ChattyDocs
ChattyDocs turns a set of PDFs, text files or crawled websites into a queryable dataset, then answers questions about it with the exact document and page cited, through a web app, an embeddable chat, Telegram or a GraphQL API.
What is ChattyDocs?
ChattyDocs is a document question-answering assistant built around one idea: the dataset, a collection of documents you assemble and then talk to. You fill it from three sources, uploading several PDFs at once from your device, adding any plain text file, or simply pasting the address of a website, which the service downloads and parses on its own. Indexing runs in the background, so nothing keeps you waiting, and the interface shows how much of your monthly upload quota each file or site will consume. Documents can be added or removed at any time.
The second step is configuration, and it is unusually detailed for a product at this price. You choose the chat model and decide how the available room is split between context, history and output. A temperature setting controls how creative or how deterministic the answers should be. A system prompt gives the assistant a name, a personality, a response language and whatever behaviour you want to impose. A dataset can also be made public and shared by link with other ChattyDocs users.
Then you talk to it. Every answer comes back with its sources: for a multi-page PDF, the document name and the page number; for a crawled site, a link. Chat sessions are multiple and persistent, so a conversation resumes where it stopped, and a session context can be written once and injected into every query. The same dataset is reachable from the web application, from a chat embedded on a page and from a Telegram bot, with all sessions visible in one place. Everything asked and answered exports in a single click.
Developers get a documented GraphQL API at api.chattydocs.com/graphql, with mutations to create datasets, add or delete documents and query a dataset, authenticated by a bearer key, from the middle plan upwards. On that route the text extraction is yours, since the API accepts text only. A public Telegram demo bot lets anyone try the product before signing up, and organisations that cannot let data leave their walls are offered a Docker self-hosted package on request.
What it does
- Turn PDFs, text files and entire websites into a single queryable dataset
- Answer questions on that dataset and cite the source document and page behind each answer
- Serve the same dataset through a web application, an embedded chat and a Telegram bot
- Shape the assistant with a system prompt setting its name, personality and response language
- Tune the chat model, the context, history and output sizes, and the temperature
- Keep several persistent chat sessions and pick any of them up later
- Export every question and answer in one click, or drive the whole thing from a GraphQL API
When to use ChattyDocs / When not to
A quick filter to help you decide if ChattyDocs is the right fit.
When to use ChattyDocs
- Researchers and academics working through a large body of papers, books or reports who need every answer traced back to a document and a page number
- Support and knowledge-base owners who want a Telegram bot or a chat embedded on a web page answering from their own documentation
- Solo consultants, freelancers and small teams looking for a document assistant with a 9 USD monthly entry point rather than an enterprise contract
- Developers who would rather drive datasets, uploads and queries from a documented GraphQL API than from a user interface
- Organisations under strict hosting constraints, who can take the Docker self-hosted package and keep the database, the vector store and the services in-house
When not to use ChattyDocs
- Anyone looking for a permanently free tool: the trial stops after 20 queries and 900 kilobytes of text, and every plan after that is paid
- Teams with heavy monthly volumes on a small budget: the entry plan caps at 7 megabytes of uploads and 1,000 queries per month, and holds no API access
- Buyers who need a signed data processing agreement, a named EU Article 27 representative or a SOC 2 or ISO 27001 certification, none of which exists here
- European controllers uneasy about United States processing: the privacy policy states that data is transferred there and processed there, without naming a transfer mechanism
- Users expecting native mobile applications or office integrations such as Slack, Notion or Google Drive: Telegram is the only third-party channel offered
How to use ChattyDocs
A typical end-to-end flow, from setup to results.
