Frequently asked questions

docAnalyzer is a source-grounded AI workspace for working with your documents. You upload your sources (PDFs, Office files, EPUB, spreadsheets, and more), then work with them through three chat modes that hand work back and forth:

• Ask docAnalyzer: the workspace-aware starting point. It knows everything you've uploaded, helps you figure out what you want, and opens the right surface for the job.
• Focus chat: deep analysis on a chosen set of sources. It searches, reads, and cross-references across multiple rounds; every claim links back to the exact page or section it came from; and any answer can attach downloadable files (PDFs, spreadsheets, reports) you can reuse.
• Co-work chat: collaborative editing on a note. The AI drafts and revises alongside you in a full editor, turn by turn.

Beyond a single chat, Workflows run the same task across every document in a set (summarize, extract structured data, audit against a reference, and more), with results you open and keep working with in Focus chat.

It's built for expert practitioners: researchers, analysts, legal and finance teams, anyone reasoning across hundreds of long documents rather than skimming one.

Sign in, then upload your documents from the Add Documents panel (drag and drop, paste a URL, paste from clipboard, or import from Zotero or Mendeley). Each one is analyzed and indexed automatically.

From there:

• Ask Ask docAnalyzer what you want to do; it knows your whole workspace and opens the right surface for the job.
• Open a Focus chat on a document, a label, or any set of sources to ask questions and get cited answers, plus downloadable files when you need them.
• Save an answer as a Note, then refine it with Co-work chat.
• Run a Workflow when you have the same task to do across many documents at once.

Team and Enterprise plans add shared workspaces so colleagues can work from the same sources.

You can upload PDF, Word (DOCX), PowerPoint (PPTX), OpenDocument (ODT/ODP), Excel (XLSX), EPUB, HTML, Markdown, Rich Text (RTF), and plain text (TXT). CSV and JSON are available with experimental mode on. The platform handles everything from intricate legal contracts to financial reports and scholarly papers.

Scanned or image-based pages are processed with OCR automatically during analysis, and the model also sees charts, diagrams, and figures inside your documents. Enhanced OCR is available when a document needs higher extraction accuracy.

Standalone image, audio, and video files aren't supported as uploads yet. If your document is in a format we don't support, converting it to PDF gives you full compatibility.

OCR (Optical Character Recognition) converts images of text into machine-readable data. docAnalyzer uses a two-tier approach:

Automatic OCR runs during the standard analysis of every document. When pages are detected with little or no selectable text, OCR is automatically applied to extract content. This is included at no extra cost and works well for most scanned documents.

Enhanced OCR is available as a premium option for documents where automatic processing wasn't sufficient. It provides higher accuracy extraction and can be launched from the document action menu. Enhanced OCR uses credits based on the number of pages processed.

OCR is useful for:

• Scanned PDFs: Documents created by scanning physical papers, making the text appear as images rather than selectable text
• Image-based documents: Screenshots of documents, photos of papers, or PDFs created from images
• Legacy documents: Older documents digitized through scanning without text recognition
• Handwritten content: OCR can recognize printed text from handwritten documents (support varies depending on language and quality)

Without OCR, these documents would be impossible for AI to analyze since the text exists only as pixels in an image. With automatic OCR built into the analysis pipeline, most documents are ready to chat with immediately after upload.

Both automatic OCR and Enhanced OCR support text extraction in over 40 languages, making them suitable for processing documents from around the world. The supported languages include:

European Languages: Afrikaans, Albanian, Belarusian, Bulgarian, Catalan, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Icelandic, Italian, Latvian, Lithuanian, Macedonian, Norwegian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swedish, Turkish, Ukrainian

Asian Languages: Arabic, Bangla/Bengali, Chinese, Hindi, Indonesian, Japanese, Korean, Malay, Marathi, Nepali, Persian, Tagalog/Filipino, Vietnamese

This extensive language support ensures that scanned PDFs and image-based documents in these languages can be accurately converted to searchable text, enabling full AI analysis and interaction capabilities. The language is automatically detected and the appropriate recognition model is applied for optimal accuracy.

Absolutely, docAnalyzer welcomes documents in various languages, not just English. Its capability extends to parsing and analyzing content across a broad spectrum of languages. However, please note that the intricacy of the language and the document's complexity could influence the precision of the analysis. In particular, documents in less commonly used languages or those with highly complex content may yield less optimal results.

