Question 1

Why should I use Irys instead of Claude, GPT, Perplexity, and Lexis separately?

Accepted Answer

Each of those tools is strong at one thing. but every one of them is amnesiac and isolated. Claude forgets everything when you close the tab. Perplexity has no matter context. GPT has no legal structure. NotebookLM has no output workflow. And none of them know your firm, your client, or your history. More critically: using consumer AI tools for client work without proper DPAs may violate your professional conduct obligations under ABA Model Rule 1.6. Irys is built with data isolation, zero retention, and architecture designed for attorney-client privilege. One platform, one subscription, full compliance.

Question 2

How does Irys compare to Harvey or Legora?

Accepted Answer

Harvey and Legora are general-purpose legal AI tools built for large law firms. Irys is purpose-built for the full legal workflow. research, drafting, redlining, matter management. with a three-tier reasoning architecture and citation verification against 50M+ cases. Irys is also significantly more accessible on pricing, with no enterprise-only gating.

Question 3

What is Irys?

Accepted Answer

Irys One is a unified AI platform for legal work. research, draft, redline, manage matters, and collaborate in one system built specifically for legal professionals. It replaces five tools with one.

Question 4

How is this different from general AI?

Accepted Answer

General AI tools like ChatGPT and Claude have no legal-specific guardrails, no citation verification, no matter management, and no audit trail. Irys starts with your matter. documents, facts, and legal context. and produces real legal analysis with sourced citations against 50M+ cases.

Question 5

Will Irys replace my legal team?

Accepted Answer

No. Irys augments your team, not replaces it. It handles the time-consuming research, first-draft generation, and document comparison so your lawyers can focus on strategy, judgment, and client relationships. Think of it as giving every attorney a tireless associate.

Question 6

Do you have your own LLM, or do you use existing models?

Accepted Answer

We do not build our own LLM from scratch. We use the best available foundation models and are model-agnostic. we route to the right model for the right task. What we own is the infrastructure around it: the retrieval pipeline, the citation verification engine, the knowledge graph, the matter context layer, and the reasoning orchestration. The LLM is the engine. We built the car.

Question 7

What does unlimited context actually mean?

Accepted Answer

It is not a metaphor. Our three-layer architecture (Worker / Coordinator / Synthesizer) processes documents across tiers so that matter context is preserved regardless of how many documents you upload. No token limits, no chat resets. Firms upload hundreds of documents into a single matter and the AI works from the full matter every time.

Question 8

Is Irys a wrapper around ChatGPT?

Accepted Answer

No. Wrappers take a general-purpose LLM, add a prompt template, and charge a markup. Irys is purpose-built infrastructure: a three-tier reasoning architecture, proprietary retrieval pipeline, citation verification against 50M+ cases, a knowledge graph engine, and a matter-native context layer. We rent the intelligence from the best AI models available. We own the legal system around it.

Question 9

Do I need technical expertise or IT support to use Irys?

Accepted Answer

No. Irys is designed for lawyers, not engineers. The interface is intuitive and requires zero technical setup. Most users are productive within minutes. For enterprise deployments, we handle onboarding and configuration.

Question 10

What is Agentic Retrieval?

Accepted Answer

Agentic Retrieval (AR) is a complete replacement for the previous AI Assistant, not an update to it. Instead of answering in a single pass, AR runs a structured investigation: it plans the research, pursues multiple lines of inquiry in parallel, validates case law against primary sources, and synthesizes a legal memorandum with inline citations linked back to their source. The result is better reasoning, more reliable citations, and answers grounded in verifiable sources rather than model inference alone.

Question 11

Is Agentic Retrieval included in my Irys plan?

Accepted Answer

Yes. Agent mode is available on all Irys plans at no additional cost.

Question 12

Do I need to do anything to start using it?

Accepted Answer

No. Agent is now the default mode when you open the AI Assistant. You don’t need to change any settings, it’s on automatically.

Question 13

What are the three modes and when should I use each one?

Accepted Answer

Agent, Use this for almost everything. It’s the right choice whenever you need a thorough, cite-ready answer: analyzing matter documents, validating case law, or any question you’d want to show a colleague. Agent is the new default and where all future improvements will land. Research, Use this for quick, single-pass lookups: finding a statute, pulling a regulation, or a fast “are there cases on X” scan where a short answer is enough. Deep Research, Use this if you specifically want a long-form narrative report and are willing to wait for it. In most cases Agent will produce the same depth of output faster and with better citations. We recommend trying Agent first. If you’re unsure, start with Agent.

Question 14

What’s the difference between Agent and Research?

Accepted Answer

Research is a single-pass lookup, fast and lightweight, good for quick questions. Agent is a full investigation: it plans, pursues multiple leads in parallel, validates case law against CourtListener, and returns a structured memorandum with inline citations. For anything you’d actually cite or show to someone else, use Agent.

Question 15

What’s the difference between Agent and Deep Research?

