Pont du Gard, Roman aqueduct at golden hour
Ǣ ǣwita (n.) — Old English. A counselor of the law.

Legal AI that won't make things up.

Your firm's work product. Your firm's methodology.
Research, drafting, and work product in one platform.

Try it free for 14 days Request a demo
No seat minimums. · Cancel anytime. · Just $99 per month.

You shouldn't have to send your clients' files to third-party LLMs
to get work done faster.

  • Your Data Never used to train a model. Ever.
  • Your AI We host a top-scoring reasoning model so your work stays secure.
  • Your Terms Stop choosing between power and privacy. Try it free for 14 days
From 1665 A.D. through Today
Every U.S. case and court opinion on record.
The Full American Legal Corpus
Current statutes across all jurisdictions, state and federal.
AI That Does Not Hallucinate
Controlled experiments have never produced a hallucination.
The product

One login. One answer. Every citation.

Aewita unifies research, drafting, and your firm's playbooks into a single workspace. Watch a real query resolve — from question to sourced answer — in under nine seconds.

app.aewita.com  ·  /assistant Grounded in your documents
Three verbs

Find it. Think it through. Write it.

Three ways Aewita meets you in the work. One system of record. One citation layer under all of them.

01

Find it.

Ask anything about your case. Get the answer with the citation.

📚 Library Grounded
Query
What's the 2d Circuit rule on equitable tolling for §1983 actions?
Three-prong test. Diligence + extraordinary circumstance + causation.[1]
1
Pace v. DiGuglielmo 544 U.S. 408 · Supreme Ct.
✓
02

Think it through.

Talk to an AI that remembers every matter you've worked on.

💬 Assistant Memory on
Does this argument contradict anything in Kline Holdings?
Æ
Two conflicts in §4.3 covenants.
↳ Diff view Cite the clause
03

Write it.

Tell it what you need. Get a first draft with every clause sourced.

✎ Scrivener Drafting
Motion to Dismiss — Diversity, Amount in Controversy
[3]
[7]
12 pp·34 citations·✓ all verified
Architectural privilege

No OpenAI.
No Anthropic.
No Google.

We run the AI so nobody sees your clients' files. Not even us.

01
Zero external LLM calls.
02
State-of-the-art reasoning power. We tested all frontier models including Claude Opus 4.6 and Gemini 3.0 Pro and selected the top performer to power Aewita.
03
Never train on your data. We built Aewita so all lawyers can work securely. You don't need an enterprise contract to opt out of training.
Read our security practices →
Hallucination rate

All legal AI hallucinates. Aewita doesn't.

Industry hallucination rates range from 5% to 35%+. In controlled experiments, Aewita's infrastructure has never produced a hallucinated output. Statistically speaking, we can say under 0.3% with 95% certainty.

General-purpose LLM (upper bound)
~35%
Typical legal AI wrapper
~15%
Best-in-class retrieval-augmented
~5%
Aewita
<0.3%
Methodology: 800 controlled, real-world queries across 22 practice areas. Each answer was individually checked and verified to produce only real and accurate citations and analysis.
Playbooks

Your best partner's brain, searchable.

When a senior partner leaves, their methodology shouldn't leave with them. Aewita's playbooks let a first-year associate execute a tenth-year partner's methodology — searchable, systematized, firm-wide.

See how playbooks work →
Scrivener / Playbooks / Series A — Kline style Acme  ·  Series A ▾
📚
💬
✎
⚏
◎
Series A Term Sheet — Kline style
v 14 · edited by JK  ·  5 steps
Playbook running
  1. 01 ✓
    Confirm cap-table post-money vs. pre-money framing
    Pulled from Kline deals · 2023–2025
  2. 02 ✓
    Compare liquidation preference to last 3 Kline deals
    3 precedents found · 1× non-participating
  3. 03 ✓
    Flag founder-vesting any term < 4 yr / 1 yr cliff
    0 flags raised
  4. 04
    Draft investor-rights section from template A3
    Generating · 120 of ~180 lines
  5. 05
    Review w/ Kline before send — partner habit
    Waits for step 04
3 of 5 steps complete Open draft →
ABA Model Rules

Compliance by architecture, not by promise.

Privilege preserved by design. Not by a vendor's terms of service.

Rule 1.1

Competence

Every answer traces to the primary source. A lawyer's duty to verify is met without leaving the workspace.

Rule 1.6

Confidentiality

Client files never reach a third-party model. No subprocessor. No data-processing agreement for an LLM vendor because there isn't one.

Rule 5.3

Supervisory Responsibility

Audit logs show what the AI did, what it cited, and what a human approved. Reviewable by the responsible attorney.

What Aewita replaces

One login. One bill. One platform.

Westlaw Lexis Harvey CoCounsel MidPage Generic DMS AI
→
ÆWITA
Pricing

Plain numbers. No seat minimums.

Every plan starts with 14 days free. We charge on day 15 unless you cancel.

Enterprise
Custom
Custom pricing for law firms
  • White-glove onboarding & training
  • Dedicated customer engineer
  • On-premise & private-cloud deployments
  • Deploy inside Azure, AWS, or GCP tenancies your firm already approves
  • SSO, SCIM, audit-log export, custom DPAs
  • Volume pricing & master service agreements
Request a Demo
Most lawyers choose this
Individual
$ 60 /mo
$720 billed annually — save $468
  • Full research across every U.S. case & statute
  • Unlimited AI chat on your matters
  • Drafting for 792 document types
  • Firm-wide playbooks
  • DMS integrations (NetDocuments, iManage)
  • MCP connectors for Claude & other assistants
  • No seat minimum. No setup fee.
Try free for 14 days
See full pricing details →
Sam Anderson, CEO and Founder of Aewita
Samuel Anderson
CEO & Founder
From the founder

Built by an inventor to solve the problems with Legal AI.

Our CEO is the sole inventor behind the three patented innovations that make Aewita purpose-built to elevate your practice.

" Every other legal AI tool asks you to trust a subprocessor. Aewita doesn't have one. — Sam Anderson, CEO & Founder of Aewita
Read the full story →
FAQ

Answers to what partners ask first.

What is Aewita?+

Aewita (from Old English ǣwita, "counselor of the law") is a legal AI platform built on a self-hosted frontier reasoning model. Research, drafting, playbooks, and DMS integrations unified under one login. No API calls to OpenAI, Anthropic, or any external LLM.

How is Aewita different from Harvey AI?+

Harvey runs on OpenAI. Aewita runs on its own self-hosted model — no third-party LLM is involved. Your clients' privileged information never touches an external AI provider. Transparent pricing ($99/month, no seat minimum) vs. Harvey's enterprise-only contracts.

Does Aewita train on my data?+

No. Never. Not as a policy preference — as a build constraint.

Is Aewita really hallucination-free?+

All LLMs hallucinate at some rate. Industry rates range from 5–35%+. In controlled experiments, Aewita's infrastructure has never produced a hallucinated output. Statistically we can state with 95% certainty that our rate is below 0.3%.

How much does Aewita cost?+

$20/mo (or $180/yr) for students and faculty. $99/mo (or $720/yr) for attorneys. 14 days free. We charge on day 15 unless you cancel. No seat minimums. No setup fees.

What legal sources does Aewita cover?+

All U.S. case law and court opinions from 1665 to today, plus all current federal and state statutes across 50 states and D.C. The platform supports 792 document types across 22 practice areas.

Ready?

Fourteen days free. Get started in minutes. See why 100% of firms who saw Aewita before launch signed up.

Try it free for 14 days Request a demo