Signing in sends a 6-digit code to your email, which you enter on the code page. If it doesn't work:

1. Check your spam or junk folder, then wait a moment — delivery can lag.
2. Codes expire and are single-use. If yours is rejected, request a new one and enter the most recent code.
3. Ask IT to allow emails from docanalyzer.ai and add [email protected] to safe senders.

Once you're in, you can set a password in your account settings and sign in with email + password instead. Still stuck? Contact [email protected].

Data security is our utmost priority at docAnalyzer. We adhere to rigorous data protection and privacy standards to ensure your documents and information remain secure. Our storage infrastructures is compliant with ISO/IEC 27001, 27701 and healthcare data hosting. For enterprise customers with specific security requirements, our Enterprise plan includes additional security features like SAML sign-on. For more detailed information, please check out our Privacy Policy.

To delete a workspace, you must first ensure it is completely empty. This involves two main steps:

1. Delete all documents: Remove every document currently stored within the workspace.
2. Delete all labels: Remove all labels that have been created or applied within that workspace.

Once all documents and labels have been successfully deleted from the workspace, the option to delete the workspace itself will become available on the main documents page if the workspace is empty. Please be aware that deleting a workspace is an irreversible action. All data associated with that workspace, including any remaining configuration or chat history, will be permanently removed.

Yes, our platform is designed to be compatible with all devices. You can upload documents and interact with the AI on your mobile device, desktop, or tablet.

Yes. We ship improvements continuously, shaped by user feedback and advances in the underlying AI models. We can't pre-announce specifics, but updates appear in the app as they land.

docAnalyzer's Community plan is free forever (not a trial) and includes document analysis, chat, notes, and monthly credits, with monthly upload limits. For more headroom, the paid plans (Basic, Pro, Pro Plus, Team, Enterprise) raise the limits and add features like sharing, the API, Lab models, and team workspaces. See pricing for current limits and prices.

docAnalyzer has a free Community plan plus five paid tiers, each adding headroom over the one below:

• Community (free): document analysis, chat, notes, and monthly credits, within monthly upload limits.
• Basic: the paid entry point with the full feature set (chat sharing, Chatbots, Notes, Lab models, BYOK, API access), but with caps on multi-document chat size, document upload size, storage, workspaces, labels, and monthly credits.
• Pro: everything in Basic with those caps lifted: no multi-document chat-size limit, larger uploads and storage, unlimited workspaces and labels, and more monthly credits.
• Pro Plus: Pro with higher credit ceilings and priority models.
• Team: everything in Pro, plus shared team workspaces, collaboration, and per-seat storage.
• Enterprise: everything in Team, plus SAML sign-on, dedicated support and SLA, and custom seat counts.

Every paid plan includes API access. The exact per-tier limits, credits, and prices live on the pricing page, which is always current.

"Cancel Anytime" means that there is no long-term contract or commitment when you subscribe to any paid plan. You can choose to cancel your subscription at any time without any penalties or additional charges. Please note that cancellation will take effect from the next billing cycle, and no refunds are provided for the current billing period.

Canceling your subscription with docAnalyzer is a straightforward process. Navigate to your account page and click on the "Cancel Plan" button. This will take you to a dedicated cancellation page where you can provide feedback about your reasons for canceling and confirm your decision. After confirming your cancellation, you'll continue to have access to your current plan features until the end of your billing period. Please note that while your subscription will be canceled, your account and any free features will remain accessible.

docAnalyzer uses the Stripe payment solution that accepts all major card networks like Visa, MasterCard, AMEX, Discover and Union Pay. Depending on your location, other methods and wallets (like Apple Pay and Google Pay) may also be available.

Credits are a spending currency on docAnalyzer. You set a credits budget, a spending cap for each request. The AI uses what it needs and stops when the budget runs out. Simple questions may use only a fraction of your budget.

Credits are consumed by: advanced AI models, API calls, OCR (optical character recognition), and AI agent workflows. Whether a chat interaction costs credits depends on two factors: the model and the dataset size. Small datasets with the standard model are typically free, but larger datasets require the AI to work through multiple agentic cycles, which costs credits regardless of the model.

You can purchase credit bundles as needed. Bundle credits never expire. All subscription plans also include a monthly allocation of free credits, with Team and Enterprise plans offering significantly higher allowances. See credit add-ons.

