Robin AI and Magic Circle: A Side-by-Side for Cross-Border Legal Teams

Robin AI and Magic Circle compete in roughly the same product category — agent-driven legal AI focused on contract review — but they were built around different buyer profiles. Comparing them directly is useful for firms that are evaluating both, but the comparison is less about feature parity and more about which buyer profile the firm fits into.

This analysis examines how the two tools approach the work, where they overlap meaningfully, and where the differences in geographic focus and deployment model become decisive.

Background

Robin AI was founded in London and grew up serving English-language contract review at mid-sized law firms and in-house teams across the United Kingdom, Europe, and parts of North America. The product is sharply focused: it does contract review and clause extraction well, with playbook automation and clause library features that mature firms quickly come to rely on. Pricing is structured to be accessible relative to Harvey, and the cloud-only model fits the typical English-language buyer well.

Magic Circle was built for a different center of gravity. The starting point was the practical reality of Vietnamese and Southeast Asian legal work: bilingual matters, stamped scanned documents, regulatory frameworks rooted in civil-code traditions, and deployment requirements driven by data-residency rules and the security policies of large institutional clients. The product covers a broader workflow scope than Robin — including diligence summarization and regulatory comparison alongside contract review — and ships in four deployment models including on-premise.

Where the products genuinely overlap

For English-language contract review on cloud-hosted matter files, the two products do similar work. Both ingest contracts, both run agents that read across the document, both extract structured clauses, and both produce citations linked back to the source text. Both support tenant isolation, both offer API access at higher tiers, and both can integrate with the major document management systems. A buyer evaluating only this slice of functionality, with no Vietnamese documents in scope and no deployment constraints, will find the two products roughly comparable on capability and could reasonably select either based on commercial terms and team chemistry.

This is the case in which Robin is often the more natural choice. The product is tightly focused on contract review, the user interface assumes English-language work, and the pricing has historically been positioned below Harvey while above smaller competitors. For a London-based mid-firm or a New York in-house team doing English contracts, Robin tends to win on alignment with the existing workflow.

Where the products diverge meaningfully

The picture changes once any of three factors enter the evaluation: Vietnamese documents, bilingual matters, or deployment constraints.

On Vietnamese documents, the gap is substantial. Robin's OCR was trained primarily on English text in clean layouts. Stamped scans, Vietnamese diacritics on lower-resolution copies, and bilingual two-column layouts produce systematic errors that propagate up to the citation layer. A user can still get answers, but the citations may point to extracted text that does not accurately reflect what is actually printed on the page. For a Vietnamese law firm whose median contract has these features, this is not a minor degradation — it is the difference between a tool that can be used in real work and a tool that cannot.

On bilingual matters, the difference is structural. Cross-border practice in Southeast Asia routinely involves matters where the same agreement exists in Vietnamese and English versions, where Vietnamese statutory references appear within English-language commentary, or where one matter file contains exhibits in both languages. Magic Circle reasons across the two languages within a single matter as a native capability. Robin's model is single-language, requiring either splitting the matter or pre-translating documents before processing.

On deployment, the difference is binary. Robin is cloud-only. Magic Circle ships in managed cloud, customer-controlled cloud tenant, on-premise within the firm's data center, and air-gapped configurations. For any firm whose client mix includes parties subject to data residency under NĐ-13/2023, or whose CISO has veto rights on cloud AI vendors, the cloud-only constraint of Robin is dispositive — the evaluation ends before it begins.

Pricing comparison

Both vendors decline to publish full pricing, so any comparison rests on industry reports and observed tier structures. Robin's published starting tiers fall roughly in the thirty to one hundred US dollars per user per month range, with enterprise agreements negotiated above that. Magic Circle's published Pro tier in international pricing sits at seventy-nine US dollars per user per month with annual discount, the Business tier is sales-led, and the Enterprise tier — the only configuration that includes on-premise deployment — is priced as a base platform fee plus per-seat.

For Vietnamese-billed firms, Magic Circle pricing is denominated in Vietnamese Dong, with per-seat rates calibrated to the Vietnamese market rather than transplanted from US benchmarks. Robin's published pricing is in pounds and dollars, and at typical exchange rates falls above what most Vietnamese mid-firms are accustomed to spending per seat on legal software.

The pricing comparison is most meaningful when held alongside the workflow comparison. If the firm's work is genuinely English-only and cloud-acceptable, Robin's pricing is competitive. If the work includes Vietnamese matters or requires deployment beyond cloud, the cheaper headline price of Robin becomes irrelevant — the product cannot do the work that needs doing.

Support and onboarding

Robin operates support and customer success out of the United Kingdom and the United States, in English, on London and Eastern Time business hours. This works well for firms aligned with those time zones and language preferences, and friction increases for teams operating outside them. Magic Circle provides support in Vietnamese and English on Southeast Asia business hours, with dedicated customer success management at Enterprise tier.

For firms doing implementation work — onboarding lawyers, training paralegals, configuring playbooks — the difference in language and time zone alignment is more meaningful than it appears in vendor demos. It affects the speed of pilot ramp, the quality of escalation handling, and the firm's ability to retain operational knowledge after the vendor's initial training engagement.

How to evaluate both fairly

The most reliable way to compare these products is to run them in parallel on a real matter for a defined period — typically two to four weeks — with success criteria established before the evaluation begins. The criteria should include citation accuracy measured against the firm's actual document distribution rather than against vendor-curated samples, OCR quality on stamped Vietnamese scans (if any are in scope), handling of bilingual content, total time from matter setup to first usable answer, and reviewer trust as expressed by senior associates after using the tool for one week.

In our experience, evaluations conducted this way produce clearer results than evaluations based on demo days. The vendor demo controls for everything that makes the buyer's actual documents difficult; the parallel pilot reveals what each tool does on the documents the firm actually needs to process.

Choosing between them

The simplest way to express the choice: if the firm's work is English-only, cloud-acceptable, and contract-review-focused, Robin AI is a credible default and may be the better fit on workflow alignment. If the work includes Vietnamese documents at any meaningful volume, bilingual cross-border matters, or any deployment requirement beyond cloud, Magic Circle is built specifically for that scope and Robin will require workarounds that compromise the value the firm is paying for.

Book a Magic Circle comparison demo on your own documents →