AI knowledge brief

Hours of sources.
One expert brief.

Drop in PDFs, lectures, images, notes, YouTube. Edusion keeps the knowledge. Throws away the rest.

Create Knowledge Brief⏎
No account for the first briefSources never leave your session
Session · CRISPR delivery vectors4 sources

Core takeaways

Delivery is the bottleneck, not the nuclease.

  • AAV fails on cargo size and pre-existing immunity.
  • LNPs default to liver via ApoE corona.

Protein corona

The layer of serum proteins that adsorbs to a nanoparticle in circulation. ApoE adsorption routes LNPs to hepatocytes via LDL receptor, which is why liver is the default destination.

Final cheat sheet

Delivery, not editing, is the rate-limiting step in vivo.

Proof of compression

Everything in. Only what matters out.

A 68-page review, a 102-minute lecture, a photographed page of notes and a pasted readout. One brief you can hold in your head.

1:42:09
compressing

Knowledge brief

CRISPR delivery vectors

Core takeaways

Delivery is the bottleneck, not the nuclease.

Three vector families, three distinct failure modes.

Transient payloads beat persistent ones on safety.

Key concepts

Tropism · tissue preference set by capsid residues or lipid chemistry. Packaging limit · ~4.7 kb for AAV.

Final cheat sheet

Choose the cell type first. Everything else is a consequence.

How it works

Gather. Compress. Use.

Three steps. No configuration, no prompt engineering, no model picker.

01 / Gather

Drop the raw material

PDFs, DOCX, screenshots, photographed pages, pasted notes, a YouTube URL. Mixed formats in one session is the normal case.

Drop sources or paste a YouTube URL
Doudna_2024.pdfLecture 02

02 / Compress

Extract, then discard

Text, structure and timestamps are pulled from every source. Repetition, filler and slide furniture are dropped.

Doudna_2024.pdf
Lecture 02
LNP_notes.jpg

03 / Use

Then the tools unlock

The brief arrives first. Flashcards, quiz, report, mind map and chat build on it and re-read the original sources.

Core takeaways

Delivery is the bottleneck, not the nuclease.

FlashcardsQuizReportMind MapChat
The brief

Written like an expert wrote it for themselves.

Core takeaways, key concepts, hard facts, process, relationships, and a cheat sheet at the end. Nothing that was not in your sources.

grounded in sourcesno invented examples
Brief · 4 sources

CRISPR delivery vectors: why editing works and delivery decides

4 sources · 142 pages equivalent · generated in 38s

Core takeaways

  • Delivery is the bottleneck, not the nuclease. Editing efficiency in vitro routinely exceeds 80%. In vivo outcomes collapse to single digits outside liver and eye. The limiting variable is cargo arriving intact in the right cell type, not Cas9 activity.
  • Three vector families, three failure modes. AAV fails on cargo size and pre-existing immunity. Lipid nanoparticles fail on extrahepatic targeting. Virus-like particles and electroporation fail on scale and manufacturing consistency.
  • Transient beats persistent for safety. Ribonucleoprotein and mRNA payloads clear within days, which shortens off-target exposure. AAV-delivered Cas9 can persist for months, and persistence is now treated as a liability rather than a feature.

Key concepts

Tropism
The tissue preference of a vector, set mostly by capsid surface residues in AAV and by ionizable lipid chemistry plus protein corona in LNPs.
Packaging limit
AAV carries ~4.7 kb of single-stranded DNA. SpCas9 plus promoter and guide occupies nearly all of it, which forces dual-vector splits or smaller nucleases.
Protein corona
The layer of serum proteins that adsorbs to a nanoparticle in circulation. ApoE adsorption routes LNPs to hepatocytes via LDL receptor, which is why liver is the default destination.
RNP delivery
Pre-assembled Cas9 protein plus guide RNA. No transcription or translation step, activity peaks within hours, clearance within roughly 72 hours.