Credits for the plan are renewed at the beginning of every calendar month. For example, if you purchased your plan on June 12, your credits would first be renewed on July 1 and then on the first of each subsequent month. Please note that unused monthly credits do not roll over to the next month. However, credits acquired through bundles do not have an expiration date.

Monthly credits included in your subscription plan are valid only for the current calendar month. Regardless of when you subscribe during a month, your monthly credits are usable until the end of that calendar month. For example, if you subscribe on the 20th of May, your credits are valid until May 31st. At midnight UTC on the first day of each new month (June 1st in this example), your monthly credits automatically reset to the full amount allocated by your subscription plan. Any unused credits from the previous month do not roll over and cannot be carried forward. This reset happens automatically without any action required on your part.

The "fair use" policy in the Basic plan allows you to ask up to 10,000 questions per month. This limit has been set to ensure that all users get fair and optimal access to our services. If a user consistently exceeds this limit, we may need to review their usage and potentially introduce additional charges or restrictions to maintain a balanced and efficient service for all users. For users requiring higher usage volumes, we recommend upgrading to our Pro, Team, or Enterprise plans, which offer unlimited chat interactions.

You can download your invoices and receipts directly from your billing portal. Simply navigate to the account page and click on the "Billing Portal" button. This will redirect you to your personal billing management area where you can view and download all available invoices and receipts.

Whether a chat interaction costs credits depends on the model and the dataset size:

• Standard model + small dataset: Typically free. When your documents fit within the AI's context, it answers in a single pass with no credits consumed.
• Large datasets (any model): The AI enters an agentic cycle: searching, reading, and reasoning across your documents in multiple steps. Each step costs credits. The larger the dataset, the more steps may be needed.
• Advanced models: Always cost credits, regardless of dataset size. They are clearly marked in the model selection dropdown.
• Other cost factors: Higher context density (2×, 3×) and higher thinking effort increase credit usage by processing more data or performing deeper reasoning.

The chat interface displays your credits budget (maximum cap) before you send. Actual spend may be lower if the AI completes quickly. Team and Enterprise plans include 600 free credits per month to support advanced usage.

Yes. Open a Focus chat from any document, note, or label, or build a dataset from any combination of them. The AI searches and cross-references across the whole set in one conversation, grounding its answers in the specific sources rather than making you open each file by hand. Every claim links back to where it came from.

docAnalyzer combines best-in-class AI models from multiple providers (Anthropic, Google, OpenAI, and others) with our own agentic pipeline that searches, retrieves, and verifies information from your documents across multiple rounds. This multi-step approach grounds every answer in actual document content rather than relying on a single retrieval pass. While our system provides highly accurate analysis, we recommend users independently verify critical information for assurance.

The system now automatically determines the optimal response length based on your question and the available context. No manual setting is needed in most cases.

If you find that answers are too short for your needs:

• Increase your credits budget: A higher budget gives the AI more room to research thoroughly and produce detailed responses.
• Be specific in your prompt: Ask for detailed explanations, step-by-step breakdowns, or comprehensive analysis to signal that you want longer output.
• Increase context density: Higher density (2×, 3×) lets the AI process more of your documents per agentic cycle, which can improve answer depth.

Yes! A single document can have multiple labels at once, which allows it to be part of different chats or sets simultaneously. This is called cross-labeling, and it makes it easy to reuse the same document in different conversations without duplicating it.

Hitting the cap because you pasted document content? Add it as a document instead — use the Add as document button on the message box, or Add Documents → Clipboard — then ask your question about it. The AI can search and cite documents; oversized prompts just crowd out document context.

By default, the maximum prompt length is capped at 10% of the per-turn context capacity. The remaining 90% is reserved for document content to keep answers grounded.

Ways to get a longer prompt:

• Increase context density in chat settings. Higher density (2×, 3×) scales the per-turn capacity, so the prompt cap grows proportionally.
• Advanced settings (paid plans): adjust Prompt Capacity from 5% up to 60% in the chat settings' Advanced panel. Higher percentages give more room for your prompt but leave less capacity for document context.

The system now automatically determines the optimal response length, so truncation is rare. If it does happen, it's usually because the AI model reached its maximum output capacity for that response.

To reduce the chance of truncation:

• Try a different AI model. Some models support longer outputs than others.
• Break complex questions into smaller, focused questions.
• Increase your credits budget to give the AI more room to work.

