Sharly AI
Sharly AI is a research assistant that turns dense documents into source-backed answers. Upload PDFs, Office files or Notion exports, compare claims across sources, switch citation styles instantly, and share findings in an encrypted team workspace.
What is Sharly AI?
Sharly AI is a document research assistant published by VOX AI Inc. and used through the web application at app.sharly.ai. It accepts PDFs, DOCX, PPTX, XLSX, Google Docs and Notion exports, and it will also ingest a web page pasted as a link. Files can be dropped in directly or synced from Google Drive, Dropbox or Notion.
The moment a document lands, Sharly produces a summary, a set of topics and a few suggested follow-up questions. From there the work happens in a chat panel that reads across every document in the workspace at once. One-click prompts cover the common moves - compare, summarise, highlight, simplify, enhance, critique - and a highlighted sentence can be explained or questioned without leaving the page. Tables and charts can be captured section by section. Three answering modes are available: Quick for lookups, Balanced for everyday questions, and Thinking for slower, more thorough reasoning. A separate setting decides where answers may come from: Focused uses only the uploaded files, while Expanded lets the model add general knowledge.
Traceability is the product's centre of gravity. Every answer carries inline citations back to the source sentence, a document viewer lets that source be checked visually, and a reference panel switches between APA, MLA and Chicago formats. Sharly also pulls out titles, authors and DOIs automatically, alongside semantic extraction of topics, entities and figures. Optical character recognition, text-to-speech and speech-to-text appear in the feature grid.
Teams get shared libraries, real-time annotation, role-based permissions and activity logs; organisations add SSO and SAML, an admin dashboard, version history and private deployment. Files are encrypted with AES-256 at rest and TLS 1.3 in transit, workspaces are isolated, and personally identifiable information is redacted in real time. Sign-in runs exclusively through Google, Microsoft or Apple. A Slack integration exists, documentation lives on docs.sharly.ai, and the site is published in English and Spanish. There is no public API and no mobile app.
What it does
- Summarise dense reports, papers and transcripts into structured insights on upload
- Ask one question across several documents at once and get a single answer thread
- Compare metrics and claims between sources and surface contradictions or gaps
- Cite every statement with an inline link to the exact source sentence
- Switch citation format between APA, MLA and Chicago from a dropdown
- Restrict answers strictly to uploaded files with Docs-only mode
- Share a document library and annotate it with a team in real time
When to use Sharly AI / When not to
A quick filter to help you decide if Sharly AI is the right fit.
When to use Sharly AI
- Academic researchers, postdocs and faculty running literature reviews who need APA, MLA or Chicago citations pulled straight from the source
- Compliance officers, internal auditors and legal reviewers cross-checking contracts, audit trails and interview transcripts for contradictions
- Analysts, policy specialists and think-tank researchers extracting figures and comparing claims across long reports
- Investors and due-diligence teams reading decks, memos and market reports at volume
- Procurement and category managers lining up specifications and spotting clause deviations across supplier documents
When not to use Sharly AI
- Developers who need programmatic access: Sharly publishes no API and its documentation contains no API reference
- Mobile-first users: there is no iOS or Android application, only a browser workspace and a Slack integration
- Organisations with strict procurement rules, since the terms of service, privacy policy and cookie policy are all unreachable
- Teams that must document a specific data-hosting jurisdiction, as no hosting country or region is disclosed anywhere
- Anyone unable or unwilling to sign in through Google, Microsoft or Apple, since Sharly offers no standalone password account
How to use Sharly AI
A typical end-to-end flow, from setup to results.
