Unstract

Login Schedule a Demo Start for free Turn Unstructured Documents into Structured Data. Instantly. Turn Unstructured Documents into Structured Data. Instantly. The easiest way to parse, structure, and automate document extraction. Production-grade document processing powered by LLMs, built for accuracy, scale, and compliance. Try Demo Playground Start for free Unstract is open-source on Leave us a star Watch 3-min demo Trusted by forward-thinking engineers Handle a wide variety of document formats without manual annotations Bank statements from 200 different banks? Same form with changes across 50 different states? We’ve got you covered with the power of LLMs. Explore Prompt Studio Give your Agents superpowers Empower agents to turn complex, real-world ‘human-only’ tasks into fully automated solutions. Learn more about our n8n integration Do differentiating work—pick your battles Call Unstract APIs and get clean, structured data. Don’t waste your time dealing with document complexities. Reduce turnaround times and improve document processing accuracy for a faster and more efficient claims processing and underwriting Unstract automation use cases for insurance Claims processing → Insurance underwriting → Insurance Triaging → KYC processing and onboarding → TrustWe bring trust to LLM responses NULL is better than wrong: Unstract’s LLMChallenge uses two separate LLMs to extract and challenge, always giving you the right value or no value at all. Say goodbye to hallucinations: Since LLMChallenge uses two LLMs to arrive at a consensus before an extracted field value is returned, hallucinations are caught and discarded early in the process. Learn More EFFICIENCYGo big on scale— and small on bills Reduce token usage by up to 7x—powered by LLMs! SinglePass Extraction: Read all your field extraction prompts to construct a large, single prompt. Summarized Extraction: Automatically constructs an extremely compact version of the input document. Learn More EFFICIENCYGo big on scale— and small on bills Reduce token usage by up to 7x—powered by LLMs! SinglePass Extraction: Read all your field extraction prompts to construct a large, single prompt. Summarized Extraction: Automatically constructs an extremely compact version of the input document Learn More EFFICIENCYGo big on scale— and small on bills Reduce token usage by up to 7x—powered by LLMs! SinglePass Extraction: Read all your field extraction prompts to construct a large, single prompt. Summarized Extraction: Automatically constructs an extremely compact version of the input document Begin your AI driven document processing automation today Start for free Schedule a demo FlexibilityYou’re in full control — flexibility max Choose the best LLM, Vector DB, Embedding Model and Text Extraction service based on your needs. AGPL 3.0 LICENSEOpen-Source Unstract is an open-source, no-code platform that lets you automate document processing workflows at any scale. Unstract leverages cutting-edge AI to surpass the current capabilities of IDP Intelligent Document Processingand RPA Robotic Process Automation. Quick Start Join us on Slack Welcome to Prompt Studio The prompt engineering environment purpose-built forstructured document data extraction. Build generic prompts at speed Prompt Studio is an environment designed for prompt engineers to create generic prompts quickly from a small sample of representative documents. Learn more Versioning built-in Stop maintaining prompts in spreadsheets. Test new versions of your prompts thoroughly. Rollback easily should you spot a problem. Learn more Multi-LLM support View and compare responses from and the cost of multiple LLMs side-by-side. Learn more Know the cost, comparatively As you build. keep an eye on how much your extraction is costing you side-by-side comparison for your chosen LLMs. Learn more LLMWhisperer: Get complex documents ready for LLM consumption LLM output is as good as the input you provide it. The perfect companion service to LLMs, it produces highly optimized output from input documents in a way LLMs are best able to understand. A unique layout-preserving mode lets LLMs understand multi-column layouts, forms and tables. State-of-the-art handwritten text detection means you can process challenging documents with ease. Checkbox and radio button detection means you can process forms easily. Can deal with scanned PDFs and smartphone camera
-captured documents with high fidelity. We help fit unstructured documents into your workflows APIs Call APIs can structure unstructureddocuments from your existing apps. ETL Pipelines Have unstructured documents in cloud filestorage? Structure them and push to datawarehouses and databases. Prefer MCP? We got you covered Unstract MCP Server Get a standard, structured JSON back
irrespective of the variants. Learn more LLMWhisperer MCP Server Prepare documents for easy consumption by agents Learn more Secure and Compliant, Always Unstract adheres to the strict rules and regulations of various compliance authorities.
