Chat Thing
Chat Thing is a no-code platform for building AI support agents that learn from your website, docs, FAQs and Notion pages, then answer customer questions around the clock on your site, Slack, Discord, WhatsApp, Telegram or via API.
What is Chat Thing?
Chat Thing is a no-code platform for building AI customer-support agents, published by Chat Thing Ltd from Newquay in Cornwall, United Kingdom. The premise is simple: you point it at the content you already have and it learns to answer like someone who works there. Sources include a crawled website or sitemap, a Notion workspace, uploaded files (.pdf, .csv, .json, .md, .html, .mht, .docx, .txt, .zip, .epub), a YouTube channel, an RSS feed, Google Drive documents or text you paste in by hand. Scheduled auto-sync re-reads those sources and detects what changed, so a bot does not quietly go stale.
Deployment is deliberately broad. One bot can serve an embeddable web widget, Slack, Discord, WhatsApp, Telegram, a dedicated email address and a general-purpose API channel. The web widget installs with a single line of code and works on WordPress, Shopify, Squarespace or a custom site; the vendor claims a five-minute setup. Beyond answering, power-ups give the bot the ability to act: call any API, trigger Zapier AI Actions, read, write and query Notion, look up Shopify orders, customers and products, run Slack and Discord actions, browse the web, search Google, route a question to a specialist bot, or escalate to a person.
Model choice is a selling point. Bots default to GPT-4o Mini, and from the Standard tier upward you can pick among models from Anthropic, OpenAI, Google, Mistral, Meta, Cohere, DeepSeek, Perplexity, xAI, Qwen, MoonshotAI and Z.ai, switching per bot. Requests are routed through OpenRouter with Zero Data Retention enforced at account level, so prompts are processed in memory and discarded; if no compliant endpoint is available the request fails rather than falling back to one that would store data. Voice features are the documented exception, calling OpenAI directly.
Capability is tiered. Human takeover, messaging channels and API access start at Standard; conversation analytics with sentiment and topic clustering at Pro; automated bot evaluations at Enterprise. Billing runs on message tokens for conversations and storage tokens for syncing content, with à-la-carte add-ons to raise a single limit. An MCP server, a chatbot SDK, HMAC-signed webhooks and eight free standalone tools round out the platform.
What it does
- Answer customer questions 24/7, grounded in the content you supplied rather than guesswork
- Ingest a crawled website, sitemap, help centre, Notion workspace, files, YouTube channel or RSS feed
- Take real actions through power-ups: look up an order, update a record, trigger a workflow
- Hand a conversation over to a human and let a teammate take it live, then hand it back
- Run on your website, Slack, Discord, WhatsApp, Telegram, a dedicated email address or the API
- Browse a live web page and run a Google search inside a conversation
- Report on what customers ask: sentiment, recurring topics and where your documentation falls short
When to use Chat Thing / When not to
A quick filter to help you decide if Chat Thing is the right fit.
When to use Chat Thing
- Support teams drowning in the same repetitive questions about hours, passwords, orders and delivery
- Small businesses and solo founders with no developer, who need a widget added in one line of code
- Product and documentation teams who want their help centre and API docs to become answerable
- Online retailers who need a bot that can look up an order or a product, not just recite a FAQ
- HR and internal-knowledge teams building staff-facing bots on top of Notion, with role-based access
When not to use Chat Thing
- Anyone needing crisis or distress support, which the vendor itself explicitly advises against
- Organisations that require EU data residency, since primary storage sits in the United States
- Teams standardised on Microsoft Teams, which is not supported and must be built via the API channel
- Users expecting a native mobile or desktop app, as Chat Thing ships only as a web product
- Budgets capped below the Standard tier, which gates messaging channels, API access, model choice and live human takeover
How to use Chat Thing
A typical end-to-end flow, from setup to results.
