The honest version. Where FoxChat wins, where the alternatives still win, and what changes for a team after a month of running it.
Customer chat has been around long enough that buyers can describe what they hate before they describe what they want. They hate help docs that nobody reads. They hate canned chatbots that route in circles. They hate live chat queues that take six hours to respond. They hate AI add-ons that bill per resolution and somehow make the bill go up the more useful the AI gets. None of these complaints are unfair, and the existing tools do not really refute them.
The category answer to those complaints has been to bolt new features onto old products. The vendor that started as a ticketing tool added AI. The vendor that started as a live chat added a knowledge base. The vendor that started as a knowledge base added bots. The result is a long feature list that is hard to compare and a procurement cycle that is harder still. The team buying chat ends up evaluating products that were never built for the same job, and the deciding factor often becomes which sales rep was nicest on the demo call. FoxChat exists because there is a cleaner version of this category, and the cleaner version is what we are trying to ship.
Static help docs solve the problem of writing the answer down. They do not solve the problem of finding the answer at the moment the visitor needs it. The math is brutal: most visitors do not search your docs. They scroll a product page, they hit a button that did not do what they expected, they get confused, and they leave. The doc you wrote three months ago that explains exactly what they should have done is invisible to them, because it lives in a separate help centre that they will never visit.
Even when a visitor does search the docs, the search is usually a literal keyword match against article titles, and the visitor types the question in their own words rather than your headings. The article exists. The search misses it. The visitor concludes the answer is not there, and the support inbox fills up with a question that was already written down. FoxChat reads the docs you wrote and answers from them in the words the visitor used, which is the half of the job static docs were never meant to do.
The first generation of chatbots was decision trees. A visitor typed a question, the bot matched it against a set of keywords, and routed to one of five pre-written answers. When the match failed, the bot offered a menu. When the menu did not contain what the visitor wanted, the bot offered to escalate to a human, who would arrive six hours later. Every visitor noticed within two messages that they were not talking to a person.
The cheapness was not the bot. The cheapness was the gap between what the visitor was trying to do and what the bot could do. A bot that can answer two questions in a hundred makes everyone feel worse. It signals to the visitor that the company has not really thought about them, and it signals to the operator that the chat surface is a containment mechanism rather than a help mechanism. The script-shaped bots most teams replaced when they bought FoxChat were producing that exact signal.
Every chat tool now claims an AI agent. They are not lying, and they are not all the same. The interesting question is what the AI is doing on top of, and what happens when it cannot answer. An AI bolted to a ticketing platform is going to optimise for resolving tickets, because the platform measures tickets. An AI bolted to a marketing chat is going to optimise for lead capture, because the platform measures leads. An AI inside a knowledge base will optimise for serving articles, even when the visitor wants something else. The AI itself is not the differentiator. The shape of the thing it is plugged into is the differentiator.
The other place AI-alone falls flat is the failure mode. A good AI on a bad foundation still has to fail sometimes, and the failure has to land somewhere a human can pick it up without losing context. The chat tools that ship AI as a feature, but ship a fragmented support tool underneath, end up dropping the conversation into a ticket queue when the AI gives up, and the ticket queue is the experience the visitor was trying to avoid in the first place.
The honest framing is that FoxChat is not a chat widget with an FAQ bot attached. It is a customer-success concierge that happens to live in a bubble. The difference shows up the moment a visitor needs to actually do something rather than just read an answer, and it rests on three things most chat tools do not have.
It runs guided how-to tours, not paragraphs. When the real answer to a question is a sequence of steps — set up an account, change a setting, finish a checkout — the FoxChat AI Operator (you name it yourself) does not paste a wall of instructions. You record a real task once, and your AI operator plays it back inside the page as a step-by-step walkthrough, giving the next action, waiting for the visitor to do it, and moving on when they are ready. A static help article tells someone what to do; a guided tour walks beside them while they do it. That is the difference between a question being answered and a task being finished.
It highlights the exact element on the page. Because your AI operator can see the live page the visitor is on, it does not say "look for the button in the top right" and hope. It lights up the specific button, field, or menu the person needs next, right where it sits on screen. When a visitor clicks the wrong thing or drifts onto the wrong page, your AI operator notices and points them back instead of plowing ahead. This is the part rigid, pre-recorded product tours get wrong: the instant a visitor steps off the script, the tour breaks. Your AI operator adapts to what the person actually did.
It remembers each visitor and where they got stuck last session. Your AI operator recognizes returning visitors. If someone was halfway through setting something up last week and comes back today, your AI operator can pick up where they left off — "Welcome back, last time we were setting up your account, want to finish that?" — instead of treating them like a stranger. Nobody has to re-explain themselves and nobody starts over. That continuity is the line between a tool that feels like a fresh blank form every visit and one that feels like it knows you.
Underneath all three, the product is one shape, not three bolted together. The widget, the brain, the inbox, the takeover, and the editor are one tool. The visitor asks, your AI operator answers from your help content with the source attached, and if confidence is too low your AI operator says so and offers to take a message or hand off to a person live. The operator sees the same retrieval results your AI operator saw, takes over in one click, and the conversation continues without restarting. When the operator answers a question your AI operator could not, that answer joins the knowledge base on the next save. End to end, the visitor's problem actually gets resolved — that is the whole point of calling it a concierge rather than a bot.
The flat-fee pricing is the second half of the difference. Charging per AI resolution sets up a hostile incentive: the operator hides the widget so it gets used less, the visitor gets a worse experience so the bill goes down, and the support inbox fills back up. FoxChat charges a flat monthly fee and answers as many questions as the visitors ask. The unit economics work because the support cost we are displacing is operator hours, and operator hours do not scale linearly with conversations the way per-resolution billing pretends they do. See the plans page for the current tiers.
FoxChat is not the right tool for every team, and we would rather you know up front. If you are running a support organisation of fifty agents on a ticket-first model, where the entire workflow revolves around assigned queues, SLA timers, and complex routing across multiple teams, a mature ticketing platform with deep workflow tooling will still outperform us. If you are an enterprise B2C buyer with a budget in the millions and an in-house AI team that wants to train its own agent on your own data, the bespoke vendors with custom training pipelines are a closer fit than our productised SaaS. If your support model is built around outbound conversational marketing — chat as the top of the funnel for an SDR team — a chat tool built around opportunity routing will fit your team better than ours. We do not pretend to compete in those three lanes. We are the right fit for sites and teams whose support volume is real but not gigantic, whose help content is good but not getting read, and whose operators want one inbox that handles everything.
The pattern we hear from teams a month in is similar across verticals. The first week is the surprise that the auto-imported knowledge base actually answers questions correctly without much editing. The second week is the surprise that the unanswered queue is small and that fifteen minutes of editor work each Friday makes your AI operator noticeably better the next week. The third week is the surprise that the support inbox is quieter, not because conversations went down but because the operator only sees the ones that needed a human. By week four, the team has stopped thinking of FoxChat as a chat widget and started thinking of it as the way visitors talk to the site. That shift — from "support tool we installed" to "how the site talks" — is the one outcome we look for in every account.
If this matches what you are trying to solve, the next move is the 14-day trial. No credit card, no demo call, no procurement cycle. Paste the script tag, watch the auto-import populate the knowledge base, and answer one or two questions in the inbox. If it works for your site, the conversion to a paid plan happens from the dashboard with one click. If it does not, no cleanup is required — the widget just stops at the end of the trial. The point of the trial being free and self-serve is that you can evaluate the product on your own site, in front of your own visitors, before deciding whether it deserves a place in your stack.
No credit card. One script tag. Every feature on every plan.
Start your trial