Early access

Learn anything deeply.
Not just faster.

Gregator researches, verifies, and organizes knowledge from dozens of sources into a clear learning path — with sources, confidence levels, and conflicting viewpoints included.

A real course, not a chat reply

Here is an actual chapter from a Gregator course. Each claim links to a source. Each source has a credibility score. You read it like an article, not a conversation.

8 sources 5 peer-reviewed 3 min read

Chapter 2: Lipid Nanoparticle Delivery

Naked mRNA is inherently unstable in biological environments. Ribonucleases in the bloodstream degrade unprotected mRNA within minutes, making direct injection ineffective for most therapeutic applications.[1] The solution is encapsulation in lipid nanoparticles (LNPs) — roughly 80–100nm spheres composed of ionizable lipids, helper lipids, cholesterol, and PEG-lipid conjugates.[2]

How LNPs enter cells

LNPs are taken up by cells through receptor-mediated endocytosis. Once inside the endosome, the ionizable lipid component becomes positively charged in the acidic environment (pH ~5–6), destabilizing the endosomal membrane and releasing the mRNA cargo into the cytoplasm.[1][3] This is where ribosomes can translate the mRNA into the target protein — in the case of COVID-19 vaccines, the SARS-CoV-2 spike protein.

Strong evidence
Endosomal escape mechanism confirmed across 3 independent cryo-EM studies (2019–2021) with consistent results.

Stability and storage challenges

Despite their effectiveness, LNPs introduce practical constraints. The lipid formulation is sensitive to temperature — the BNT162b2 (Pfizer-BioNTech) vaccine originally required storage at -70°C, though reformulation improved this to standard refrigeration for shorter periods.[4]

Mixed evidence on reformulation
Two sources cite improved stability at 2-8°C for 10 weeks; one FDA review notes the data is based on a single lot study.

Pick a topic. Walk away understanding it.

You bring the curiosity. Gregator does the reading. A few minutes later you have a structured course you can open, trust, and keep coming back to.

1

You name what you want to understand

A question, a concept, a whole field. Gregator shapes it into a real learning path — a handful of chapters, each aimed at a piece of the picture worth owning.

2

It finds the people worth listening to

For every chapter Gregator reads the web, the news, and academic databases, keeps only the strongest sources, and tells you why each one is credible — or why it isn't.

3

It writes you chapters you can trust

Every claim is tied to the source it came from. Where serious people disagree, Gregator tells you — it won't paper over the fight with a confident summary.

4

You read it, question it, come back to it

Move through the chapters at your own pace. Click into any claim to see the source. Ask Greg anything — grounded in your course, not a general chatbot. Your understanding compounds.

How does mRNA vaccination work? Researching
Decomposed into 5 chapters 1.2s
Search queries dispatched 0.8s
42 sources gathered and deduplicated 3.4s
Scoring credibility — 18 of 24 done ~20s
Writing chapter narratives

The tool you use after the quick answer

Different tools are good at different things. Gregator is the one you reach for when you want to actually understand — and still have something to come back to tomorrow.

The fast-answer chatbot

A quick reply. You close the tab. In a week you'll only vaguely remember what it said — or whether it made it up.

The AI search summarizer

A cited paragraph that points at sources. Useful, but the real reading still lives somewhere else, and you have to do it yourself.

Trust you can see, learning you can keep

The course is the core. Underneath it, every source is scored, every claim is traceable, and your understanding compounds topic by topic.

Source scoring you can inspect

Lipid Nanoparticle Delivery Systems for mRNA Therapeutics Peer-reviewed
Source type Peer-reviewed journal article — Nature Reviews Drug Discovery, 2021
Evidence Multi-cohort clinical data. 95% confidence intervals reported. 3 independent trials cited.
Signals
Quantitative claims backed Multiple independent citations Single-disease focus
Flags None detected. No promotional language, no conflicts of interest disclosed in the scoring pipeline.

Every source in your course has a panel like this. Source type (peer-reviewed, institutional, journalism, community), evidence quality, positive and negative signals, and any flags the scoring pipeline found — marketing language, missing citations, or single-source claims.

Multi-source, not single-search

Academic databases, the open web, and news are read in parallel for each chapter. Duplicates are dropped and the strongest sources rise to the top before a single word is written.

