v0.8.0 in production

Discovery

Center Deep Research

Available now Does the work of a research analyst

Ask a hard question, get a cited report with per-claim confidence, synthesized on hardware you own. Your AI agents can use it too.

Search free at centerdeep.ai Email us about Research seats

Center Deep search is open to everyone. Research seats are invited in small batches while credit pricing is finished.

Center Deep private search
Center Deep private search, live at centerdeep.ai. Research rides on top.

Who it is for

For researchers who need sources, not vibes.

Researchers, analysts, journalists, operators, and agent builders who need real answers with receipts, not a tracking-first search box or a vague AI summary.

Why it wins

Depth you can cite.

01

Reports you can defend. Every claim carries its source and a confidence score, so you know exactly what's solid and what's thin.

Turns one question into multi-angle search, reranking, and a synthesized report with a citation and confidence grade on every claim.

02

Three speeds, one engine. Quick search when you need an answer now, deep mode when the answer has to hold up.

Pulls from graded academic (CORE), medical (PubMed, EuropePMC), and legal (CourtListener) authorities plus the open web, then loops back to fill gaps.

03

Sovereign by default. Synthesis runs on local GPUs we own, so your questions never train someone else's model and cloud LLM cost is roughly zero.

Remembers what it found. Evidence persists to your research corpus, so every new question starts smarter than the last.

How it works

Ask. Trace. Deliver the brief.

01

Ask

Ask your question and pick a mode: search, balanced, or deep.

02

It digs

The engine fans out across web, academic, medical, and legal sources, reranks the evidence, and cross-checks authorities against each other.

03

You get receipts

You get a cited, confidence-graded report, and the evidence lands in your corpus for next time.

See it

Inside the product.

Center Deep private search
Center Deep private search, live at centerdeep.ai. Research rides on top.

Proof and access

What's proven today.

18signed MCP tools for agents
400backend tests passing
~$0cloud LLM cost, synthesis runs local

Proof points

  • Version 0.8.0 running in production at research.unicorncommander.ai, with a health check you can hit.
  • Agent-native from day one: a signed MCP manifest exposes 18 tools, and our own Brigade agents already use it as their research arm.
  • 400 backend tests passing at the latest integration merge.
  • Hard cost of a balanced run is about 3 cents of upstream search. Synthesis runs on our local Qwen GPUs, so the LLM bill is roughly zero.
  • We run our own company's research on it every day, and prior findings compound in a persistent per-space corpus.

Pricing and access

Center Deep search is free and open at centerdeep.ai today, no account needed. The Research surface runs our own company's research daily; external seats are invited in small batches while credit pricing is finished.

Shipping log

What just shipped.

  • June 2026: v0.8.0 'Living Intelligence' shipped to production with graded academic, medical, and legal source verticals.
  • July 2026: persistent research corpus went live, so evidence from past runs now compounds into future answers.
  • July 2026: dedicated customer production stack launched at research.unicorncommander.ai with invited external seats.

FAQ

Plain answers.

How is this different from Center Deep search?

Center Deep is our private search engine, live at centerdeep.ai. Center Deep Research sits on top of it: instead of a results page, you get a synthesized report where every claim is cited and confidence-graded.

Does it actually cite sources?

Yes, that's the whole point. Every claim traces to its source, cross-checked across academic, medical, legal, and web authorities, with a confidence score so you can see where the evidence is strong and where it's thin.

Where does my research data go?

Synthesis runs on GPUs we own, not a frontier cloud LLM, and your evidence persists in your own research corpus. Your questions don't become someone else's training data.

Can my AI agents use it?

Yes. The engine ships a signed MCP manifest with 18 tools, and agents get the same search, deep research, and corpus capabilities humans do. Our own Brigade agents already run on it.

What does it cost?

Nothing charges today. We're finishing credit-based pricing now, and early users will hear about it first.