The problem every doctor knows
Every clinician has been there. You remember reading something relevant — a recommendation in a guideline, a finding in a study, a figure in a textbook chapter — but you cannot remember exactly where. So you search. You flip through PDFs. You open tabs. You scan pages. Twenty minutes later, you either find it or give up.
Medical knowledge is growing faster than anyone can keep up with. Guidelines are updated yearly. New evidence is published daily. And the documents that inform clinical decisions — the ones you actually trust — sit scattered across your hard drive, your institution's shared folders, and your email attachments.
Why general AI tools fall short
Consider a familiar scenario: you are preparing a presentation or a set of slides for a teaching session. You have ten PDF windows open — guidelines, journal articles, textbook chapters — and you are trying to find the right figure, the right recommendation, the right data point. You know you read it somewhere. But where? After toggling between windows for fifteen minutes, the context blurs. You lose track of which document said what. The task that should take minutes stretches into an hour.
Modern AI tools like ChatGPT and Gemini have started adding citations to their responses. On the surface, this looks like progress. But look closely at what they actually cite.
Analyses of AI chatbot citations reveal a striking pattern: the majority of sources come from health media websites, commercial pages, and institutional blogs — not from peer-reviewed research. Nearly a third of citations in health-related queries point to health media sites like hospital blogs and wellness portals. Another quarter come from commercial and affiliate-driven websites. Actual academic and research sources make up less than a quarter of all citations.
Put differently, when you ask a public AI tool a medical question, it relies more on summaries and interpreters of science than on the science itself. The sources it finds are overwhelmingly open-access web pages — the kind of content that ranks well in search engines, not the kind that survives peer review. The single largest academic source, PubMed Central, accounts for a fraction of one percent of all citations.
This means the AI is citing blog posts written about a study, not the study itself. It is citing a hospital's patient-facing summary, not the guideline the summary was based on. It is citing an abstract, not the full-text article with the methods, results, and nuance that matter.
For a clinician preparing a presentation, writing a report, or making a treatment recommendation, this is not good enough. You need the original source — the full-text journal article, the complete guideline chapter, the peer-reviewed material that your colleagues and trainees can verify. An answer built on blog posts and commercial websites is not evidence-based medicine.
These tools also sometimes fabricate references entirely — a phenomenon known as hallucination. In medicine, a hallucinated citation is not just unhelpful. It is dangerous. And even when the citation is real, it often points to a resource you cannot access, verify, or trace back to the specific page and passage that supports the claim.
Medevidex works differently because it answers from your own full-text documents — the peer-reviewed journal articles, the complete guideline chapters, the textbook sections you actually trust. When it cites a source, it sends you back to the exact page in your own PDF. No guessing, no hunting, no ten-window shuffle.
What Medevidex does differently
Medevidex takes a different approach. Instead of answering from general knowledge, it answers from your documents — the PDFs you upload, the guidelines you follow, the literature you trust.
When you ask a question, Medevidex searches through your uploaded documents, finds the most relevant passages, and generates an answer that cites the exact source, page number, and passage. Every answer points back to something you can verify.
This is what we mean by "grounded in your literature." The AI does not make things up. It reads what you have given it and tells you what it found.
Not another document chatbot
There are already tools that let you upload documents and ask questions about them. Some are built by the largest technology companies in the world. So why build another one?
Because those tools are designed for general use — summarising meeting notes, analysing business reports, researching market trends. They work well enough for that. But medical literature is not a business report.
A clinical guideline has figures that contain treatment algorithms. A journal article has tables that summarise trial outcomes. A textbook chapter has diagnostic images that are essential to understanding the text around them. General-purpose document tools either ignore these entirely or treat them as decorative elements to be skipped.
Medevidex was built specifically for medical documents. Its ingestion pipeline understands that a figure labelled "Management of T2 Bladder Cancer" is not clip art — it is the clinical decision pathway. It extracts, indexes, and cites figures and tables alongside text, because in medicine, they are often where the actual answer lives.