- Create a free account: no credit card is required on the Free plan
- Create your first agent from the dashboard
- Connect your sources: a website URL to crawl, a sitemap, a Notion workspace, files, a YouTube channel or an RSS feed
- Let the first sync index the content, which consumes storage tokens
- Write the system prompt and user prompt, and set the bot's creativity level
- Choose the underlying model, available from the Standard plan upward
- Switch on the power-ups you need: any API, Zapier, Notion, Shopify, web browsing or Talk to a Human
- Style the widget: primary and secondary colours, light or dark theme, avatars, custom CSS and example messages
- Test it from the bot dashboard, or build a full test suite if you are on Enterprise
- Go live by pasting one line of code into your site, or enable Slack, Discord, WhatsApp, Telegram, email or the API, then watch conversations and take over when a human is needed
Pros & Cons
Pros
- Genuinely fast to launch: one line of code, no developer, and a permanent free plan with no card required
- An unusually explicit privacy posture: never trains on customer data, enforces Zero Data Retention account-wide, and fails a request rather than routing it to an endpoint that would retain it
- Complete and readable legal apparatus: published DPA, dedicated GDPR page, dated sub-processor list with places of processing and 30 days' notice of change
- Very wide model choice, switchable per bot, with no vendor lock-in
- True omnichannel from a single bot, rather than one bot per platform
- Real tool calling that looks up orders and updates records, not just retrieval-based answers
- À-la-carte add-ons let you raise one limit without jumping a whole pricing tier
Cons
- Token-based billing on two separate meters, message and storage, makes real cost hard to estimate up front
- The Standard tier at $49/month gates messaging channels, API access, model choice and live human takeover
- Analytics is Pro-only and automated bot evaluation is Enterprise-only
- The Free plan includes no power-ups, so neither escalation route to a human is available on it
- No EU data residency: primary storage is in the United States
- Voice features fall outside Zero Data Retention, calling OpenAI directly under its standard abuse-monitoring retention
- The site contradicts itself on two figures: 100+ versus 20+ available models, and 80k versus 100,000 free-plan message tokens
Pricing & Plans
A permanent free plan is available at $0 per month, with one agent and one data source and no credit card required. The cheapest paid tier is Basic at $14 USD per month, falling to an equivalent $11.67 per month on annual billing, which grants two months free across all tiers. All charges are billed in US dollars. No time-limited free trial is documented: the free plan is the trial.
- 1 agent
- 1 data source
- 80
- 000 message tokens (the FAQ table states 100
- 000)
- web channel included
- no power-ups
- no credit card required
- 3 agents
- 3.5M message tokens
- 3M storage tokens
- 2 team members
- website widget with custom theming
- power-ups
- email support
- 6 agents
- 19M message tokens
- 10M storage tokens
- 4 team members
- model choice across GPT
- Claude and Gemini
- live human takeover
- all channels plus API access
- 20 agents
- 38M message tokens
- 40M storage tokens
- 6 team members
- chat analytics and higher limits throughout
- 50 agents
- 120M message tokens
- 120M storage tokens
- 8 team members
- bot testing with evaluations
- unlimited data sources
- Custom plans are available on request by writing to hello@chatthing.ai
- $250/year per 5M message tokens
- $50/year per 5M storage tokens
- $80/year per extra bot
- $40/year per extra data source
- $20/year per extra team member
- $250/year per custom domain
Data, GDPR & hosting
A consolidated view of how Chat Thing handles your data.
GDPR overview
Compliance is claimed explicitly: a dedicated page states that Chat Thing is UK-based, stores data in the US and "complies with the GDPR framework". The concrete evidence is unusually complete for a company this size: registration with the UK Information Commissioner's Office under number ZB671744, a named data protection officer (Jozef Maxted, zef@chatthing.ai), a published Data Processing Agreement with a signed copy on request, a sub-processor list giving each vendor's purpose and place of processing, 30 days' notice before any sub-processor changes with a right to object, a processing-activity table with legal bases, and Standard Contractual Clauses cited for transfers. No SOC 2 or ISO 27001 certification is claimed. The privacy policy and terms are both dated 21 March 2024, and the policy visibly reuses multi-jurisdiction template blocks whose storage claims contradict each other.
Who owns the data?
You keep ownership. The terms state that you retain your intellectual property rights over the content you submit and that Chat Thing will never claim ownership of it, but you grant the vendor a non-exclusive, royalty-free, transferable, sub-licensable, worldwide licence to use, distribute, modify, copy, display, translate and create derivative works of it. That licence ends when you delete the content or your account, unless the material has already served commercial or sponsored content. For end-user conversations the roles reverse: the GDPR page and the DPA make the bot's creator the data controller and Chat Thing only the processor. The privacy policy also commits that collected data is never used to train AI models.
Reuse rights
You may reuse your own content and your bot's output freely: nothing in the terms restricts it, and every conversation gets a unique shareable link. You can export or delete your data from the dashboard at any time, and request full account deletion by writing to support@chatthing.ai. The vendor's own reuse is limited by an explicit commitment never to train AI models on collected data. The restrictions run the other way, on how you use the service: no reverse engineering, no mirroring of materials, no harvesting of other people's data without their consent, no spam, and no hateful, violent or racist content. Note that purchases are non-refundable and cancellation only takes effect at the end of the paid term.