- Create an account on the ChattyDocs client application and start the free trial, which asks for no credit card
- Create a dataset, then upload your PDFs and text files or paste the address of a website to be crawled
- Watch the quota indicator to see how much of your monthly upload allowance each file or site consumes
- Let the background indexing finish, adding or removing documents at any point without waiting
- Choose the chat model and set how context, history and output share the available room
- Set the temperature to make answers more creative or more deterministic
- Write the system prompt that gives the assistant its name, its personality and its response language
- Open a chat session, ask your questions, and check the document and page cited under each answer
- Reuse persistent sessions to resume a conversation, and define a session context injected into every query
- Publish the dataset by link, attach it to a Telegram bot, or call the GraphQL API with a bearer key for programmatic use
Pros & Cons
Pros
- Every answer carries its sources, down to the page number inside a multi-page PDF
- One dataset, three ways in: web application, embedded chat and Telegram bot, with all sessions gathered in one view
- Control rarely offered at this price: chat model, context, history and output sizes, temperature and system prompt
- A whole website becomes a data source by pasting its address, with indexing then running in the background
- Free trial without a credit card, and monthly plans cancellable at any time
- Publicly documented GraphQL API, complete schema and a working Python example on the site
- Docker self-hosted option for organisations that cannot let their data leave their own infrastructure
Cons
- No permanent free plan: the trial stops at 20 queries and 900 kilobytes of text
- Tight quotas on the entry plan, 7 megabytes of text and 1,000 queries per month, and no API access before the 49 USD tier
- No postal address, no contact page and no legal notice: a single email address serves support, legal, GDPR and DMCA matters
- No published data processing agreement, no EU Article 27 representative and no security certification of any kind
- Data is transferred to and processed in the United States, with no transfer mechanism named in the policy
- Answers already produced survive the deletion of their source document and may still contain excerpts of it
- Site, sitemap and legal texts all date from 2023, and no list of supported interface or processing languages is published
Pricing & Plans
There is no permanently free plan. A free trial is available without a credit card, limited to 20 queries and 900 kilobytes of text for uploading. The lowest paid entry point is 9 USD per month, billed monthly and cancellable at any time.
- 20 datasets
- 30 documents per dataset
- 1 Telegram bot
- 7 megabytes of text uploaded per month
- 1
- 000 dataset queries per month
- export of queries and answers
- 100 datasets
- unlimited documents per dataset
- 10 Telegram bots
- 35 megabytes of text per month
- 5
- 000 dataset queries per month
- export and API access
- unlimited datasets and documents
- 30 Telegram bots
- 200 megabytes of text per month
- 30
- 000 dataset queries per month
- export
- API access and priority support
- full self-hosting with Docker covering database
- vector store
- web application and all services
- priced on request by email
- 20 queries and 900 kilobytes of text available for uploading
- no credit card required
Data, GDPR & hosting
A consolidated view of how ChattyDocs handles your data.
GDPR overview
GDPR is addressed head-on but sparingly. Section 10 of the privacy policy is devoted to residents of the European Union and the EEA, links to Regulation 2016/679 and lists the rights of access, update, erasure, rectification, objection, restriction, portability and withdrawal of consent, all exercised by email to support@chattydocs.com after a possible identity check, with the right to complain to a national supervisory authority stated in full. ChattyDocs declares itself the data controller. Beyond that the implementation is thin: no data protection officer is named, no Article 27 EU representative is designated, no data processing agreement is published, and section 7 announces a transfer to the United States without naming any transfer mechanism. Separate CalOPPA and CCPA sections cover Californian residents. The policy has been in force since 2 April 2023.
Who owns the data?
The terms are explicit that you keep every right to the content you upload, and that you remain responsible for its legality and for protecting those rights. ChattyDocs reserves the right, but takes on no obligation, to monitor and edit user content. On personal data, the privacy policy names ChattyDocs as the data controller, states that nothing is sold or rented to third parties, and confirms that you are the sole owner of your personal data and may request its disclosure or deletion at any time. Uploaded documents and the embeddings derived from them sit in a database the vendor describes as being under its own control, with no third party involved. Feedback is the one exception: sending it transfers its ownership to the company.
Reuse rights
Nothing in the terms restricts what you may do with the material you bring in or with the answers you get back. You keep your rights to your own content, and the interface offers a one-click export of every question and answer, so reuse needs no permission. The counterpart sits on the vendor's side. Uploaded files are parsed and their extracted text stored in a database ChattyDocs controls; that text is then sent to the external service OpenAI to compute embeddings, which return to the same database. Through the API the extraction step is yours, since only text is accepted. Personal data, meaning your email address, cookies and usage data, serves service delivery, support, analytics, billing, account notices and marketing for similar products that you may opt out of. Google Analytics measures audience, Stripe processes payments without ChattyDocs storing card details, and Do Not Track signals are honoured.
