Use 3Dogs Nexus inside your assistant

Live now. 3Dogs Nexus works with Claude (every plan, including Free) and Grok. Set it up in a few clicks below; 3 days free with just your email, no card. Then try the fresh-facts test on the proof page.

What it does

3Dogs Nexus is a verification-and-reasoning layer that sits behind your AI assistant. You stay in your assistant—whether it’s Claude or Grok—and it acts as the coordinator. When you face a hard question, you hand it to 3Dogs. Your assistant decides when and how to send the work, combines the results, and explains everything back to you. 3Dogs runs the multi-model research: it grounds your question in live evidence, convenes a panel of independent models for debate, reconciles contradictions, and returns a cited, evidence-backed answer. Your host assistant remains in charge, and can even participate as a named voice in the analysis.

New: a free six-lesson video course - see 3Dogs inside Claude and Grok, shared memory, the pro-level audit move, the Library and big files. Watch the lessons.

See the proof: test 3Dogs on what your assistant can't know, then read a full recorded session. See the proof.

Set up in Claude

Available on all Claude plans - Free (one custom connector), Pro, Max, Team and Enterprise. Team/Enterprise owners must enable connectors in Organization settings first.

In Claude (web or desktop), go to Settings > Connectors.
Click Add custom connector.
Name it "3Dogs Nexus". For the Server URL, enter: `https://connect.3dogs.ai/mcp`
Click Add, then Connect. You'll be prompted to sign in via an emailed link.

If you added the connector before October 29, 2026, remove it and re-add it once to access the new memory tools.

Set up in Grok

Available now at grok.com.

Go to grok.com/connectors.
Click New Connector > Custom.
Paste the Server URL: `https://connect.3dogs.ai/mcp`
Save it, then sign in with the emailed link.

Existing users should click Reauthenticate once to enable new features.

The four phrases

Interact with 3Dogs Nexus through simple, natural language prompts to your assistant:

"Run this in 3Dogs" – For deep research, strategy, diligence, or analyzing large documents (up to ~800,000 characters). Your assistant sends the question and any material.
"Audit this with 3Dogs" – To independently cross-check an answer from your assistant or another source. 3Dogs provides an AUDIT verdict.
"Challenge that 3Dogs answer" – If a result seems off, you can push back. Your assistant sends a structured CHALLENGE for a fresh panel to rule on.
"Run it deeper" – For decisions with real stakes. 3Dogs suggests it when a question needs it: 9 models from 8 makers answer independently from up to 8 live searches, reviewers check every claim against the sources, and you see where the panel split. About 3 minutes.

Memory that follows you across assistants

Your 3Dogs memory is private, persistent, and personal. Notes you store (`remember_3dogs`) are tied to your sign-in email and can be recalled (`recall_3dogs`) in any chat, in any connected assistant. If you use both, it is a shared memory—the same memory follows you between assistants (tested: stored in Claude, recalled in Grok on the same account). Run a 3Dogs query in Claude, then audit the whole thing in Grok (or the reverse)—two independent assistants plus the 3Dogs panel, with the memory carrying the context across. Every operation is audited. Sensitive data like credentials is rejected, and erasing memory requires explicit confirmation.

What people use it for

Strategy Decisions: *"Should we expand into a second region this year, or consolidate first? Show the evidence on both sides."*
Risk Analysis: *"What are the three biggest risks to this project plan and which one should I address first?"*
Supplier Comparison: *"We are considering two suppliers with different terms. Which one leaves us less exposed if demand drops 15%?"*
Document Deep-Dive: *"Review this 200-page RFP and list the top 5 compliance gaps."* (Use `send_to_3dogs` for large files).
Fact-Checking & Audits: *"Audit the executive summary of this market report for overstated claims."*
Technical Research: *"Compare the latest implementations of technique X across three recent papers."*
Creative Stress-Testing: *"Challenge the weakest premise in this campaign concept."*
Personal Knowledge Base: *"Remember that my competitor's keynote is on November 12th."*

Data and limits

The engine uses over 100 models across 5 clouds. Your questions and material are processed by outside providers (currently Amazon Web Services, Microsoft Azure and Anthropic), and processing may take place in the United States; this data is not used to train AI models and is not sold. 3Dogs Nexus provides research support, not professional legal, tax, financial, or medical advice. For users 18+. Use is subject to plan limits. 3 days free at the Professional allowance with just your email—no card to start. A card is only needed when you subscribe at the end.

Works with Claude and Grok.

Claude is a trademark of Anthropic, PBC. Grok is a trademark of xAI. 3Dogs Nexus is independent and not affiliated with, endorsed by, or sponsored by Anthropic or xAI.

Claude, Grok, Gemini, ChatGPT/GPT, Mistral Le Chat, DeepSeek and Alexa+ are trademarks of their owners. 3Dogs Nexus is independent and not affiliated with or endorsed by them.