goods: e-scooters, ergonomic chairs, razors, socks, and jewelry. That's not an accident – for buyers of physical products, the communication format Hark offers turns out to be far more natural than calling a support line or writing a ticket.

The idea: customers record short video clips – the same format as social stories – to show and explain whatever issue they're experiencing. They use the Hark app to do it, and the recording lands directly in the seller's support queue.

Recording a video is obviously easier than writing out a problem with a physical product. You show it, you explain it. Support gets the full picture in two minutes without playing email ping-pong. No follow-up questions like "what were you doing when this happened?" – the context is right there on screen.

This works for positive feedback too. A video review or product testimonial is far easier to give than filling out a survey. As Hark puts it, customer communication should feel like a natural conversation – not a formal business process.

Of course, having customers record videos is only half the story. Watching every incoming video, transcribing the problem, routing it to the right team member – that would create more work, not less.

That's where Hark's AI comes in. It processes incoming clips automatically, extracting the relevant details: which product, what the issue is, and even the emotional register of the message.

In many cases, the AI also suggests a resolution, drawing on the seller's knowledge base and past support interactions. The support agent often just needs to confirm the recommendation and send it – editing as needed – with one click.

All incoming requests are consolidated on a single dashboard, ranked by frequency and severity. Clicking on any issue opens a dedicated page: a full list of related requests, video clips, a timeline showing volume over time, and links to adjacent issues.

Each issue page functions as a rich brief for internal teams. Support agents can send these pages to product, engineering, or ops to communicate not just what the problem is, but how widespread it is, how urgent, and exactly how customers describe it.

The results are measurable: Hark's customers see an average 39% reduction in resolution time and a 43% drop in cost per support ticket. First-contact resolution – problems solved in a single reply – increases by 181%.

Pricing starts at $500/month, scaling with volume. The issue-insights module is currently in closed beta.

Hark's latest raise of $3.5M brings total funding to $5M.

02 /Why It Matters

The customer support software market is growing at roughly 21% annually. In 2024 it's worth around $15B; by 2031 it's projected to approach $70B.

AI is flooding into the category. Regardless of how you slice the numbers, it's clear that AI's role in customer support is expanding fast – and it will likely embed more deeply than most current projections suggest.

Hark has chosen a large, growing market with obvious AI tailwinds. The competitive field is crowded: chatbots, virtual agents, ticket analysis platforms, sentiment tools. Most of them layer additional capabilities onto traditional support processes.

What's different about Hark is that it changes the process itself. Rather than upgrading what happens after a request is received, Hark redesigns the input: customers describe their problems in a format that's already native to their behavior, and one that carries far richer information than text.

Knit, [covered here](/review/budut-li-oni-jeto-pokupat) in 2022, built a video feedback platform targeting Gen Z and raised $5.6M in their first round. They've since added stronger AI analysis tools for surfacing insights from the recordings.

Orson, [covered here](/review/horoshie-istorii-prinosjat-horoshie-dengi) last June, took yet another angle – an "AI director" that conducts video interviews with customers, selects the strongest moments, and edits them into compelling short testimonials. They raised $8M.

03 /Opportunities

AI in customer support is inevitable – the market is large, it's growing, and the opportunity is real.

The general direction is clear: build AI-powered platforms for customer support operations.

What's appealing about Hark's approach specifically is that they started from the customer's end of the funnel. They made it easier for customers to reach out first – which unlocks better data and faster resolutions downstream. Every other capability flows from that.

The reason this sequencing matters: when the bar to give feedback is high, sellers only hear about the most catastrophic problems – or they hear nothing until customers have already moved on. Lower the bar and volume increases. Handling that increased volume without adding headcount requires genuine automation – which is why this kind of platform has to solve the intake problem and the processing problem together.

The most direct opportunity: build a Hark equivalent for physical-goods manufacturers.

But the concept is portable. Consider a support platform for cloud software, where the classic ticket reads "I did something and now everything is gone." You can't reconstruct what the user actually did from a one-line description, and the user probably can't either.

A better approach: run a background screen recorder that loops and overwrites old footage. When the user hits a problem, one tap captures the preceding 30–60 seconds of activity and surfaces it as a clip they can annotate with voice and blur for sensitive info before sending. Support gets the exact sequence of actions that triggered the bug. Add AI to find patterns across hundreds of such clips, and you're building something genuinely useful.

The general principle: improving customer support starts at the top of the funnel – making it easy and natural for customers to communicate in a way that's useful to both sides.