Chatref

What it is

Chatref is a chatbot builder for customer support. The company's FAQ puts it plainly: "Chatref is an AI support agent for your website. It learns from your own content and answers customer questions day and night, with your team able to step in any time." It is written for small support teams and website owners who have help docs or a site full of answers and want an agent that uses them without writing code or drawing conversation flows. The site does not say where the company is based or how many customers it has.

What you build and how

You build one AI agent, and the work is mostly choosing what it knows. The site's feature list reads like a build order: AI Agents ("Your own AI assistant, trained on your business"), Knowledge Base ("Teach your agent everything your business knows"), Website Widget, Shared Inbox, Omnichannel, Custom Actions, Lead Capture, Conversation Tags, Insights and Analytics, Multilingual, Customization and Pay as you go.

The knowledge base comes first. The FAQ says you add "your help docs, website pages, and files, and it answers from them." Another passage adds that knowledge can be extended with website URLs, files or text, and that the system retrains automatically when new content is added. Answers can link back to the page they came from, and you can test the chat privately before it goes live. A comparison post on the company's blog describes the setup as content-first: "You add your knowledge sources" such as a website URL, help center or PDFs, and the agent trains on them in minutes.

Because the agent answers from your content, the company says it will not invent replies: "Chatref answers from the content you give it, so replies stay true to your business instead of guessing." You can set the tone and voice of responses to match brand guidelines, and the same post says you can set the agent's tone and personality without writing rules. The interface can also show suggested questions that customers click instead of typing.

Going live takes a snippet. The FAQ says you set up the agent "from a simple dashboard and add it to your site with one snippet," and that "Most teams are live the same day." The widget can be customized to match your brand, with no code.

The same agent reaches customers on more than the website. The company says it answers on web, Slack and email, and the FAQ names WhatsApp as well. You do not build a separate bot per channel.

The Shared Inbox is the collaboration piece. The feature list describes it as "Watch chats live and step in when a human is needed," and the trust section says that when a chat needs a person, "your team steps in on the same thread. No clunky handoff, no lost context." Conversation Tags label chats by topic automatically so you can filter and report, and Insights and Analytics show what people ask and how the agent is doing.

Custom Actions let the chat "collect info, link out, or do tasks," and Lead Capture turns chats into contacts. The company is candid that this is not its focus. The Landbot comparison says Chatref "can capture leads during a chat, but it's not a dedicated lead-gen tool."

Languages: the site says it helps customers in 11 languages automatically, and the FAQ lists Spanish, French, Arabic, Hindi and Chinese among them, replying in the visitor's own language.

What the pages state

Pricing is by usage. The site says "Pay as you go. Start free. You only pay for the answers your chatbot gives. No per-seat fees, and no yearly contract to sign." The pricing summary lists "$50 in free credit to start," top-ups "anytime with simple prepaid credits," your whole team added "at no extra seat cost," and the ability to "Scale up or pause whenever you want." The site gives no per-answer credit price or any plan tiers. The blog's comparison with another vendor says each answered question costs "a small amount of credits," so the exact cost per conversation is something to confirm on the full pricing page.

On data, the FAQ says "Each workspace's data is kept private and isolated, and you stay in control of what your agent can see." The site names no compliance standard, so a buyer who needs GDPR, SOC 2 or similar paperwork has to ask.

How it contrasts with flow builders is spelled out in the Landbot post: "Landbot runs on flows you design. Chatref runs on knowledge you already have." It adds that with Chatref, "You don't need to build conversation paths," while "Landbot gives you more creative control over the exact dialogue." A second post about YourGPT argues that per-seat charges and overage fees are where support-tool bills quietly grow, and offers prepaid credits as the alternative. Treat both posts as the company's own framing of the field.

2 more details come from those posts. The home page describes what happens on each question: Chatref "checks the content you taught it, answers in your own words within seconds, and lets a real person on your team step in at any time." And the YourGPT post says you can "start free and test with real chats before you buy credits," which suits a team that wants to see real answers on its own site before spending anything. The same post says the AI's answers come from your own content, so they stay factual, and that anyone on the team can take over a conversation instantly at no extra cost.

Who it suits and who it does not

Chatref suits a small team with existing documentation and a support load that rises and falls. Paying by answered question, with the whole team in the inbox at no seat charge, fits a business that has quiet months. The FAQ says it can handle "any question covered by your knowledge base," so adding pricing, comparisons and demo content lets it answer sales questions too.

It is a weaker fit for someone whose main goal is a designed conversation, such as qualifying prospects through several branching steps. The company says so itself and points those teams at Landbot or a separate lead tool. It also needs real content to learn from: an agent grounded in your pages is only as good as those pages. And because the site publishes no plan tiers or credit prices, a finance owner will want to test the free credit before committing.