Rest assured, we have policies, systems, and processes to ensure that your data is always safe, secure, and private. Manual processes belong to a pre LLM era. Welcome to the future Schedule a Demo Start for free We use cookies to enhance your browsing experience. By clicking "Accept", you consent to our use of cookies. Read More. Decline Cookie Settings Accept For more information on how Google's third party cookies operate and handle your data, see: Google's Privacy Policy Necessary Always Active Necessary cookies help make a website usable by enabling basic functions like page navigation and access to secure areas of the website. The website cannot function properly without these cookies. Marketing Marketing Marketing cookies are used to track visitors across websites. The intention is to display ads that are relevant and engaging for the individual user and thereby more valuable for publishers and third party advertisers. Analytics Analytics Analytics cookies help website owners to understand how visitors interact with websites by collecting and reporting information anonymously. Preferences Preferences Preference cookies enable a website to remember information that changes the way the website behaves or looks, like your preferred language or the region that you are in. Unclassified Unclassified Unclassified cookies are cookies that we are in the process of classifying, together with the providers of individual cookies. Save And Accept Cookie Settings --- Simple pricing for all your complex document processing needs Need help choosing the right plan? Talk to us Unstract CloudLLMWhisperer Get 2 Months Free With Annual Billing! Monthly Annual COMPARE PLANS STARTER $499/month Billed monthly $416/month Billed annually Start 14-day free trial GROWTH $2249/month Billed monthly $1874/month Billed annually Start 14-day free trial COMPARE PLANS STARTER $499/month Billed monthly $416/month Billed annually Start 14-day free trial Number of pages/month Number of pages/Year 5,000 60,000 GROWTH $2249/month Billed monthly $1874/month Billed annually Start 14-day free trial Number of pages/month Number of pages/Year 25,000 300,000 Number of pages/month Number of pages/year 5,000 60,000 25,000 300,000 Overage/page $0.1 $0.09 Includes LLMWhisperer. You bring your keys for: LLMs, vector databases, and embedding models. Unstract Enterprise On-Premise Our Enterprise Plan is tailored to meet your unique business needs and for those seeking to self-host Unstract on-premise. You can leverage the power of Unstract while maintaining full ownership of your data, infrastructure, and security. Request pricing quote Claim $10 free credit! To help you get started, we’re offering a $10 free credit on Azure OpenAI GPT-4o and free access to Postgres, Azure OpenAI Embedding, and LLMWhisperer text extractor Start 14-day free trial Simple pay-as-you-go pricing.No flat fee or upfront commitment needed. Start free: Process up to 100 pages daily at no cost. No credit card required! Monthly Annual PROCESSING MODES NATIVE TEXT $199/month $1/1000 pages Start free Cost per page $0.0500 Cost per page above quota $0.0525 Supported document types PDFs (not scanned) PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress Recommended use cases Low latency requirement All documents are PDFs PDFs are native text PDFs Cost sensitive application LOW COST $5/1000 pages $5/1000 pages Start free Cost per page $0.0475 Cost per page above quota $0.0500 Supported document types PDFs PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress Recommended use cases High quality scanned PDFs High quality scanned images HIGH QUALITY $7/1000 pages $10/1000 pages Start free Cost per page $0.0450 Cost per page above quota $0.0475 Supported document types PDFs PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress Recommended use cases Medium/low quality scanned PDFs Medium/low quality scanned images Handwritten documents HIGH QUALITY WITH FORM ELEMENTS $15/1000 pages $15/1000 pages Start free Cost per page $0.0425 Cost per page above quota $0.0450 Supported document types PDFs PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress Recommended use cases Checkbox and radio button detection Medium/low quality scanned PDFs Medium/low quality scanned images Handwritten documents Supported document types PDFs (not scanned) PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress PDFs PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress PDFs PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress PDFs PDF forms Images (JPEG, PNG, TIFF) MS Office Document MS Office Excel MS Office Powerpoint LibreOffice Writer LibreOffice Calc LibreOffice Impress Recommended use cases Low latency requirement All documents are PDFs PDFs are native text PDFs Cost sensitive application High quality scanned PDFs High quality scanned images Medium/low quality scanned PDFs Medium/low quality scanned images Handwritten documents Checkbox and radio button detection Medium/low quality scanned PDFs Medium/low quality scanned images Handwritten documents ADD-ON Source Document Highlighting API $0.01/ line coordinate data lookup With a little help from an LLM, you can implement source document highlighting with LLMWhisperer.