A knowledge graph that grows with you

Every topic you research connects to the others — prerequisites, contradictions, shared sources. Your understanding isn't a pile of tabs; it's a map you own.

A weekly horizon

Based on what you've been studying, Gregator suggests adjacent topics worth knowing — the things you'd never have thought to search for.

Greg — a companion grounded in your course

Ask anything about the material you've just generated. Greg answers from your sources and your claims, not from generic model memory. If the evidence is thin, Greg says so — and points you to the source that convinced it.

Greg is grounded in
your course, not the internet

Greg is an AI companion loaded with the chapters, sources, and claims from your research — not generic model memory. When you ask something, Greg answers from the material Gregator already vetted for you, and points at the exact source behind every reply.

Scoped to what you researched — if a source is weak or two sources disagree, Greg tells you instead of smoothing it over with a confident-sounding reply.
Push back on your own course — ask Greg to steelman the other side and surface the counterarguments hiding in your sources.
Greg Ch. 2 — Lipid Nanoparticles
The chapter says endosomal escape is well-established. How confident should I be in that?
Fairly confident for the core mechanism. Three independent cryo-EM studies (Hou 2021, Eygeris 2022, Yanez Arteta 2019) show consistent results for ionizable lipid protonation triggering membrane disruption.

The uncertainty is in efficiency — only 2–10% of LNPs successfully escape the endosome. The rest are degraded in lysosomes. This is an active area of research.
Nature Rev. Drug Discovery — peer-reviewed
Ask about this chapter...

Learning in the age of AI is broken

AI makes it easy to get answers. It doesn't make it easy to understand. Gregator is built to close that gap.

For your understanding,
not your attention

No engagement loops, no algorithmic feeds, no time-on-site metrics. Gregator succeeds when you learn something and leave.

Synthesis, not just summaries

Gregator builds interpretations grounded in sources — and when evidence conflicts or a source has weak backing, it tells you. You see where confidence is high and where it isn't.

Every claim is checkable

Click any citation to see the source, its credibility tier, and what the scoring found. Nothing is hidden behind a "trust me."

Answers to what people ask first

How is this different from a regular AI chatbot?

Chatbots give you a quick reply and vanish. Gregator gives you a structured, cited course you can come back to — chapters, inline citations, and conflict flags where serious people disagree. It's not a reply. It's something you keep.

Where do the sources come from?

Live web search, news outlets, and academic databases are read in parallel for every chapter. Sources aren't pre-packaged; they're gathered the moment you submit a topic, so what you read reflects what's out there right now.

What happens when sources disagree?

The disagreement is surfaced as a visible signal on the claim — not smoothed over with a confident-sounding summary. You see which sources support each side and how strong the evidence is for each.

Can I trust the claims?

Every claim is linked to its original source, and every source is scored for credibility — type (peer-reviewed, institutional, journalism, community), evidence quality, corroboration, and signals like promotional language. You can audit any claim in one click.

What counts as a "course"?

A multi-chapter document on one topic, built from dozens of sources, with claim-level citations and optional chat with Greg. Most courses finish in 2–4 minutes and span 3–8 chapters depending on complexity.

Can I download my courses?

Yes, on the Pro plan. Courses export to PDF and Markdown so you can read them offline, archive them, or feed them into your own tools.

What happens when I use up my quota?

You can keep reading and chatting about every course you've already generated — nothing disappears. Generating new courses is paused until the next billing cycle, or you can upgrade to unlock more.

Do you store or train on my data?

We store your generated courses and chat history so you can come back to them — that's the point. We do not train models on your data. Gregator uses several third-party LLM providers (DeepSeek, MiniMax, Gemini, Kimi) through their APIs; none of our usage opts in to provider training.

Start free. Upgrade when you need more.

Every plan includes the full research pipeline, claim-level citations, and structured multi-chapter courses. Paid plans add ongoing capacity and tools for long-term learning.

Free
Try it out — forever
$0
  • 3 courses, lifetime
  • Structured chapter output
  • Claim-level citations
  • Multi-source synthesis
Start free
Pro
For serious learners
$19.99 / month
  • 20 courses per month
  • Everything in Starter
  • Quizzes on every chapter
  • Export courses to PDF and Markdown
Start Pro
Try it

Pick a topic you've been meaning to understand

Three courses on us. No credit card needed.

Generate your first course