The support ticket is the last unautomated line item in cross-border e-commerce
Every cross-border operator I know has quietly built the same stack: a helpdesk, a chatbot layered on top of it, a knowledge base nobody updates, and a Slack channel where the real escalation happens. The chatbot deflects the easy WISMO tickets — where is my order, can I change my address, why is my tracking stuck — and then everything else falls through to a human who has to open three tabs to figure out what actually went wrong. That human is expensive, and in a market where Shopify brands are fighting 40%+ return rates and Amazon sellers are eating ODR dings for slow replies, that human is the bottleneck on your review score.
So when I see a support tool repositioning itself around investigation rather than answering, my ears prick up. Gleap — the German support-and-feedback platform that’s been quietly shipping since 2021 — just launched Kai Resolve, an agent that reads the screen recording, the console logs, the click trail, and your actual code before it replies. On paper that’s a developer-tool story. In practice, it’s a story about whether the 60% of cross-border tickets that aren’t simple WISMO can finally be handled without a human in the loop — and whether the economics of $0.49 per investigation change how you staff a support desk in Manila or Lisbon.
What Gleap is actually selling, and why it’s not another Intercom clone
The pitch from co-founder Isabella Salzmann is precise: Kai Support already handled the easy cases cheaply, but “some tickets need more than a good answer: someone has to dig into the data.” That’s the wedge. Kai Resolve watches the session replay, reads the console and network logs, checks the account data read-only in a sandbox, points to the exact file and line in your code, then either answers the customer, flags a likely bug for your team with cause and evidence, or files the feature request to your roadmap via MCP. Gleap claims it resolves 75% of all tickets automatically based on its own data — and when it isn’t sure, it says so rather than guessing.
The pricing is the tell. Roughly $0.49 per investigation on average, billed for work done, included from the Team plan, with a 7-day free trial and no credit card. That’s not seat-based SaaS pricing — that’s outcome pricing, and it’s the same shift we’ve watched sweep through Klarna’s AI assistant rollout and every AI SDR tool that followed. When you bill per resolved unit of work instead of per seat, the vendor’s incentive flips: they want fewer, better investigations, not more logins.
Why Amazon sellers should care more than Shopify ones
Here’s the counterintuitive part. Shopify DTC operators have always had the widget-and-replay luxury — Gorgias, Re:amaze, Zendesk all plug into your storefront and capture the browser session. Amazon sellers don’t get that. Your buyer messages arrive through Amazon Seller Central with no session context, no console logs, no click trail — just a text box and a 24-hour response clock. So Kai Resolve’s core value proposition (investigate the evidence attached to the report) is structurally unavailable to the pure-Amazon operator. The sellers who benefit are the hybrid ones: Shopify storefront plus Amazon storefront, where the Shopify side generates the rich telemetry and the Amazon side generates the volume. If you’re 100% FBA, this launch is interesting but not yet actionable.
How it stacks up against the incumbents you’re probably already paying
The obvious comparison is Intercom and its Fin agent, and Gleap’s own reviewers make it for me. One reviewer on the launch page says Kai “outperforms Intercom’s Fin”, and another long-time user who switched from Intercom calls the pricing “affordable and provides excellent value, especially compared to similar tools like Intercom.” That’s a directional signal, not a benchmark — but it lines up with what I hear from operators: Fin is good at deflection, weak at diagnosis, and priced per resolution in a way that gets uncomfortable at scale.
The more interesting comparison for cross-border sellers is Gorgias, which owns the Shopify helpdesk mindshare and has been bolting on AI macros and auto-responses. Gorgias wins on e-commerce-native integrations — order lookups, refunds, subscription management inside the ticket. Gleap wins on technical depth: console logs, code reading, session replay, the MCP plumbing into monitoring and roadmaps. For a store selling physical goods with occasional checkout bugs, Gorgias is the safer pick. For a brand running a custom storefront, a subscription app, or a headless build where “the discount code didn’t apply” is a real engineering question, Kai Resolve is doing something Gorgias doesn’t.