Data retention & training
Hosting summary
The site names no hosting provider and no data centre. What it does state, in section 7 of the privacy policy, is the jurisdiction: users located outside the United States are told that their information, personal data included, is transferred to the United States and processed there. No transfer mechanism, whether standard contractual clauses, an adequacy decision or a framework certification, is named anywhere. Uploaded documents and the embeddings derived from them are stored in a database described as being under the vendor's sole control, with no third party involved; the only outbound step is the extracted text sent to OpenAI to compute those embeddings. The public website is served behind Cloudflare, whose anycast address says nothing about where anything is actually stored. Organisations that cannot accept these terms are offered a Docker self-hosted package covering the database, the vector store, the web application and all services, priced on request.
Things to keep in mind
Risks and trade-offs to weigh before adopting ChattyDocs.
- Answers already generated outlive the documents they came from and may keep excerpts of text you believed deleted; removing them takes an email request rather than a click
- Uploaded text leaves the vendor's database to reach OpenAI for embedding generation, and the site never states whether customer data is used to train models
- Data is transferred to and processed in the United States with no transfer mechanism named, something a European controller has to assess before uploading anything sensitive
- A single email address stands behind support, legal, GDPR and DMCA requests, and no postal address or company registration appears anywhere on the site
- Sourced answers feel authoritative, and a reader who stops opening the cited page ends up trusting a retrieval step that is only as reliable as the documents behind it
- Handing a whole corpus to an assistant can quietly replace reading it, and the mental map you build by working through documents yourself is exactly what gets outsourced
- Subscriptions renew automatically each month and fees are non-refundable, while strict monthly quotas push towards the next tier as soon as usage grows
Setup & Integrations
Technical difficulty
Low on the standard route: create an account, drop in files or paste a website address, and start asking. Indexing is automatic. The advanced settings, covering chat model, context, history and output sizes, temperature and system prompt, are open to everyone but assume you understand what you are adjusting. The API route is developer work: fetch a key from the site, call the GraphQL endpoint with a bearer header, compress and encode your text, and handle PDF extraction yourself since only text is accepted. Self-hosting with Docker is an infrastructure project of its own.
Deployment
Integrations
Behind ChattyDocs
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What can I put into a ChattyDocs dataset?
Do the answers say where they come from?
Can I use ChattyDocs outside a browser?
Is there a free plan?
How much does the cheapest paid plan cost?
Which plans include the API?
Can I host ChattyDocs myself?
What happens when I delete a document?
Who processes my data?
Is there a minimum age?
Should you pick ChattyDocs?
ChattyDocs does one thing and makes it legible. You gather documents into a dataset, you shape how the assistant should behave, and you ask questions that come back with the page they were drawn from. The control on offer, covering the chat model, the split between context, history and output, the temperature and a full system prompt, is unusual at a 9 USD monthly entry point, and the three access routes, web application, embedded chat and Telegram bot, let one knowledge base serve an internal team and an outside audience without duplication. The publicly documented GraphQL API, with its complete schema and a working Python example, makes the product genuinely usable as a component rather than only as an interface.
The reservations are about governance rather than features. There is no permanently free plan, and the entry tier caps quickly at 7 megabytes of text and 1,000 queries a month. More importantly, the publisher exposes no postal address, no contact page and no legal notice; a single email address answers for support, legal, GDPR and DMCA matters. No data processing agreement, no EU Article 27 representative and no security certification is published, and the privacy policy states plainly that data is transferred to and processed in the United States without naming a transfer mechanism. Answers already generated outlive the documents behind them and may keep excerpts of deleted text.
The site, its sitemap and its legal texts all date from 2023, yet the application, the API endpoint and the public demo bot were all verified in service at the end of August 2026. For an individual, a small team or a developer with non-sensitive material, ChattyDocs is a well-shaped and affordable option. For a regulated organisation, the Docker self-hosted package is the route that answers most of the objections above.
- Choosing a selection results in a full page refresh.
- Opens in a new window.