The main factors that influence credit usage:

• AI Model: Advanced models cost more per token than the standard model. Models that cost credits are clearly marked in the dropdown.
• Dataset size: Larger document sets require the AI to work through more agentic cycles: searching, reading, and reasoning across your documents. Each cycle costs credits.
• Context Density: Higher density multipliers (2×, 3×) let the AI process more context per agentic cycle, which uses more credits but enables deeper analysis.
• Thinking Effort: Higher thinking levels enable more thorough reasoning but require more computational resources.

The chat interface displays your credits budget (maximum cap) before you send. This is a spending cap: the AI uses what it needs and stops. Simple questions may use only a fraction of the budget. Team and Enterprise plans include 600 free credits per month.

Context density controls how much of your documents the AI can process in a single agentic cycle. When you chat with documents, the AI works in cycles: each cycle it reads a portion of your documents, reasons about it, and may run additional cycles to cover more. The density multiplier scales how much it reads per cycle.

• Standard (1x): The default. Works well for single documents or small collections. The AI processes a comfortable amount per cycle, keeping strong attention to detail.
• 2x–3x: Good for medium-sized datasets (a collection of 20+ documents, or longer reports). The AI covers more ground per cycle, which means fewer cycles overall and faster results.
• 5x–10x+: For very large datasets. The AI takes in much more per cycle, significantly reducing the number of cycles needed.

The trade-off: Higher density lets the AI handle more information per cycle and improves overall speed, but it may lose track of finer details within that larger context. Lower density means the AI examines less text per cycle but with sharper focus on specifics.

When to increase it:

• The AI is running many cycles to answer your question
• You need a broad overview across many documents rather than precise details from one
• Speed matters more than granularity

When to keep it standard:

• You're asking about specific clauses, numbers, or details in a document
• Precision matters more than speed
• Your dataset is small enough that the AI covers it in one or two cycles anyway

The available density options depend on the model you've selected. Models with larger context windows offer higher multipliers.

Short answer: almost never. The "auto" default is the best choice for the vast majority of use cases.

Modern AI models are good at adapting their response length to your question: a factual lookup gets a short answer, a detailed analysis gets a long one. The max answer length setting is a hard ceiling, not a target: the model stops when its answer is complete, regardless of the cap.

This setting is also difficult to reason about because each model has its own built-in output limit (which can't be exceeded regardless of this setting), and for models that support thinking/reasoning, the thinking tokens count toward the total, so the effective answer length is less than the number shown.

Instead of adjusting max answer length, consider adjusting the credits budget. The credits budget controls how many agentic cycles the AI can run, which directly affects how thoroughly it explores your documents. A higher budget means more cycles and deeper analysis, which naturally leads to more comprehensive answers. This is almost always a more effective lever than changing the output cap.

The rare cases where changing it may help:

• Capping it lower (Brief/Concise) when you want deliberately short answers for batch extractions or quick factual lookups. This can save credits on large-scale operations.
• Raising it higher when you notice answers being cut off mid-sentence, which is rare with auto but can happen with certain models on very detailed prompts.

To enhance the quality of responses from docAnalyzer, consider these tips:

• Be Specific: Aim for precise and focused questions. The clearer your question, the more accurate the response.
• One Topic per Question: Stick to asking about one topic at a time. This helps the AI to provide more relevant and detailed answers.
• Step-by-Step Approach: For complex inquiries, adopt a step-by-step interrogation technique. Break down your larger question into smaller, more manageable ones.

Remember, well-structured questions lead to more insightful and useful answers.

When working with large datasets, it's important to tailor your questions for the best results:

• Concept-Oriented Queries: Focus on the concepts or themes of your inquiry rather than exact word matches. Our AI is adept at interpreting and responding to the essence of your question.
• Focused Queries: Clear, discriminative questions help the AI select the most relevant document segments during its agentic cycle.
• Understand Processing: Our AI works through agentic cycles: searching, reading sections, and refining its approach across multiple steps. Even so, not all data can be processed in a single query for very large datasets. If you need a full scan, see Can I ask docAnalyzer to parse, extract, or translate a very large dataset all at once?
• Increase Context Density: Higher density (2×, 3×) lets the AI process more of your dataset per agentic cycle. You can also increase the credits budget to allow more cycles before stopping.

By following these guidelines, you can navigate through extensive information more effectively.

