Most legal professionals who have used AI tools know the friction point: the analysis is done, but the result is sitting in the wrong window. It needs to be copied, pasted, reformatted, and repositioned inside the actual contract - manually. That is exactly the problem CASUS, a Swiss legal AI platform, is designed to solve. The principle is straightforward: CASUS in the document, not in the tool.
A Live Demonstration: IusBubble Webinar #29
In IusBubble Webinar #29, Fabian and Celeste from CASUS demonstrated the approach live, hosted by Daniel Brugger. The core message was clear: the difference is not in what the AI analyzes, but in what happens next.
Fabian showed the Word add-in: the Agent Mode executing a complex instruction like "make the closing deliveries more seller-friendly," with markup appearing directly in the document - semantically correct, correctly formatted, without any manual follow-up. Celeste demonstrated the same approach in the web app: Proofread across a 50-page SPA, Risk and Quality Review with party-specific perspective, and the AI Data Room for parallel analysis of many documents.
Both surfaces - Word add-in and web app - offer the same functionality. Teams working in SharePoint use the add-in; teams preferring a browser-based approach download documents as DOCX files with tracked changes.
Why the Location of AI Output Matters
Legal AI only creates value when its output becomes part of the work product. Analysis that stays inside a chat window or a separate interface does not improve a contract - it creates an additional step between insight and action.
That step is not trivial. Copy-pasting AI suggestions into Word breaks formatting. Numbering goes wrong. Defined terms get reproduced inconsistently. Cross-references drift. Every manual transfer introduces errors and costs time.
CASUS is built around the principle that AI output should land directly in the document - in Microsoft Word via an add-in, or in the web app where documents are returned as DOCX files with tracked markup changes. No intermediate steps. No formatting corrections.
The Swiss Legal Context
Swiss law firms and in-house legal teams work with high document volumes under time pressure. Contract negotiations, due diligence, compliance reviews - these all happen predominantly in Word documents, stored in SharePoint or on local servers.
At the same time, Swiss legal teams face specific constraints. Data protection requirements under the revised nFADP, confidentiality obligations, and internal IT governance often make it impossible to use tools that transfer data to the US - regardless of how capable those tools might be.
CASUS is hosted exclusively in Switzerland and the EU. There is no data transfer to the US, zero data retention, and no human review of content (abuse monitoring opt-out available). These characteristics reduce the internal approval cycles that typically slow down tool adoption at IT, data protection, and risk teams.
Three Requirements for a Productive In-Document AI Workflow
For AI assistance to genuinely integrate into legal practice, three conditions need to be met.
Placement – changes must be inserted at the correct location in the document. Not somewhere approximate, but semantically correct: in the right clause, in the right section, in the right position relative to surrounding provisions.
Structure – numbering, heading hierarchy, formatting styles, and document logic must remain intact. An AI suggestion that breaks numbering or incorrectly reproduces a defined term creates more work than it saves.
Traceability – lawyers need to understand why a suggestion was made. Answers and changes must be linked to the relevant passages in the source document, so every AI output can be verified before it is accepted.
These three requirements define what "CASUS in the document, not in the tool" means in practice.
How CASUS Works Inside Microsoft Word
The CASUS Word add-in stays open while a document is being edited. It provides access to all core workflows without switching applications or windows.
AI Chat with Agent Mode
The AI Chat answers questions about the document - for example, whether a liability cap exists, what the notice period is, or what the cost implications are over the full contract term. Answers are linked to the relevant passages in the document and clickable, so every AI output can be verified against the source text.
In Agent Mode, tasks can be delegated directly. A user can instruct CASUS to "increase the liability cap to CHF 50,000" or "make the closing deliveries more seller-friendly." CASUS executes these changes as tracked markup in the document - in the correct language, with correct formatting, placed at the right location. The chat explains which contract sections were affected and why.
CASUS works semantically, meaning defined terms are reproduced correctly and changes respect the existing document structure. The markup appears under the user's name and can be accepted, rejected, or further adjusted.
Benchmark
The Benchmark workflow compares a contract against an internal playbook or standard - for example, a buyer-friendly SPA standard or an NDA template. CASUS surfaces deviations, missing clauses, and incomplete provisions as structured findings. Each finding comes with a recommendation, and clauses can be inserted directly into the document at the right location, with the right formatting.
