An on-device alternative to cloud legal AI.

They are capable and well engineered, and this page is not an argument against them.

• Spellbook. Around $149–$199 per seat per month for the Professional and Enterprise tiers, with Enterprise starting at 10 seats. A 5-lawyer firm: roughly $5,000–$12,000/year.
• Harvey. Indicative pricing in the range of $40,000/year for ten users; some published estimates put the per-lawyer cost at $1,000–$1,200/month with twenty-seat minimums. Enterprise contracts only.
• Legora. Around $3,000 per user per year, ten-seat minimum, $30,000 annual contract minimum.
• CoCounsel. Roughly $225+ per seat per month, sold through Thomson Reuters’ sales team.
• Muet. $390 per device per year: one licence and one annual invoice per device, with no minimum, contract, or procurement.

• Spellbook. Firm-wide playbooks, clause libraries, and high-volume contract drafting inside Word. If the firm drafts hundreds of NDAs a month, this is the tool.
• Harvey. Multi-document review at scale (Vault handles 100,000-document repositories), configurable workflows, and the SSO/audit-log/IP-allowlist set an AmLaw IT team expects.
• Legora. Structured tabular extraction across thousands of documents at once. M&A diligence and portfolio review.
• CoCounsel. Integrated legal research inside the Westlaw ecosystem. If the firm already lives in Westlaw, the integration is real value.

Muet doesn’t try to be any of these. It is a different tool for a different shape of work.

• Pre-engagement pitches and rainmaking correspondence before a matter is opened.
• Off-the-record calls and internal notes the partner is working through.
• Internal investigations the partner is personally running.
• Sensitive HR and partner-track matters they don’t want associates touching.
• Recruitment correspondence and lateral-hire negotiations.
• Equity, succession, and partnership disputes.
• Board work for outside organisations the partner sits on.
• Communications with their own counsel.

Today this work is either pasted into a personal ChatGPT account or done without AI. Muet fits there, on the partner’s laptop beside the firm tools.

• No third party sees the document. Processing happens on the lawyer’s Mac, with no vendor in the loop for the work itself.
• No DPA, no sub-processor list. With no processor relationship there is nothing to negotiate, which helps when a corporate client’s security team would otherwise need weeks to clear a new vendor.
• Works offline. On a plane, in a tunnel, in a regulated facility, or on weak Wi-Fi, the work continues.
• Costs $390 per Mac per year. A one-line software expense, with no procurement cycle or seat negotiations.
• Sits next to whatever else you use. Most lawyers run several tools, and Muet handles the work the others can’t.

• Solo and 2–9 lawyer firms. Often Muet alone, because the price-to-privacy ratio is hard to beat and the firm doesn’t have the deal volume to justify Spellbook or the procurement weight to deploy Harvey.
• Mid-size firms (10–100 lawyers). Often Muet alongside Spellbook, with Spellbook for contract drafting and Muet for the partner work the matter system never touches.
• AmLaw and large in-house teams. Muet on partner laptops alongside the existing Harvey, Legora, or CoCounsel deployment, for the personal-laptop work the firm tools never see.
• Civil-law EU practices. Often Muet as the only AI tool, because secret professionnel and §203 StGB make cloud-based options legally awkward in a way they aren’t in the US.
• In-house counsel. Muet for investigations, NDAs, and pre-deal work, which the company’s cloud tool is rarely cleared for.