Focus chat is built for precise, focused questions, even on large datasets. It isn't designed to parse, extract, or translate an entire large dataset in a single answer: there are practical limits to how much any AI model can cover in one conversation.

For that, use a Workflow. A Workflow runs the same task across every source in a set, one document at a time, so each gets its own full pass. It's more thorough than a single chat answer for whole-dataset extraction or translation, though it isn't a real-time conversation. Workflow results open in Focus chat, where you can keep working with them.

Yes. docAnalyzer treats structured content, including Excel spreadsheets (XLSX), as first-class documents. We parse sheets into structured data so you can chat about tables, summarize key metrics, and run Workflows like Data Extractor on the same file.

For the best results, keep clear header rows, use one table per sheet when possible, and mention the sheet name or range in your prompt for complex workbooks. If your spreadsheet relies heavily on formulas or pivot tables, exporting to values can improve accuracy.

Yes. When the agent processes your documents, it can generate downloadable files alongside its written reply: PDF reports, charts, diagrams, Word documents, Excel spreadsheets, and CSV extracts. Each one appears as a download card in the chat.

Today these downloads are session-scoped: they live with the conversation that produced them. Long-term saving to your workspace is on our roadmap.

The dataset is set automatically when you spawn a chatbot and it cannot be changed afterward. If you spawn from a document menu, the dataset is that single document. If you spawn from a label menu, the dataset includes all documents and notes attached to that label. If you need a different dataset, create a new chatbot from the desired document or label.

No, at docAnalyzer, we respect your privacy and control over your data. When you delete a chat, it's permanently removed from our active servers. We do not keep a copy or backup of deleted chats. Your interactions are your own, and you have full authority to manage them as you see fit. Please note that this irreversible action, once executed, cannot be undone. For more details, refer to our Privacy Policy.

No, at docAnalyzer, we prioritize your privacy and data sovereignty. When you choose to delete a document, it's completely removed from our active servers. We don't keep a copy, backup, or any residual data from deleted documents. We believe in giving our users full control over their data. Please be aware that this action is irreversible and cannot be undone. For more detailed information, refer to our Privacy Policy.

A Workflow runs the same task across every source in a dataset, one document at a time, so each gets its own full pass instead of competing for room in a single answer. That makes it the right tool for whole-collection jobs: summarizing every document, extracting the same fields from each, auditing a set against a reference, rewriting to remove AI tells, or generating metadata. Workflows handle any document type docAnalyzer supports and return a structured result you open in Focus chat.

From a Focus chat (or a document, label, or selection), open the Workflows picker, choose the one you want, and fill in its short form (for example, the question to ask each document, or the fields to extract). It runs across your dataset and the result opens in Focus chat, where you can read it, download files, and keep working.

docAnalyzer ships six Workflows, and we add more over time:

• Summarizer: condense each document, or a whole set, into a clear summary, with control over length, format, and tone.
• Data Extractor: pull the same fields from every document into structured data (JSON or a spreadsheet) from a template you define.
• Blueprint: audit each document against a reference or checklist and report where it matches or doesn't.
• Humanizer: rewrite text to remove detectable AI patterns while keeping every fact intact.
• SEO Metadata: generate titles, descriptions, and metadata for each document.
• Individual: ask the same question of every document and get one answer per source, as prose or structured JSON.

Pick and configure a Workflow from the Workflows picker in a Focus chat.

Yes. Workflows always use credits, since each document gets its own pass. The cost scales with the number of documents, their size, and the model. You see an estimate before you run, and you can top up with credit bundles that never expire. Paid plans include a monthly credit allowance; see pricing.

Summarizer condenses documents of any length into clear, concise summaries while preserving the most important information. You can configure:

• Custom summary lengths
• Single or multiple related documents
• Automatic key-point extraction
• Output format (paragraph, bullets, or hybrid)
• Tone (neutral, academic, business, or casual)
• Batch processing across many documents

Useful for academic research, legal review, business intelligence, and processing literature.

Data Extractor identifies and pulls specific information from documents, turning unstructured content into structured, actionable data. You can configure:

• Custom templates that define exactly what to extract
• Structured JSON output ready for further processing
• Multiple data types (strings, numbers, booleans)
• Batch processing across many documents at once
• Spreadsheet import to define extraction templates
• Context-aware extraction that pulls only relevant information

Ideal for invoice processing, contract analysis, research data collection, and resume parsing.