The output also includes a percentage score showing how closely the document aligns with the reference standard - for example, "85% aligned with standard." This makes internal communication and prioritization straightforward.
Risk and Quality Review
The Risk and Quality Review analyzes risks from the perspective of a specific party. CASUS identifies the contracting parties and assesses risks from each party's viewpoint - not in a generic way. Findings are prioritized by severity (low, medium, high) and come with concrete drafting suggestions. Each suggestion can be applied directly in Word, without copy-paste.
Proofread
The Proofread workflow checks language and formal consistency across the entire document - spelling, grammar, Swiss spelling conventions (ss instead of ß), inconsistent terminology, incorrect cross-references, missing definitions, and open placeholders. On a 50-page SPA, this kind of review catches issues that are difficult to find manually: for example, a data room folder referenced as "Financial Information" that actually contains HR data, based on the annex index.
The Proofread module is not a legal review of the substantive legal situation - it is a quality and consistency check of the document before it goes out.
Practical Implications for Law Firms and Legal Departments
Efficiency comes from eliminating transfer steps
The largest efficiency gain is not the analysis itself - it is the elimination of manual steps between analysis and implementation. No copy-paste, no formatting correction, no manual clause positioning. This matters especially at high document volumes, where small inefficiencies accumulate.
Quality becomes reproducible
When benchmark checks, risk reviews, and proofreads consistently follow the same standards, quality becomes reproducible across a team. Internal alignment is simpler when a document can be described as "85% aligned with the standard" - a concrete, traceable basis for negotiations and internal sign-off processes.
Data compliance is built in, not bolted on
Because CASUS hosts in Switzerland and the EU, transfers no data to the US, and offers zero data retention, data compliance is not a barrier to adoption. The typical friction of IT and data protection approval cycles is significantly reduced.
Who benefits most from this approach
In-house legal teams managing high contract volumes and needing consistent, reproducible quality benefit directly. Transaction teams in law firms working under time pressure on SPAs, NDAs, or DPAs benefit from the combination of speed and in-document accuracy. Teams involved in due diligence or compliance processes - where many documents need to be analyzed in parallel - can use the AI Data Room workflow for structured extraction across large document sets.
Try CASUS
CASUS is available to try free of charge, with no commitment. Both the Word add-in and the web app are accessible immediately. Documents can be tested without setup effort and without any data transfer to the US.
FAQ
What does "CASUS in the document, not in the tool" mean?
CASUS delivers AI output directly inside Microsoft Word as tracked markup, rather than in a separate chat window or interface. Clauses are inserted at the correct location, and formatting and numbering are preserved - without manual intermediate steps.
How does the CASUS Word add-in work?
The CASUS Word add-in stays open while editing a document and provides access to AI Chat with Agent Mode, Benchmark, Risk and Quality Review, Proofread, and other workflows - all operating directly within the document.
Can CASUS make changes to a document autonomously?
In Agent Mode, CASUS executes changes on instruction - for example, inserting clauses, adjusting wording, or adding content. Changes appear as tracked markup under the user's name and can be accepted, rejected, or further adjusted.
Where is data stored when using CASUS?
CASUS is hosted exclusively in Switzerland and the EU. There is no data transfer to the US. CASUS offers zero data retention and no human review of content (abuse monitoring opt-out available).
What is the difference between CASUS and other legal AI tools?
Many legal AI tools deliver results in a separate interface that requires manual transfer into the working document. CASUS works directly in the Word document or returns documents as DOCX files with tracked markup changes - semantically correct, correctly formatted, and placed at the right location.
Which document types does CASUS support?
CASUS is designed for contract documents - SPAs, NDAs, DPAs, and other standard contract types. It supports both single-document workflows (Risk Review, Benchmark, Proofread, Chat) and parallel analysis of large document sets via the AI Data Room.
Is CASUS usable without the Word add-in?
Yes. CASUS offers a web app with equivalent functionality. Documents are uploaded, analyzed, and downloaded as DOCX files with tracked markup changes. The web app is suited to teams that prefer a browser-based workflow.
Is CASUS a Swiss alternative to tools like Harvey or Legora?
CASUS is positioned as a Swiss alternative to tools such as Harvey, Legora, and Spellbook. The key differentiators are Swiss and EU data residency, in-document execution of AI actions, and a focus on the specific needs of Swiss law firms and in-house legal teams.