Generally implemented as a side-by-side view with the source document and the extracted data, this is incredibly useful to speed up manual reviews. COMPARE PROCESSING MODES NATIVE TEXT LOW COST HIGH QUALITY HIGH QUALITY WITH FORM ELEMENTS Checkbox and Radio button detection Line reproduction in output Supported Languages All(Unicode) 120+
English default. Reach out to us to enable more languages 300+ 300+ Image preprocessing (median filter and gaussian blur) Auto Auto Line splitting strategy choice Layout preserving output Extraction performance Very fast Fast Medium Medium Handwriting recognition Basic
support AI/ML based enhancement Rotation and skew compensation Auto repair PDFs Dense textcontent Best
Performance Very good Very good Very good High entropy content (each page contains large variery of text sizes) Best
Performance Very good Very good Very good Responsive Comparison Table Native Text Low Cost High Quality High QualityWITH FORM ELEMENTS Checkbox and Radio button detection Line reproduction in output Supported Languages All(Unicode) 120+English default.Reach out to us to enable more languages 300+ 300+ Image preprocessing(median filter and gaussian blur) Line splitting strategy choice Layout preserving output Extraction performance Very fast Fast Medium Medium Handwriting recognition Basic support AI/ML based enhancement Rotation and skew compensation Auto repair PDFs Dense text content Best Performance Very good Very good Very good High entropy content (each page contains large variety of text sizes) Best Performance Very good Very good Very good Native Text Low Cost High Quality High QualityWITH FORM ELEMENTS Checkbox and Radio button detection Line reproduction in output Supported Languages All(Unicode) 120+English default.Reach out to us to enable more languages 300+ 300+ Image preprocessing(median filter and gaussian blur) Line splitting strategy choice Layout preserving output Extraction performance Very fast Fast Medium Medium Handwriting recognition Basic support AI/ML based enhancement Rotation and skew compensation Auto repair PDFs Dense text content Best Performance Very good Very good Very good High entropy content (each page contains large variety of text sizes) Best Performance Very good Very good Very good LLMWhisperer Enterprise On-Premise Our Enterprise Plan is tailored to meet your unique business needs and for those seeking to self-host LLMwhisperer on-premise. You can leverage the power of LLMWhisperer while maintaining full ownership of your data, infrastructure, and security. Request pricing quote Free tier: 100 pages/day To help you get started, we offer a free tier that lets you process up to 100 pages daily at no cost. No credit card is required to get started! Try forever free Secure and Compliant, Always Unstract adheres to the strict rules and regulations of various compliance authorities.
