Skimle is an AI-powered qualitative data analysis platform that structures unstructured documents (interviews, data room files, customer comments, public consultation statements, legal documents and so on) into organised, analysable datasets. Unlike chatbots that retrieve random passages, Skimle systematically processes your entire document set, analysing each section, extracting insights, and organising them into hierarchical theme taxonomies with full traceability. See how Skimle works for a full overview.
Skimle is built for knowledge workers who analyse qualitative data: academic researchers conducting thematic analysis, management consultants synthesising expert interviews, policy analysts processing public consultations, market researchers analysing customer feedback, HR teams running engagement surveys, and anyone who needs systematic, defensible insights from unstructured data.
Most simple AI analysis tools use RAG (Retrieval-Augmented Generation) to retrieve relevant passages at query time, meaning you get different answers each time you ask the same question. Skimle processes your entire dataset systematically upfront, creating a stable structure where every theme links to verified quotes. You get comprehensive coverage, consistent results, and full traceability instead of black-box answers. Read more about why RAG-based tools fall short for serious analysis.
No. Skimle is designed for researchers and analysts, not programmers. The interface is intuitive (think spreadsheet + document viewer), and all AI processing happens automatically. If you can use Google Docs and Excel, you can use Skimle.
Most users are productive within 15–30 minutes. Upload documents, review AI-generated themes, refine categories — the workflow is straightforward. Unlike traditional qualitative analysis software (NVivo, ATLAS.ti) which can take weeks to master, Skimle's interface is designed for immediate productivity.
Skimle accepts PDF, .docx, TXT, RTF, audio files (MP3, M4A, WAV), and video files (MP4, MOV). For audio and video, Skimle transcribes automatically before analysis begins, keeping your entire workflow in one place. You can mix formats in a single project.
Yes. Skimle's built-in transcription handles MP3, M4A, WAV, MP4, and MOV files. Upload your recording and Skimle transcribes it before analysis begins. The transcript is stored alongside the source file so you can verify quotes against the original. No separate transcription service needed.
Up to 1,000 documents per project. Most use cases involve 10–200 documents (typical research studies, consulting projects, policy consultations), but Skimle scales to handle large document sets like data room reviews or comprehensive literature analyses.
100+ languages. Upload documents in any language — Skimle analyses them in the original language while creating unified theme structures. Particularly useful for EU-wide consultations, global market research, or multilingual academic studies. No translation required.
Yes. Upload English interviews, German reports, and Spanish survey responses in the same project. Skimle creates cross-language theme categories while preserving original-language quotes for verification.
The Skimle table is an interactive spreadsheet where each row represents a document (e.g., one interview) and each column represents a theme category (e.g., “pricing concerns”). Cells contain extracted insights with direct links to source quotes. This structure makes it easy to see patterns across documents, compare segments, and drill into specific topics.
Absolutely. Skimle's AI suggests initial categories based on your data, but you have full control to merge, split, rename, delete, or create categories from scratch. You can also provide your own category structure upfront (e.g., based on your interview guide or theoretical framework) and have Skimle extract insights accordingly.
Two-way transparency means you can navigate both directions: (1) click a theme to see all supporting quotes across documents, and (2) click a document to see all themes extracted from it. Every AI decision is traceable — no black boxes. This is what makes Skimle-assisted analysis defensible in academic, commercial, and regulatory contexts.
Skimle's chat interface has awareness of your structured data (themes, categories, metadata). You can ask questions like “What do enterprise customers say about pricing?” or “How do opinions differ by region?” The AI answers using both the structured theme data and original documents — giving you more accurate, contextual responses than generic RAG chatbots.
Skimle Ask is an AI-powered interview tool that lets you collect qualitative data at scale. You define the interview guide and Skimle conducts structured conversations with participants asynchronously. Responses feed directly into the analysis pipeline, so data collection and thematic analysis happen in the same environment — no copy-pasting between tools.
Yes. Skimle exposes a Model Context Protocol (MCP) endpoint at https://skimle.com/api/mcp. Any MCP-compatible tool — Claude Desktop, Cursor, Windsurf, Claude Code, and others — can connect and use the full set of read and write tools: browsing documents, searching insights, reorganising categories, and more. Authentication is via OAuth (no token needed) or an API key generated in your account settings. You choose which projects to expose. Credits are shared with the rest of Skimle — no separate billing.
Yes. Claude Code and Cowork both support MCP, so you can connect them to Skimle using the endpoint https://skimle.com/api/mcp. Once connected, the agent can list your projects, search across documents, explore category hierarchies, create and update insights, and save notes — all from within your coding or writing environment. See the MCP setup page for step-by-step instructions, or read our step-by-step agentic workflow guide.
Word (comprehensive reports with themes, quotes, and summaries), PowerPoint (executive summaries with theme breakdowns), Excel (data tables with coding matrices), and REFI-QDA — the interoperable standard recognised by NVivo, MAXQDA, and ATLAS.ti, so you can continue manual coding in a traditional QDA tool if needed. All exports maintain source attribution and full traceability.
Yes. Skimle Anonymise provides AI-powered pseudonymisation and anonymisation with cross-file consistency, six identifier categories (names, roles, locations, organisations, dates, and other), configurable transformation modes, and a PDF audit report. The anonymised files feed directly into the analysis workflow. This covers GDPR and HIPAA requirements and produces the documentation ethics boards and IRBs expect.
Yes. Skimle is built by academics for academic use. Our co-founder Henri Schildt is a professor with 30+ publications using qualitative methods. Skimle's approach is inspired by established thematic analysis (Braun & Clarke) and grounded theory (Gioia) methodologies — systematic, transparent, and reproducible. See our academic researchers use case and guide to AI in academic qualitative research.
