DocJuicr

A chatbot answers questions about a document. DocJuicr squeezes the whole pile into finished work, and shows you the page for every line.

The general assistants are upfront about their own limits. Microsoft says that past 20 files, Copilot will search the full contents of the 20 most relevant files only (learn.microsoft.com). A NotebookLM notebook holds 50 sources on the free tier and 300 on Pro (support.google.com). Claude Projects switch from reading everything to a retrieval search once project knowledge exceeds the model’s context window (support.anthropic.com).

DocJuicr is built to work through a whole set instead – hundreds of documents and thousands of pages, including emails, scans and spreadsheets – and every AI-drafted entry quotes the passage and page it came from, so you can check it.

Anyone who needs to turn a pile of documents into a timeline, a who’s who or a spreadsheet: people running their own dispute or complaint, workplace and HR investigators, family and local historians, students and researchers, and professionals such as lawyers, analysts, accountants, auditors and professional writers.

• Building a timeline from the dates in a set of documents and emails - by rule, or drafted by AI - and AI-drafted 'who's who' network charts of the people, companies and places, with every AI entry quoting the passage it came from
• Chatting with your documents and saving the cited answers as a Word file
• Summarising a document or a bundle in the style you need (memo, brief, LinkedIn post and more)
• Flow charts of money movements from bank statements and similar records
• Redacting or pseudonymising names, companies, money and other categories, with AI finding the ones the rules miss
• Highlighting key terms in customisable colours, tabulating the highlighted passages with your notes, or transposing them back onto your original Word document
• Transcribing audio and video recordings, and running OCR on scans with no text layer
• Threading a mailbox into one de-duplicated, chronological table
• Saving a playbook or a workflow, so a function - or a sequence of functions - runs the same way every time
• Extracting tables to Excel, pulling out images, inspecting metadata, and cleaning, merging, splitting and unlocking PDFs
• Working from PDFs, Word, Excel, PowerPoint, email, text, scans and recordings - converted on the way in

You can create an account in three ways: with a Google account, with a Microsoft account, or with any email address and a password.

Google or Microsoft is the quickest way in, and the one we suggest – there is no password for you to manage, or for us to hold. Any personal Gmail account works. For Microsoft, a personal Hotmail, Outlook or Live account works; a corporate Microsoft 365 account may need your administrator’s permission first.

If you would rather use your own email address, choose “Sign in with Username and password” on the sign-in page, enter your email address and a password, and confirm when asked to create the account. Accounts made this way need an authenticator app – Google Authenticator, Microsoft Authenticator or 1Password, for instance – to supply a code each time you sign in. You are given recovery codes when you set it up; keep them somewhere safe.

Once you've created your account, choosing the right plan is easy. By default you will start on a free demo plan. We have three paid subscription plans, designed for light, medium and heavy users. The demo plan and the paid plans all include the full set of document tools and their own monthly AI page allowance for the AI features – AI timelines, network charts and summaries: 50 pages on the demo plan, 500 on Small, 1,000 on Medium and 2,500 on Large. The plans have different limits such as the page limit for each document, a monthly limit of pages processed, and an overall storage limit. These limits are explained on the Subscription page. You can upgrade at any time, and downgrade with effect from the next billing cycle date. All prices shown include GST and tax invoices can be downloaded on demand.

DocJuicr uses machine learning language models together with defined patterns, and large language models to draft timelines, network charts, summaries and chat answers from your documents. Every AI-drafted entry quotes the passage it came from, and where a quote cannot be matched back to the document it is flagged rather than given a page number.

No. Your documents, prompts and results are never used to train AI models - not by us, and not by our AI providers. Anthropic’s commercial terms expressly bar it from training models on customer content from its API, which is how DocJuicr connects, and Amazon Bedrock (which powers parts of our document chat) states that customer inputs and outputs are not used to train any model. Your documents are processed to produce your result, stored for you in your account, and nothing more.