Rest assured, we have policies, systems, and processes to ensure that your data is always safe, secure, and private We use cookies to enhance your browsing experience. By clicking "Accept", you consent to our use of cookies. Read More. Decline Cookie Settings Accept For more information on how Google's third party cookies operate and handle your data, see: Google's Privacy Policy Necessary Always Active Necessary cookies help make a website usable by enabling basic functions like page navigation and access to secure areas of the website. The website cannot function properly without these cookies. Marketing Marketing Marketing cookies are used to track visitors across websites. The intention is to display ads that are relevant and engaging for the individual user and thereby more valuable for publishers and third party advertisers. Analytics Analytics Analytics cookies help website owners to understand how visitors interact with websites by collecting and reporting information anonymously. Preferences Preferences Preference cookies enable a website to remember information that changes the way the website behaves or looks, like your preferred language or the region that you are in. Unclassified Unclassified Unclassified cookies are cookies that we are in the process of classifying, together with the providers of individual cookies. Save And Accept Cookie Settings --- Turn any document into structured data in minutes Turn any document into structured data in minutes Extract structured data from any document using Natural Language, with unparalleled accuracy and cost-efficiency. Try Demo Playground Start for Free Watch 3-min demo Trusted by organizations that demand precision and scale Your IDE can’t handle professional prompt engineering The reality today Writing prompts with standard IDE is slowand fragmented. You're constantly switching between sample documents, prompt text files, and terminals. Testing against document variations is done manually, and enforcing a consistent output schema means writing and maintaining complex logic. On-demand intelligence Instantly extract the data you need fromany document. Iterate quickly with prompts, documents, and outputs all in one view. Test prompts against documents simultaneously. Enforce any output type with a simple toggle. Deploy projects as a production-ready API or ETL pipeline. A purpose-built prompt engineering environment Unstract transforms every document into structured data that flows through your infrastructure. Deploy outputs the way you want: lightweight APIs or enterprise ETL. Full Page Scroll with Image Switching Build and test prompts within a single canvas Write prompts for any use case—from line items on an invoice to clauses in a legal contract—and see instant results. And refine them on the fly without writing a single line of code. Try Free for 14 Days Spot and discard hallucinations quickly Run prompts through 2 different models at the same time with LLMChallenge and get an output only if both agree. Eliminate hallucinations and bring the highest level of trust for outputs. Book a Demo Control LLM Costs with token-saving toggles Prompt Studio helps you keep LLM costs down with SinglePass and Summarized extraction. Both modes let you balance speed and cost to reduce token usage by up to 7x. See Documentation Enforce consistent schema for any input Obtain the exact data structure you need—from simple text to nested JSON—and transform arbitrary inputs from any document into clean, reliable, and system-ready data. Talk to Us View the exact source of extracted results Instantly verify outputs with Source Document Highlighting to make reviews faster, ensure a completely auditable trail and build absolute trust in your automated workflows. Start for Free Ensure consistent and reliable extraction every time View how prompts perform across documents with Prompt Coverage and quickly optimize underperforming ones to simplify cross-document verification and quality assurance processes. Get in Touch Build and test prompts within a single canvas Write prompts for any use case—from line items on an invoice to clauses in a legal contract—and see instant results. And refine them on the fly without writing asingle line of code. Try Free for 14 Days Spot and discard hallucinations quickly Run prompts through 2 different models at the same time with LLMChallenge and get an output only if both agree.Eliminate hallucinations and bring the highest levelof trust for outputs. Book a Demo Control LLM Costs with token-saving toggles Prompt Studio helps you keep LLM costs down with SinglePass and Summarized extraction. Both modes let
you balance speed and cost to reduce token usage byup to 7x. See Documentation Enforce consistent schema for any input Obtain the exact data structure you need—from simple text to nested JSON—and transform arbitrary inputs from from any document into clean, reliable, and system ready data. Talk to Us View the exact source of extracted results Instantly verify outputs with Source Document Highlighting to make reviews faster, ensure a completely auditable trail and build absolute trust in your automatedworkflows. Start for Free Ensure consistent and 
reliable extraction every time View how prompts perform across documents with Prompt Coverage and quickly optimize underperforming ones to simplify cross-document verification and quality assurance processes. Get in Touch Get unprecedented control Go beyond basic prompting with a full suite of advanced controls built for accuracy and workflow efficiency. Output Analyzer Compare source documents to the extracted data and make changes on the go. Preamble and Postamble Use preambles to guide the LLM’s approach before each prompt, and Postambles to format the final output. Combined Output Consolidate the results from all your prompts into a single, structured JSON object. Clone, Share, 