Then there’s the bug-reporting lineage. Gleap started life in 2021 as an instant-replay, screenshot bug-reporting widget, and one agency owner on the launch page describes installing it on every WordPress site his agency builds, with the widget visible only to admins so clients can file bug reports with automatic screenshots or video. That’s the DNA of Kai Resolve: it’s not a chatbot that learned to read logs, it’s a bug-reporting tool that learned to write replies. Whether that lineage is an advantage or a limitation depends entirely on how much of your ticket volume is actually technical — and for most physical-goods sellers, it isn’t.
Where the math breaks
Let me be the pessimist for a paragraph. $0.49 per investigation is an average, and Gleap’s own team confirms in the comments that “a really packed ticket can cost a bit more and a simple one a bit less.” For a seller doing 8,000 tickets a month, that’s roughly $3,900 in investigation fees on top of the Team plan — real money, but likely cheaper than two FTEs. The problem is the 25% that aren’t auto-resolved. Those tickets still consume the investigation fee and a human’s time, so your effective cost per resolved ticket is meaningfully higher than $0.49. Run the numbers against your current cost-per-ticket before you get excited.
The duplicate-bug problem is the one to watch
The sharpest question in the entire launch thread comes from a commenter asking whether Kai can correlate multiple reports of the same bug into one investigation. The answer is honest and revealing: “For bugs, each ticket still gets its own investigation right now. Grouping all reports of one bug into a single investigation… is the next logical step and something we’re looking at.” For cross-border sellers this matters enormously. When a payment gateway breaks at 2am and 400 customers in three time zones file tickets before your team wakes up, you want one investigation and one update broadcast — not 400 separate $0.49 investigations. Until that ships, the value is capped at the long-tail technical ticket, not the incident-response scenario where support costs actually spike.
What cross-border operators should borrow from this launch
Three things, regardless of whether you buy Gleap.
First, the evidence-attached ticket. The reason Kai Resolve can investigate is that Gleap’s widget captures the session replay, console logs, and click trail at the moment of report. Most cross-border brands capture none of this. If you’re on Shopify, install a session-replay tool like Microsoft Clarity (free) or Hotjar and start attaching replay links to your helpdesk tickets manually. Even without AI, your agents get 10x faster at diagnosis overnight.
Second, outcome-based vendor evaluation. When you’re renewing Zendesk or Gorgias this year, ask for pricing per resolved ticket, not per seat. The market is moving that direction — Gleap’s $0.49 model is the leading indicator — and the vendors who won’t move are telling you something about their margins.
Third, the closed feedback loop. The detail I keep coming back to is the one Lukas Böhler, Gleap’s co-founder, highlights: when a fix ships, “the customer hears back on the same ticket they opened — automatically.” How many of your product complaints get a follow-up when the fix goes live? Almost none, in my experience. That single behavior — closing the loop — is the cheapest review-score improvement available to any brand, and you don’t need an AI agent to do it. You need a rule in your helpdesk that fires when a linked issue closes.
What I’d watch / test next
This week, pull 50 recent tickets from your helpdesk and classify them: WISMO, returns/refunds, product question, technical/bug, other. If technical is under 10% of volume, Kai Resolve is a nice-to-have, not a priority — revisit in six months when the duplicate-correlation feature ships. If technical is 20%+, sign up for the 7-day trial and feed it your five nastiest tickets, the ones your team has been avoiding. Watch specifically for false confidence: does it flag uncertainty, or does it guess? That’s the failure mode that erodes customer trust fastest.
Second test: install Microsoft Clarity on your storefront this week and route replay links into your helpdesk tickets manually. Measure your agents’ time-to-resolution before and after. If manual replay attachment moves the needle, you’ve just built the business case for paying for automated investigation. If it doesn’t, you’ve saved yourself a subscription.
Third: audit your review-request and fix-notification flows. The cheapest win in this entire launch isn’t the AI — it’s the loop-closing behavior. Ship that first.