- Sign up at app.sharly.ai using a Google, Microsoft or Apple account - Sharly issues no password of its own
- Drop a file onto the dashboard, or use the Add Link card to paste the URL of a web page
- Connect Google Drive, Dropbox or Notion if the source documents already live there
- Read the summary and topics Sharly generates automatically, then pick one of the suggested follow-up questions
- Choose an answering mode in the chat panel: Quick, Balanced or Thinking
- Open the gear icon and set the content source to Focused for document-only answers, or Expanded to allow general knowledge
- Check each answer through its inline citation and the document viewer beside it
- Pick the citation format you need - APA, MLA or Chicago - from the reference panel dropdown
- Invite teammates on the Team or Business plan, assign roles, and annotate the shared library together
- Manage or cancel the subscription under Settings, Subscription, Manage Subscription, which opens the Stripe portal
Pros & Cons
Pros
- Every answer traces back to the exact source sentence, verifiable in a side-by-side document viewer
- Three standard academic citation formats, switchable from a dropdown, with DOI and author extraction
- Docs-only mode gives a genuine guarantee that nothing outside the uploaded files enters an answer
- Clear published position on model training: the vendor states it never trains on customer data
- Encryption is documented rather than implied - AES-256 at rest, TLS 1.3 in transit, key rotation
- A permanent free plan with no commitment, and a feature-by-feature pricing grid that is unusually detailed
- Public user documentation, a Discord community and a Slack integration for teams already working there
Cons
- The terms of service, privacy policy and cookie policy are all unreachable - the embedded Termly documents have been deleted
- The GDPR is never mentioned, and there is no DPA, no subprocessor list and no EU representative
- No postal address for the publisher appears on any page of the site
- No hosting country or region is disclosed; data residency is a Business-plan bullet with no detail attached
- No public API and no mobile application, so the product is a browser workspace plus Slack
- Sign-in is possible only through Google, Microsoft or Apple, with no standalone account
- The site contradicts itself in places: 20 versus 100+ supported languages, and 5 documents a day versus 10 a month on the free plan
Pricing & Plans
Sharly AI offers a permanent free plan at no cost. The lowest paid entry point is the Pro plan at USD 12.50 per month when billed annually, or USD 15.00 per month on a monthly cycle. The Team plan is USD 24.00 per seat per month billed annually, or USD 30.00 per seat monthly, and the Business plan is quoted on request. Annual commitment saves 20 percent across the paid tiers.
- multi-format upload
- instant summaries and AI answers
- a 10k context window
- one shared workspace and email support. The pricing page states 5 documents per day in the plan card and 10 documents plus 50 messages per month in the comparison table.
- unlimited uploads
- 1
- 000 messages per month
- a 50k context window
- advanced reasoning
- no LLM training on inputs
- advanced security protocols and email plus chat support.
- USD 24.00 per seat per month billed yearly
- or USD 30.00 per seat monthly
- up to 100 seats. Adds unlimited documents and chats
- a 124k context window
- team annotations
- knowledge retention
- semantic extraction
- analytics
- custom pricing
- unlimited seats. Adds an admin dashboard with role-based access
- version history and audit logs
- SSO and SAML
- private deployments and a dedicated support line.
- free pilot of the Team feature set
- offered by application to selected research labs
- academic teams and publishing groups.
Data, GDPR & hosting
A consolidated view of how Sharly AI handles your data.
GDPR overview
There is no mention of the GDPR anywhere on Sharly's website. A search across every page collected, including the raw HTML, returned zero occurrences of the regulation, of a data processing agreement, of a subprocessor list, of an Article 27 representative or of a data protection officer. The pages that would normally carry this information are unreadable: /terms, /privacy and /cookie-policy each embed a Termly document that no longer resolves. The only privacy-adjacent machinery still working is the Termly consent banner and its cookie resource blocker. Security controls are documented on a separate page, but security is not compliance. European buyers should treat Sharly's GDPR posture as undocumented and ask the vendor for a written position before processing personal data.
Who owns the data?
Sharly's terms of service, privacy policy and cookie policy are published only as embedded Termly documents, and all three returned Policy Not Found when checked on 25 August 2026. No archived snapshot has ever carried that text inline either, so ownership of uploaded content cannot be established from any published contract. What the vendor does state is operational rather than contractual: data is segregated by workspace and organisation through row-level access controls, workspaces are isolated from one another, and content is deleted automatically after thirty days without use. Account holders can erase everything permanently from Settings, then Security and Privacy, then the Danger Zone. Anyone needing a written ownership clause must request it from the vendor directly.
Reuse rights
No contract governing reuse is available, because the three legal documents the site links to have been removed at the provider's end. The vendor's own statements are narrower but explicit: the models Sharly uses strictly prohibit the use of input data for any model re-training purposes, and the pricing FAQ repeats that the company never trains on customer data. Uploaded files stay inside the workspace they were added to, and a Docs-only mode restricts answers to those files alone, with no external sources. Personally identifiable information is scanned for and redacted in real time. Whether a customer may republish or commercially exploit the summaries and citations Sharly produces is simply not addressed anywhere that can be read.