& Export Start new projects quickly or share it with your team in read-only mode, even if they don’t have an Unstract account. Users Unstract Lenny Hartman CTO, Tokenstreet Unstract lets us turn a wide range of document formats into clean, structured data with low integration effort and without compromising on enterprise-grade controls. Its high extraction accuracy paired with clear highlight-based validation speeds up our process a lot! Connect the tools you already use and trust See All Integrations Results you can count on 20x Improved Operational Efficiency 80% Lesser Human Touchpoints 99% Accurate Document Extraction Enterprise-grade by design Ready to transform
your document workflows? Schedule a Demo We use cookies to enhance your browsing experience. By clicking "Accept", you consent to our use of cookies. Read More. Decline Cookie Settings Accept For more information on how Google's third party cookies operate and handle your data, see: Google's Privacy Policy Necessary Always Active Necessary cookies help make a website usable by enabling basic functions like page navigation and access to secure areas of the website. The website cannot function properly without these cookies. Marketing Marketing Marketing cookies are used to track visitors across websites. The intention is to display ads that are relevant and engaging for the individual user and thereby more valuable for publishers and third party advertisers. Analytics Analytics Analytics cookies help website owners to understand how visitors interact with websites by collecting and reporting information anonymously. Preferences Preferences Preference cookies enable a website to remember information that changes the way the website behaves or looks, like your preferred language or the region that you are in. Unclassified Unclassified Unclassified cookies are cookies that we are in the process of classifying, together with the providers of individual cookies. Save And Accept Cookie Settings --- Don’t let LLM hallucinations corrupt your data Don’t let LLM hallucinations corrupt your data A dual-LLM consensus engine ensures only accurate data makes it into your API, ETL and HITL workflows. Try Free for 14 Days Schedule a Demo See it in action Trusted by Add a trust layer to AI document data extraction Full Page Scroll with Image Switching Verify outputs with two LLMs Two LLMs run your prompts in parallel: an extractor and a challenger. Get an output only if both agree—NULL if not. Configure any combination: OpenAI + Claude, Azure GPT + Vertex + more! Start For Free Track, debug & optimize with extraction metadata Get drilled-down insights by running the metadata for every extraction. Track token costs, view challenger confidence scores, or debug outputs after deployment. See Documentation Verify outputs with
two LLMs Two LLMs run your prompts in parallel: an extractor and a challenger. Get an output only if both agree—NULL if not. Configure any combination: OpenAI + Claude, Azure GPT + Vertex + more! Start For Free Track, debug & optimize with extraction metadata Get drilled-down insights by running the metadata for every extraction. Track token costs, view challenger confidence scores, or debug outputs after deployment. See Documentation Supports the 
models and adapters of your choice Loved by users Lenny Hartman CTO, Tokenstreet Unstract lets us turn a wide range of document formats into clean, structured data with low integration effort and without compromising on enterprise-grade controls. Its high extraction accuracy paired with clear highlight-based validation speeds up our process a lot! Results you can count on Production-Grade
Data Integrity Improved Automation Scaling Reduced Manual Oversight FAQs Which LLM combinations work best? We recommend pairing LLMs from different providers. Popular combinations: OpenAI GPT-4 + Google Gemini Pro, Anthropic Claude + Cohere, OpenAI + Anthropic. What happens when LLMs disagree? The field returns NULL. How long does consensus take? Typically adds 2-5 seconds to extraction time. For financial and legal documents, accuracy trumps speed. Can I use the same LLM model for both extraction and challenging? Technically yes, but you shouldn’t! Using the same model (or even models from the same provider) defeats the purpose because they tend to make similar mistakes. Provider diversity is key to catching different error patterns. What’s the difference between a NULL and an empty field? Important distinction: NULL means “we couldn’t reach consensus.” Empty means “we agreed this field has no value.” Can I see the challenge metadata for debugging? Yes. The full conversation log is available via API for every extraction. You can see why the LLMs disagreed and the confidence score given by the challenger LLM. Enterprise-grade by design Want to see
LLMChallenge in action? Get a Demo We use cookies to enhance your browsing experience. By clicking "Accept", you consent to our use of cookies. Read More. Decline Cookie Settings Accept For more information on how Google's third party cookies operate and handle your data, see: Google's Privacy Policy Necessary Always Active Necessary cookies help make a website usable by enabling basic functions like page navigation and access to secure areas of the website. The website cannot function properly without these cookies. Marketing Marketing Marketing cookies are used to track visitors across websites. The intention is to display ads that are relevant and engaging for the individual user and thereby more valuable for publishers and third party advertisers. Analytics Analytics Analytics cookies help website owners to understand how visitors interact with websites by collecting and reporting information anonymously. Preferences Preferences Preference cookies enable a website to remember information that changes the way the website behaves or looks, like your preferred language or the region that you are in. Unclassified Unclassified Unclassified cookies are cookies that we are in the process of classifying, together with the providers of individual cookies. Save And Accept Cookie Settings