PaperChat

What PaperChat is

PaperChat is a no-code builder for customer support AI agents, written mainly for online stores. Its home page says it "learns your products, policies and pages, then answers on your website, Telegram, Instagram and Messenger", looks up live product and order details, captures the lead, and hands the conversation to you when a person is needed. The reader it serves is a store owner or a small support team that wants a bot connected to the shop, not a search box with a personality. The page's own pitch is "Train an agent on your store, put it on your site and let it answer tonight." The footer says the product is made in the EU.

What you build and how

The build has three steps on the home page, and the promise is "No developer, no onboarding call, no rules to write out by hand."

1. Point it at your store. You paste your website address and PaperChat reads your pages. You can add files, help docs or plain text for anything that is not on the site, and the page lists website, PDFs, docs, sitemaps and raw text as sources. Training "takes a few minutes, not a project plan".
2. Connect what it should be able to do. Switch on your store and the agent can search the live catalog, read order status and quote your real policies. The page says there is nothing to code and "no API keys for Shopify".
3. Put it where customers already are. One line of script puts the widget on your site. Connect Telegram, Instagram or Messenger and the same agent answers there too, with every conversation in one inbox.

The most important idea is that the agent acts instead of only answering. The page sets a basic chatbot against PaperChat in five situations. When a customer types "Where is my order?", the basic bot sends people to a tracking page, while PaperChat looks the order up and reads back the status. On a size question, a basic bot repeats an old FAQ line, while PaperChat searches the live catalog with price and stock. On a return after 30 days, a basic bot guesses or invents a policy, while PaperChat answers from your own policy pages. When a customer wants a person, it hands the conversation to your team in the same thread, instead of looping. And at 2am on Instagram, the same agent replies where a website-only bot would leave the visitor waiting. During an action the customer sees "Performing action" and then a real answer, such as a refund status pulled from Stripe in the page's sample chat.

The integrations come in two groups. Shopify, WooCommerce and WordPress actions are included on every plan: Shopify lets the agent search the catalog, read policies and build a cart; WooCommerce lets it look up products, stock and order status; WordPress lets it answer from posts and pages as they change. On a paid plan you can connect Stripe, Zapier, HubSpot, Pipedrive, ClickUp, Attio, Resend, Paddle and Apify, and any MCP server. MCP is described on the page as "the open standard that lets an AI agent use another product properly, through its own API". The logo strip names Printify, Google Maps, Sanity, Storyblok, Firecrawl, Tavily and CoinGecko among them, and the page says "More than 30 servers" across commerce, payments, CRM, content, maps, travel and live web search, and says Zapier MCP reaches thousands more apps, with Slack, HubSpot, Mailchimp and Google Calendar named. You pick a server, paste a key, and you can also add any MCP server by URL.

Around the conversation, the agent asks for a name, email or phone number mid-chat and files it in your leads list with the full transcript. It answers in "Over 58 languages", matched to whatever the customer writes in, with no extra setup. Every conversation is stored and searchable, which the company pitches as a way to see the five questions costing you the most time. Telegram connects with a bot token, Instagram through direct messages and Messenger through your Facebook page.

Beyond the home page, the site carries integration pages for Shopify, Webflow, Wix, Framer and WordPress with WooCommerce, and solution pages for chatbots on those platforms plus Squarespace, and for customer support, lead generation, an AI sales agent, Telegram, Instagram and Messenger. It also lists use-case pages for real estate, SaaS and agencies, plus a set of free tools such as an AI reply generator and a text-to-speech tool. The home page even embeds a demo bot that was "built with Paperchat and trained on this website".

Plans, limits and claims

Every plan starts with a 7-day free trial, and the pricing is built around credits: "One credit per answer, no seat fees, no setup fee, no contract to sign."

• Basic, $35 a month, marked most popular. 4,000 message credits (about 1000+ conversations), 1 AI chatbot, a 25 MB knowledge base and 3 team members.
• Pro, $99 a month. 10,000 message credits (about 2500+ conversations), 3 AI chatbots, a 40 MB knowledge base and 4 team members.
• Enterprise, $399 a month. 40,000 message credits (about 10000+ conversations), 5 AI chatbots, a 60 MB knowledge base and 5 team members.

The page says you can ask for more credits, chatbots or a bigger knowledge base. Every plan, the page repeats, includes the Shopify, WooCommerce and WordPress actions.

The home page reports "270 chatbots built on Paperchat, answering 1.2 million messages and counting". That is the company's own count.

Who it suits and who it does not

PaperChat suits a Shopify, WooCommerce or WordPress store whose customers ask about orders, sizes and refunds more than they ask about how a feature works. The included store actions are the draw, and a 7-day trial lets an owner test the agent against real questions. A small team that talks to customers on Instagram, Messenger and Telegram will like having one agent and one inbox across them. The pricing counts credits, so a team can estimate cost from expected message volume rather than headcount, since the plans carry team members and no seat fees.

It is a weaker fit for a business with no store or product data to connect, because its strongest features are the live lookups. The Basic plan's 1 chatbot and 25 MB knowledge base is a tight start for a large help center, and even Enterprise stops at 5 AI chatbots and 5 team members, so an agency running many clients would need the extra chatbots the page says can be added on request. The site presents it as a customer support tool, so it is not an internal workflow or app builder.