Humanizer rewrites AI-generated text through surgical edits (deletions, rephrasing, structural changes) to remove detectable AI patterns while preserving all factual content. You can configure:

• Three intensity levels: light surface fixes, moderate rewrites, or aggressive restructuring
• Tone control: preserve the original voice or shift to casual, professional, or academic
• Keep or flexibly restructure headings and sections
• A three-phase pass targeting document-level patterns, paragraph rhythm, and surface tics
• Removal of common AI tells: stock phrases, tricolons, formulaic transitions, overused vocabulary
• All facts, citations, and data points preserved unchanged

Works with notes and any flowable document (Markdown, plain text, HTML). Ideal for blog content, academic writing, business communications, and content marketing.

Blueprint audits each document in a set against a reference or checklist you provide, then reports where each one matches, falls short, or is missing something. It's built for review-against-a-standard jobs: contracts against a clause checklist, submissions against requirements, reports against a template. You give it the reference and what to check; it runs the same review across every document and returns a structured result you can read or download.

SEO Metadata generates search metadata for each document: titles, descriptions, and related fields ready to use on a page or in a CMS. Point it at a set of documents and it produces consistent, per-document metadata in one run, instead of writing each by hand.

Document analysis tools vary widely in how they retrieve and process information, and those differences directly impact the quality of their answers. docAnalyzer is built around an agent-driven pipeline that actively searches, retrieves, and cross-references documents over multiple rounds, instead of relying on a single retrieval step and hoping the right context appears.

This iterative, evidence-focused approach produces more accurate, grounded, and detailed responses. While no system performs perfectly in every scenario, we continuously benchmark against leading alternatives and refine our methodology to ensure consistently superior results. Our Enterprise plan offers additional customization options for organizations with specific tuning requirements or industry-specific document analysis needs.

docAnalyzer uses a multi-stage pipeline to handle large and complex documents. During ingestion, it classifies each file (paginated, flowable, or structured), extracts key metadata such as language and topics, builds a navigable chunk graph, and creates optimized search indexes.

During a chat session, the AI goes beyond a single vector lookup. It operates within an autonomous tool loop, using semantic search, keyword search, table-of-contents navigation, and targeted section reading. It selects the right tools, reviews the results, and iterates until it gathers sufficient evidence to answer, much like a researcher working through a complex document.

For organizations managing extensive document libraries, our Team and Enterprise plans provide significantly larger storage capacities, enhanced processing capabilities, and collaborative features designed specifically for handling large document collections efficiently.

For big size archives, initial indexing can indeed take some time via the interface or API, but we can provide a custom ingestion script tailored to your needs for faster onboarding. This approach significantly reduces processing time for large volumes of documents and ensures efficient integration into your workflow. Let us know if you'd like to explore that route (it's offered only with the Enterprise plan). Our Enterprise plan is specifically designed to support organizations with large-scale document management needs, providing not only custom ingestion solutions but also dedicated support and advanced security features like SAML sign-on.

docAnalyzer employs a multi-stage pipeline to manage large or complex documents. During ingestion, it classifies each document by category (paginated, flowable, or structured), extracts metadata (language, topics, structure), builds a navigable chunk graph, and creates search indexes. At chat time, the AI doesn't just run a single vector lookup. It operates in an autonomous tool loop with access to semantic search, keyword search, table-of-contents navigation, and targeted section reading. It decides which tools to call, inspects the results, and iterates until it has enough evidence to answer, the same way a researcher would work through a complex document. For organizations routinely working with complex document structures or very large files, our Team and Enterprise plans offer higher document size limits (up to 300MB per document for Enterprise) and enhanced processing capabilities to ensure optimal performance.

Yes. To keep the service fast and reliable, docAnalyzer uses a fair-use and anti-DDoS throttling system that is triggered when a single account uploads a significant number of documents within a short period. This is an account-level control (not platform-wide), and it moderates the upload and processing rate only for that account until usage returns to normal. Each queued document shows an ETA. If you upload many documents on the same day, you may notice short delays as your queue is paced fairly. Current throttling resets daily at midnight UTC. When throttling is active, you can optionally use the Speed Up action to immediately process the currently queued documents in a selected workspace using credits. This only affects the current queue; future uploads remain subject to fair-use rules. Team and Enterprise plans offer higher rate limits and priority processing.