Increasingly yes, with proper disclosure. State your methodology transparently: “AI-assisted qualitative coding using Skimle, with manual validation and refinement.” Skimle 's full audit trail lets you document exactly how themes were derived, which satisfies peer review requirements for methodological rigour. Our guide on AI in academic qualitative research covers how to frame this in methods sections and ethics applications.
We provide suggested citation language upon request.
Yes. Export your coding matrix, have a second coder review a sample of documents manually, and calculate agreement (Cohen's kappa, etc.). Researchers can use Skimle for initial coding (90% of the work), then validate with manual review (10% quality check).
Yes. Skimle supports inductive coding where themes emerge from data rather than being predefined. Start with open coding (Skimle suggests initial categories), then iteratively refine through axial coding (merge/reorganise categories), and develop theoretical frameworks. The process mirrors manual grounded theory but accelerates the mechanical aspects.
Every quote generated by AI is verified against source documents. If the AI creates text that doesn't exist verbatim in your documents, our system flags it and requests re-processing until verified quotes are provided. Additionally, the systematic structure means you can spot-check any theme by reviewing its supporting quotes.
Skimle's systematic processing analyses every section of every document, reducing the risk of missed themes compared to manual coding (where fatigue and bias affect coverage). However, we recommend reviewing AI-generated themes critically. AI suggests, you validate.
Yes. Skimle supports REFI-QDA export, the interoperable standard recognised by NVivo, MAXQDA, and ATLAS.ti. You can start analysis in Skimle and continue manual coding in a traditional QDA tool, or use Skimle to pre-code a dataset before handing it off to a team working in NVivo.
Yes. The free tier lets you analyse up to 400 pages to test the platform. Perfect for trying Skimle with a pilot project before committing to a paid plan. See all plans.
Document analysis consumes credits based on document length. One credit allows you to upload around 2,000 characters of text, roughly one page. Subscriptions include a monthly or annual credit allocation, with options to purchase top-ups if needed.
Pricing starts with a free tier for trial use. See full pricing. Enterprise and institutional licences are available with custom pricing — contact us for details.
Yes. We offer discounted packs for students and academic researchers. Proof of academic status is required. Contact us or check the academic pricing tab on the pricing page.
Yes, anytime from your account settings. When upgrading, you're charged the prorated difference. When downgrading, the change takes effect at your next billing cycle.
You can purchase credit top-ups at any time, or upgrade to a higher tier. Your existing projects remain accessible; you just can't analyse new documents until you add credits.
Yes. We provide site licences for universities, research institutes, consulting firms, and government agencies. Team licences include collaborative features, centralised billing, data processing agreements, and volume discounts. Contact us for institutional pricing.
Yes. All paid plans generate invoices automatically. For institutional purchases requiring purchase orders or specific billing arrangements, contact our sales team.
Yes. Access is authenticated and logged. We follow industry-standard security practices and are happy to complete security questionnaires for enterprise customers.
All data is stored in the European Union (cloud servers in EU regions). Your data never leaves the EU, ensuring GDPR compliance and data sovereignty for European customers.
Yes, fully. We comply with all GDPR requirements: lawful processing, data minimisation, purpose limitation, storage limitation, and data subject rights. We provide Data Processing Agreements (DPAs) for institutional customers. For projects involving participant data, Skimle Anonymise helps you pseudonymise or anonymise before analysis, reducing your compliance exposure further.
No. Your documents and data are never used to train AI models. Your data remains private and is used solely for your analysis purposes.
Yes. You can delete individual projects or your entire account at any time from your settings. Deletion is permanent and immediate. We retain no copies after deletion (except for legal and accounting records such as invoices).
Yes, for enterprise customers with specific security or compliance requirements. We can deploy Skimle in a dedicated cloud environment or discuss on-premises options. Contact us for details.
Researchers can include Skimle in their IRB/ethics applications under “data analysis tools” similarly to how they would mention SPSS or NVivo. We can provide documentation about our data handling, security, and AI processes to support your ethics applications. Skimle Anonymise also produces a signed PDF audit report documenting every transformation applied, which ethics boards increasingly expect as evidence of rigorous anonymisation.
Traditional tools require manual coding of every passage — systematic but time-intensive (weeks to months). Skimle uses AI to automate the coding process while maintaining the same systematic, transparent methodology. You get NVivo-quality rigour at AI speed. Skimle's interface is also more intuitive and cloud-native, and it exports to REFI-QDA so projects can move between tools. See our full comparison of QDA tools.
Dovetail is designed for product teams doing lightweight UX research with AI features added on. Skimle is built from the ground up for rigorous qualitative analysis with academic-grade methodology. If you need to defend your findings in peer review, present to sceptical executives, or handle confidential data with GDPR compliance, Skimle provides the structure and transparency those situations demand. Read the full Skimle vs Dovetail comparison.
You can, but you'll hit limitations quickly: inconsistent answers (ask the same question twice, get different responses), no systematic coverage (might miss insights that don't semantically match your query), no traceability (can't show where conclusions came from), and a limited context window that struggles with 20+ documents. For quick summaries of one or two documents, general AI tools work. For systematic analysis, use Skimle. Read more about where LLM-based analysis works and where it fails.
For small projects (under 10 interviews), manual reading is perfectly fine. But as scale increases: (1) Time explodes — 20 interviews = 40+ hours of manual coding; (2) Consistency drops — fatigue and bias affect later documents differently than early ones; (3) Pattern recognition suffers — hard to hold 50 interviews in your head simultaneously. Skimle maintains quality at scale while freeing your time for actual analysis instead of mechanical coding.