Accepted Answer

Deep Research was designed for long-form, comprehensive reports. Agent now produces equivalent or better output faster, with the added benefit of showing you the full investigation as it runs. Deep Research remains available for users who have workflows built around it, but it’s in maintenance mode, all new development is in Agent.

Question 16

How do I switch between modes?

Accepted Answer

Click the Agent button in the bottom left of the chat panel to open the mode selector dropdown and choose your preferred mode.

Question 17

What is Thinking and how do I use it?

Accepted Answer

Thinking is the live investigation panel that opens when you submit a query in Agent mode. You can follow along in real time as AR identifies key issues, pursues each lead, and pulls in relevant sources. Once the query is complete, Thinking displays a compressed summary of every lead and cross-reference used to generate the answer, so you can see not just what AR concluded, but how it got there.

Question 18

What happens after I submit a question in Agent mode?

Accepted Answer

Agent runs a four-step process. First, the Thinking panel opens showing the planned investigation and key issues to pursue. Second, leads run in parallel, each one a line of inquiry such as a case to validate, a document section to read, or a statute to retrieve. Third, a structured legal memorandum is written with inline superscript citations. Fourth, the Sources & Citations panel populates with everything retrieved, grouped by Case Law and Web, with a search function and Copy All option.

Question 19

Does Agent mode work with my matter documents?

Accepted Answer

Yes. Agent is matter-aware by default, when you’re working inside a Matter, AR prioritizes your attached documents alongside external legal sources. Document citations link directly to the source PDF at the cited page.

Question 20

What’s the difference between Case Law and Web Search in the Sources panel?

Accepted Answer

Case Law refers to opinions retrieved and validated through CourtListener, a free, authoritative repository of U.S. legal opinions, oral arguments, and federal filings. Web Search refers to additional relevant sources located through online search. Both are grouped separately in the Sources & Citations panel so you can see exactly where each piece of information came from.

Question 21

How do inline citations work?

Accepted Answer

Hover over any footnote number in Agent’s output to see the citation and a hyperlink to the source. Click the footnote number to open the full Sources & Citations panel, where all references are listed and grouped by type. From either view, you can navigate directly to the full opinion on CourtListener or to the source document page.

Question 22

How does Irys verify case law citations?

Accepted Answer

All case law references are validated against CourtListener, which covers 50M+ U.S. legal citations going back over 200 years. Clicking any case citation takes you directly to the full opinion on CourtListener. You should always review citations independently, AR is AI, but the verification chain is visible and checkable at every step.

Question 23

Do I need a CourtListener account or subscription to use Agent mode?

Accepted Answer

No. CourtListener is a free resource and no account or subscription is required.

Question 24

Does Agent mode affect how quickly Irys responds?

Accepted Answer

Agent is medium-speed, slower than Research’s single-pass lookup, but faster than Deep Research on comparable queries. Because the Thinking panel shows the investigation as it runs, you can follow progress throughout rather than waiting for a final answer to appear all at once.

Question 25

Will my existing saved prompts and chat history still work?

Accepted Answer

Yes. Nothing changes for your existing prompts or chat history. Agent is simply the new default mode, all your prior work is unaffected.

Question 26

Can I still use Research and Deep Research?

Accepted Answer

Yes. Both modes remain available and can be selected at any time from the mode dropdown. Agent is the default, but switching is a single click.

Question 27

Is my data secure and private?

Accepted Answer

Yes. Zero data retention. we never store or train on your data. End-to-end encryption (AES-256 at rest, TLS in transit), tenant isolation, MFA, SOC 2, ISO 27001, GDPR-aligned, and CPRA-compliant infrastructure. Your data never leaves the Irys environment. Critically, Irys is a data processor, not a data retention company. LLM calls use ephemeral encrypted tokens. no content is retained after the session. Unlike AI tools that route through OpenAI or Anthropic and retain data indefinitely, your client data never accumulates on third-party servers subject to subpoena.

Question 28

Do you train on my data?

Accepted Answer

Absolutely not. Your data is never used to train, fine-tune, or improve any AI model. This is a contractual guarantee, not just a policy. Your client data stays yours.

Question 29

Can my Irys data be subpoenaed?

Accepted Answer

This is the question most firms have not asked yet. OpenAI has been successfully subpoenaed and has produced full chat histories, uploaded documents, and PDFs. Consumer AI tools. including Harvey, Legora, and CoCounsel, which are wrappers over OpenAI or Anthropic APIs. store your data on those servers. Irys operates as a data processor, not a data retention company. Approximately 80% of AI processing runs on Irys-controlled infrastructure. LLM calls use ephemeral encrypted tokens. nothing is retained after the session ends. There is no accumulated client data for a subpoena to reach.

Question 30

How accurate is Irys?

Accepted Answer

Irys uses a multi-layered architecture built for legal analysis. Document analysis extracts structured context from your matters. Agentic investigation plans the research, pulls case law, and validates every citation against primary sources. Synthesis produces cite-ready memos with inline citations linked to source. Every claim is traceable, not just plausible.

Question 31

Does Irys cite its sources so I can verify the output?