Facts worth memorising

  • AAV packaging capacity: ~4.7 kb single-stranded. SaCas9 (3.2 kb) fits with a guide; SpCas9 (4.1 kb) does not fit with regulatory elements.
  • LNP biodistribution without targeting ligands: >90% of dose to liver within 30 minutes of IV administration.
  • Neutralizing antibody prevalence against AAV2 in adults: roughly 30–60%; AAV5 is lower, which is why AAV5 appears in systemic programmes.
  • Approved editing therapies to date use ex vivo electroporation of RNP, sidestepping in vivo delivery entirely. Trial readout Q3

Process: selecting a vector

  • Define the target cell type first. Vector selection is downstream of that decision, never upstream.
  • Check cargo size against packaging limit. Over ~4.5 kb, AAV requires a dual-vector split or a compact nuclease.
  • Screen for pre-existing neutralizing antibodies if the route is systemic AAV. Titre above threshold excludes the patient.
  • Select payload format: RNP for shortest exposure, mRNA for moderate, DNA for sustained expression where persistence is acceptable.

Final cheat sheet

  1. Delivery, not editing, is the rate-limiting step in vivo.
  2. AAV: high transduction, 4.7 kb ceiling, immunity problem, effectively one dose.
  3. LNP: redosable, easy manufacturing, liver by default, extrahepatic targeting unsolved.
  4. RNP ex vivo: highest control, works today, does not scale to solid tissue.
  5. Choose the cell type first. Everything else is a consequence.
After the brief

Five ways to put it to work.

Flashcards

3 / 12

Question on one side, the source's own answer on the other. Click to flip.

Quiz

Checks whether you can use the facts, not repeat them.

An untargeted LNP is dosed IV. Where does most of it go?

Report

The long version, when you need to hand it to someone else.

Mind Map

The shape of the argument, not a bubble diagram. Hover a node to trace its edges.

CRISPR delivery vectorsAAV4.7 kb ceilingPre-existing immunityOne-shot onlyLipid nanoparticlesApoE corona → liverRedosableRNP ex vivo72 h clearanceApproved therapies

Chat

Answers from your sources. It says so when the answer isn't there.

Does the brief overstate how bad extrahepatic LNP targeting is?
No, the sources agree. Lecture 02 puts untargeted biodistribution at over 90% liver in 30 minutes. The notes flag lipid tuning for lung and spleen as early animal work, not solved.Lecture 02 · 14:32

Every tool reads the original sources, not just the summary. Nothing important is lost in compression.

PDF.pdfWord documents.docxPlain text.txtScreenshots.pngPhotographed pages.jpgYouTube lecturesurlPasted notes⌘VPDF.pdfWord documents.docxPlain text.txtScreenshots.pngPhotographed pages.jpgYouTube lecturesurlPasted notes⌘V
Use cases

Three situations, one shape of problem.

Students

Exam in 48 hours. 200 pages of slides.
14 lecture decksone brief + 40 cards

Operators

Three PDFs and a call transcript before the meeting.
90 min of reading6 min and a position

Researchers

A paper, a lecture, a dataset note. One map of the argument.
3 disconnected sourcesone mind map
About Edusion

Making Study Material Easier to Learn

Edusion is an AI-powered study assistant built to help university students turn scattered learning materials into organized, easy-to-revise study resources.

Students can provide PDFs, notes, research papers, presentations, or YouTube lectures. Edusion processes the content and generates summaries, key points, flashcards, quizzes, and exam-oriented notes.

Our Goal

More time for understanding.

Our goal is to reduce the time students spend organizing and revising study material, allowing them to focus more on understanding and learning.

Project Team

Emad KhanRoll No. 202403040
Shreyas KharatRoll No. 202403041

Technology & Approach

A unified study workflow.

Edusion combines modern web technologies, document processing, speech-to-text capabilities, and AI models. It is designed as a complete web application, with AI serving as one component of the overall system rather than the entire application.

Bring the sources.
Leave with the knowledge.

Four files and a lecture link is enough. The brief takes about forty seconds.