Data retention & training
Hosting summary
Sharly discloses no hosting jurisdiction. No country and no region is named on any page, and data residency appears only as a bullet in the compliance block reserved for the Business plan, with no detail attached. The documents that would normally carry this information - the terms and the privacy policy - are unreachable. What can be observed independently is infrastructure rather than a vendor statement: the domain resolves to an Amazon Web Services anycast address announced from AS16509, geolocated in the United States. That describes an edge node, not where files are stored. The controls Sharly does document are logical rather than geographic: data is segregated per workspace and organisation with row-level access controls, workspaces are isolated, files are encrypted with AES-256 at rest and TLS 1.3 in transit, and keys are rotated. Organisations with a residency requirement should treat hosting location as unknown and raise it with the vendor before uploading anything sensitive.
Things to keep in mind
Risks and trade-offs to weigh before adopting Sharly AI.
- The three legal documents are unreadable, so no user can know what they have agreed to regarding ownership, retention or subprocessing
- Data is deleted automatically after thirty days of inactivity - a workspace treated as an archive can quietly empty itself
- Source-backed answers feel authoritative, and the polish of a citation can discourage the reader from opening the document behind it
- Leaning on automatic summaries erodes the slow reading that produces real familiarity with a corpus, particularly for students
- Expanded mode blends model knowledge with document content, and the boundary between the two is easy to lose track of in a long thread
- Sign-in depends entirely on Google, Microsoft or Apple: losing that account locks the workspace out
- SOC 2 Type II appears as a plan feature rather than a dated attestation, and may be read as a certification the site never actually claims
Setup & Integrations
Technical difficulty
Very low for an individual. There is nothing to install: Sharly runs in the browser, sign-up takes a minute through Google, Microsoft or Apple, and the first summary appears as soon as a file is dropped in. No configuration is required. Team setup is barely heavier - invite seats and assign roles from settings. Only the organisational tier needs help: SSO and SAML, the admin dashboard and private deployment are arranged through the sales team. The Slack integration is a three-step flow: add Sharly to a channel, authenticate, then start typing.
Deployment
Integrations
Supported languages
Behind Sharly AI
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What happens if I hit my document or message limit?
How do I sign in to Sharly?
Can I stop Sharly from using sources outside my own files?
Does Sharly train its models on my documents?
What is a seat and how is it counted?
Can I pay monthly instead of yearly?
Where can I read Sharly's privacy policy and terms?
Is there an API or a mobile app?
How long does Sharly keep my data?
Are there discounts for students or universities?
Should you pick Sharly AI?
Sharly AI is a coherent, well-built research assistant with one clear idea behind it: no answer should be readable without its source. Inline citations tied to the exact sentence, a document viewer for checking them, APA, MLA and Chicago switching, automatic DOI and author extraction, and a Docs-only mode that keeps the model inside the uploaded files - together these make it a credible tool for literature reviews, contract review and analyst work where a claim has to be defensible. The collaboration layer is real too: shared libraries, live annotation, role-based permissions and activity logs, with SSO, audit trails and private deployment reserved for the Business tier. Encryption is stated precisely rather than gestured at.
The reservation is serious and it is not about the product. Sharly publishes no readable legal documents. The terms of service, privacy policy and cookie policy are embedded Termly documents that have been deleted at the provider's end, and no archived version has ever carried the text. The GDPR is not mentioned once across the entire site, there is no data processing agreement, no subprocessor list, no EU representative, no stated hosting jurisdiction and no postal address for the publisher. Buyers in regulated environments cannot complete a vendor assessment from what is published, and the SOC 2 Type II line is a plan bullet, not an attestation. Smaller frictions add up alongside this: no API, no mobile app, mandatory third-party sign-in, and a site that contradicts itself on supported languages and free-plan quotas.
For an individual researcher, the free plan makes trying it costless and the Pro tier is inexpensive. For an institution, the sensible order is to ask for the legal documents in writing first, and evaluate the product second.
- Choosing a selection results in a full page refresh.
- Opens in a new window.