Data retention & training
Hosting summary
Chat Thing Ltd is UK-based, but the GDPR page states plainly that data is stored in the US, and the terms confirm the services are hosted in the United States. The sub-processor list, last updated 29 July 2026, is the most precise source: Supabase for the primary database, file storage and authentication, Railway for compute, Cloudflare for CDN and DNS, OpenRouter for model routing, OpenAI for voice, Nylas for optional email connectivity, Stripe for payments and ConvertKit for marketing email, all processing in the United States; Plausible and PostHog handle web analytics in Germany. Data protection agreements are in place with each, and Standard Contractual Clauses are cited for transfers, with 30 days' notice and a right to object before any sub-processor changes. The DPA defers the place of storage to the base agreement and permits transfers outside the EU, EEA and Switzerland. Two caveats: the resolved domain IP is a Cloudflare anycast node and says nothing about storage location, and the privacy policy's template sections variously claim UK and Canadian storage, contradicting the rest of the site.
Things to keep in mind
Risks and trade-offs to weigh before adopting Chat Thing.
- A poorly scoped bot can state something wrong with complete confidence; the vendor itself recommends telling users plainly that they are talking to a bot
- The vendor explicitly warns against using it for crisis support, since a chatbot will not reliably recognise distress the way a trained person does
- If a bot ever claims to have changed your account, treat it as an error and verify — it has no such power by design
- Answer quality is entirely bounded by the content you feed it: outdated documentation produces a confidently outdated bot
- Your customers' conversations transit US-based sub-processors, and those end users never chose them; you remain the data controller for whatever they type, including sensitive information
- Power-ups give the bot real reach into orders and records, so a misconfiguration has consequences outside the conversation
- Outsourcing the first line of support can quietly cut you off from customer signals and erode the in-house knowledge that used to answer those questions
Setup & Integrations
Technical difficulty
Low for the intended path. The vendor claims an agent live in five minutes: paste one line of code, no developer needed, and it works on WordPress, Shopify, Squarespace or custom sites. Connecting a URL, a Notion workspace or files needs no technical skill. Optional depth exists in the system prompt, creativity setting and custom CSS. Real engineering only appears at the edges — the API channel, the SDK, HMAC-signed webhooks, custom API power-ups and a custom domain. The MCP server even lets an AI assistant build and test the bot for you.
Deployment
Integrations
Supported languages
Behind Chat Thing
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Chat Thing.
Frequently asked questions
What does Chat Thing actually do?
Is there a free plan, and what does the cheapest paid tier cost?
Where can I deploy a bot?
Which AI models can I use?
Is my data used to train AI models?
Do I need to change a setting to keep my data private?
Where is my data hosted, and is there a DPA?
Can the bot hand a conversation over to a human?
How does billing work?
Is there a mobile app, and does it integrate with Microsoft Teams?
Should you pick Chat Thing?
Chat Thing occupies a clear niche: AI customer support for small and mid-sized teams that have no developer to spare. Its strongest differentiator is not the chatbot itself, which many vendors now offer, but the documented privacy posture behind it. Never training on customer data, enforcing Zero Data Retention at account level, and failing a request rather than silently routing it to a model endpoint that would retain it, are commitments most competitors of this size do not put in writing. They come with a matching legal apparatus: a published DPA, a dedicated GDPR page, ICO registration, a named DPO and a dated sub-processor list with 30 days' notice of change. Add very wide model choice with no lock-in and genuine tool calling, and the product is more serious than its playful name suggests.
Three reservations deserve weight. Token-based billing on two meters means you should model your traffic before committing. The $49 Standard tier is the real entry point for anyone who needs messaging channels, API access, model choice or live human handoff — Basic is thin by comparison. And there is no EU data residency: primary storage is in the United States, while the privacy policy, unchanged since March 2024, carries template blocks whose storage claims contradict one another and the rest of the site.
The company is young but visibly active: incorporated in England and Wales in 2023, two directors, more than 10,000 claimed users, 4.6 out of 5 across 28 reviews, and a changelog still shipping new models in mid-2026. Recommended for a support team buried in repetitive questions that wants traceable compliance on a modest monthly budget. Not recommended where EU residency, crisis support or a mobile app is a requirement.
- Choosing a selection results in a full page refresh.
- Opens in a new window.