Accepted Answer

Yes. Every research output includes full case citations with pinpoint references. You can click through to verify against the original source. Citation verification runs against 50M+ cases.

Question 32

Can I control how Irys drafts?

Accepted Answer

Yes. You can provide templates, style guides, and firm-specific preferences. The Drafting Assistant adapts to your voice, your formatting, and your firm’s conventions. The Library module stores your templates for reuse across matters.

Question 33

What content types are supported?

Accepted Answer

Irys supports PDFs (including scanned documents via proprietary OCR), Word documents, Excel spreadsheets, plain text, and more. Our OCR converts scans into reliable, searchable text tied back to the original document.

Question 34

How many documents can I upload to a matter?

Accepted Answer

You can upload up to 500 files in a single bulk upload, with a 50 MB cap per file. There is no limit on how many documents a matter can hold over time. keep uploading as the matter grows. The AI reads every page of every document, regardless of how many you add.

Question 35

How many documents can Irys analyze at once?

Accepted Answer

Multi-Doc Analysis runs structured queries across up to 50 documents per session. Define your questions once, get a matrix of answers with citations, evidence, and confidence scores for every document. For larger reviews, run Multi-Doc Analysis in focused batches and pull the results into the matter workspace.

Question 36

Are there any limits on chats, uploads, or AI use?

Accepted Answer

No. Every paid plan gives you unlimited matter memory, unlimited AI queries, and unlimited chat length. No document caps, no token limits, no session resets. Upload up to 500 files per bulk upload. the whole matter stays in context every time you ask.

Question 37

Does Irys support voice-to-text?

Accepted Answer

Yes. Voice-to-text is live. dictate prompts, questions, or content directly into Irys without typing.

Question 38

Can I use Irys alongside other tools?

Accepted Answer

Yes. Irys includes a Word Add-In that brings Chat mode and Work mode directly into Microsoft Word for desktop. You can also export documents, research, and analysis in standard formats.

Question 39

Is the Word Add-In included in my plan?

Accepted Answer

Yes. The Word Add-In is included with Irys One and Enterprise at no additional cost. There are no add-on fees and no token limits. Install it from within Word (Insert → Add-ins → Get Add-ins → search Irys) or from the Microsoft AppSource marketplace.

Question 40

Does Irys integrate with existing document management systems (e.g., iManage, NetDocuments)?

Accepted Answer

DMS integrations are in development. For now, you can easily upload and organize documents within Irys. Enterprise customers can discuss custom integration timelines with our team.

Question 41

Can I share with clients or co-counsel?

Accepted Answer

You can export any work product. research memos, drafts, comparisons. in standard formats (Word, PDF) to share externally. In-platform sharing with external parties is coming with our collaboration features.

Question 42

When will collaboration features be available, and what will they include?

Accepted Answer

Collaboration features are actively in development. They’ll include shared workspaces, real-time co-editing, commenting, matter-level team chat, firm-wide playbooks, and a client portal. All of these ship as part of Irys One at no extra cost. your $299/month rate stays the same. Customers who sign up after these features launch pay more.

Question 43

Can I control who on my team sees which matters or documents?

Accepted Answer

Yes. Role-based access controls are included on Enterprise plans. you can restrict matter visibility by user, practice group, or department, with ethical walls and department-level silos. Firms under 25 seats that need these controls can contact us about custom multi-seat arrangements.

Question 44

How does pricing work?

Accepted Answer

Irys One is $299/month. the full legal AI platform. Unlimited matters, unlimited documents, unlimited AI queries. Enterprise is custom pricing for 25+ seats. Sign up today and your $299 rate stays fixed when multi-user team features ship. those features are included in your plan at no extra cost. Customers who join after those features launch pay more. Save 15% with annual billing ($3,049/year).

Question 45

How do trials work?

Accepted Answer

Full platform access for 7 days. No credit card required. Your data, matters, and work product carry forward if you subscribe. Start immediately.

Question 46

Is there a minimum contract length, or can I cancel anytime?

Accepted Answer

Monthly plans can be cancelled anytime. Annual plans are billed upfront at a 15% discount. No long-term lock-in on monthly billing.

Question 47

Are there rewards for referrals?

Accepted Answer

Yes. We have an active Partner Program with affiliate, reseller, and strategic tracks. Earn 10–30% commissions depending on the partnership type.

Question 48

Do you offer onboarding support or training?

Accepted Answer

Yes. Irys One includes documentation, guides, and email support. Enterprise plans include dedicated onboarding sessions, training for your team, and a dedicated success contact.

Question 49

What practice areas does Irys support best?

Accepted Answer

Irys is used across litigation, corporate/M&A, insurance defense, employment law, mass tort, real estate, IP, regulatory & compliance, privacy law, estate planning, immigration, and general practice. The platform adapts to your workflow regardless of practice area.

Question 50

Does it cover international or non-U.S. jurisdictions?

Accepted Answer

Irys currently focuses on U.S. federal and state law across all 50 states. International jurisdiction support is on our roadmap. Contact our team if you have specific international requirements.

Frequently Asked Questions

Still have questions?