General document tools give you a summary. Medevidex gives you a cited answer with the page number, the exact passage, and the figure — because in clinical practice, "I read it somewhere" is not enough.
There is also the question of trust. When you upload clinical guidelines and patient-adjacent literature to a general-purpose platform operated by a large technology company, your documents become part of their ecosystem — subject to their terms of service, their data practices, and their business model. You may not know where your data goes, how long it is retained, or whether it is used to improve their products.
Medevidex is purpose-built, clinician-operated, and architecturally isolated. Your documents stay yours. The tool serves one audience — healthcare professionals — and it is designed around the way they actually work.
Why scoping matters for continuing medical education
In continuing medical education, clinicians are expected to stay current with the latest evidence in their field. But "latest evidence" is not a single monolithic body of knowledge — it is specialty-specific, context-dependent, and often contradictory across different guidelines.
This is where scoping becomes critical. Medevidex allows you to organise your documents into collections and scope your queries to specific collections. Why does this matter?
When you are preparing for a urology exam, you do not want answers contaminated by cardiology guidelines. When you are reviewing paediatric protocols, you do not want adult dosing recommendations mixed in. When you are comparing two competing guidelines on the same topic, you want to query each one separately and see how they differ.
Scoping is not just a convenience feature — it is an intellectual discipline. It forces the AI to answer from a defined evidence base, which mirrors how clinicians actually think: within the boundaries of a specific specialty, a specific guideline, or a specific clinical question.
Without scoping, AI tools give you a blended answer from everything they know. With scoping, you get an answer grounded in exactly the evidence you chose to consult.
Privacy as a design principle
Many clinicians are understandably cautious about uploading medical literature to online platforms. Medevidex was built with this concern at its core.
Every document you upload is stored in a private, isolated environment. No other user can see your documents. No staff member can access them. Your content is never used to train AI models, never shared with third parties, and permanently deleted when you choose to remove it.
This is not a policy — it is how the system is built. Privacy is architectural, not aspirational.
All your documents, one place
Most clinicians have medical documents scattered everywhere — some on a work laptop, others on a home desktop, a few in Google Drive, and more buried in email attachments. When you need to find something, you are searching across three devices and four cloud services before you even start reading.
Medevidex gives you a single, organised home for your medical literature. Upload your PDFs once, arrange them into folders and collections by specialty, topic, or purpose, and access them from any device with a browser.
Your exam preparation materials in one collection. Your department's clinical protocols in another. The latest guidelines from your specialty society in a third. Everything searchable, everything organised, and everything available whenever you need it — whether you are at your desk, in clinic, or on the ward.
No more hunting through folders. No more emailing documents to yourself. No more wondering which version of a guideline is the latest. One library, always up to date, always accessible.
What it took to build this
Medevidex started as a personal project — a clinician's frustration turned into a weekend experiment that grew into something real.
The challenge was not just building an AI that could answer questions. It was building one that could read medical documents the way a clinician does: understanding that a table on page 12 matters as much as the paragraph above it, that a clinical algorithm in a figure is not decorative but essential, and that a citation without a page number is not a citation at all.
Getting figures, tables, and clinical images indexed alongside text was one of the hardest problems to solve. Most document processing tools treat images as decoration. In medical literature, they are often the most important part of the page.
The result is a system that ingests text, figures, and tables from your PDFs and makes all of it searchable — so when you ask about a staging system or a treatment algorithm, the answer can include the actual figure from your guideline.
Who this is for
Medevidex is for any healthcare professional who works with medical literature:
- —Clinicians preparing for exams or staying current with guidelines
- —Researchers reviewing literature for systematic reviews
- —Medical educators building teaching materials from primary sources
- —Trainees learning to navigate evidence-based practice
If you read medical PDFs as part of your work, Medevidex can help you read them faster and more effectively.
Try it
Medevidex is free to start. Upload your first PDF, ask a question, and see the difference that grounded, cited answers make.