# CORA Group (CSI)

Cora Group is a Constellation Software Inc. (CSI) portfolio company — Operating Group: Jonas Software

**CORA Group partners with raia to build a custom AI agent for NDA review and redlining -accelerating early-stage deal execution by automating first-pass legal review and ensuring consistent application of CORA’s legal playbook.** *** ### Snapshot (Planned Targets) | Metric | Before raia | After raia (Target) | | ------------------- | -------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | | **NDA Review Time** | Hours or days per document | Faster first-pass NDA turnaround via AI-generated redlined draft *(targets set after baseline review)* | | **Consistency** | Varies by reviewer and workload | Standardized to CORA’s legal playbook and historical precedent | | **Manual Effort** | High; manual clause-by-clause review and redlining | Lower; team shifts from drafting to reviewing AI-generated redlines | | **Risk Control** | Dependent on individual knowledge of preferred positions | Uniform application of approved fallback language and risk tolerances | | **Scalability** | Limited by legal team bandwidth during peak deal flow | Supports higher throughput without proportional legal headcount growth | > **Note:** This case study describes the **planned solution and target outcomes**. Measured results can be added once CORA’s agent is deployed and benchmarked. ### 1 | Company & Challenge NDAs are a frequent, time-sensitive first step in CORA Group’s deal lifecycle. The existing manual review process can create bottlenecks—slowing early-stage execution and consuming valuable legal team capacity. As CORA evaluated how to accelerate this workflow, three core challenges emerged: 1. **Slow, manual review cycles:** The legal and M\&A teams spend significant time on repetitive, clause-by-clause review, cross-referencing internal playbooks, and drafting redlines. 2. **Inconsistent application of standards:** Under time pressure and variable deal volume, ensuring every NDA reflects CORA’s latest preferred legal positions and risk tolerances is difficult. 3. **Low-value use of expert time:** Highly skilled legal professionals are pulled into routine drafting work instead of focusing on exceptions, negotiation strategy, and complex legal matters. CORA needed a solution that could automate the first-pass review, enforce consistency, and free up its legal team to focus on more strategic work—without introducing unnecessary technical or compliance risk. ### 2 | Why CORA Group chose raia CORA selected raia to build a sophisticated AI legal assistant that balances speed with the critical need for human oversight: * **Human-in-the-loop safety:** The solution is designed as a document analysis and redlining agent, not a fully autonomous editor. The AI generates a revised draft for human review, ensuring a legal professional makes the final decision. * **Playbook-driven consistency:** raia trains the agent on CORA’s NDA playbook, preferred clauses, fallback language, and historical precedent so the output aligns with internal standards. * **No risky system integrations:** The agent operates on document inputs (Word, PDF) and produces a revised document as output—without direct integrations into Microsoft Word or live collaboration platforms. * **Foundation for future M\&A automation:** This initial project creates a reusable legal knowledge framework that can be expanded to other agreement types, supporting CORA’s broader M\&A enablement interests. ### 3 | Solution Design The solution is an AI-powered legal co-pilot, trained on CORA’s proprietary data, that transforms the NDA review process from manual drafting to automated analysis and supervised revision. | Component | Role | Key Knowledge Sources | | -------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------- | | **NDA Review & Redlining Agent** | Ingests an NDA (Word/PDF), analyzes it against CORA’s playbook, and generates a revised draft with recommended, **clearly annotated** redlines and preferred replacement language. | CORA M\&A NDA playbook; previously executed NDAs (both clean and revised versions) | | **Clause Taxonomy & Knowledge Architecture** | A structured framework mapping clauses to CORA’s preferred positions, fallback language, and risk tolerances—forming the agent’s decision logic. | CORA legal standards; historical redlining patterns | | **Secure Document Processing Workflow** | A controlled, internal workflow for document intake and output within the raia platform, with version traceability and access controls. | Internal raia platform | #### Illustrative redline focus areas (examples) To increase consistency and reduce back-and-forth, the agent focuses on common NDA pressure points such as: * scope of “Confidential Information” and permitted disclosures * term and survival provisions * return/destruction obligations * residuals / use restrictions * injunctive relief and remedies language * assignment and change-of-control language * governing law / venue (where applicable) *(Final scope and preferred language are driven by CORA’s playbook and precedent.)* #### Guardrails (by Design) * **Human-in-the-loop approval:** The AI only generates drafts; a CORA team member must review and approve changes before they are shared externally. * **Conservative legal behavior:** The agent adheres strictly to the playbook and historical precedent—avoiding novel legal interpretations. * **No direct editing:** The agent does not have live access to edit documents in place. It produces a new, redlined version for review, ensuring a clear audit trail. * **Internal & controlled:** Processing remains within a secure environment with no external system integrations, minimizing compliance and security risks. ### 4 | Deployment Timeline (Phased) 1. **Phase 1 — AI Blueprinting:** Develop the core knowledge model by defining the NDA clause taxonomy, playbook-to-clause mapping, and governance rules. 2. **Phase 2 — Train AI Agent:** Train the agent on historical NDAs to learn clause-level reasoning and the correct application of preferred language (including when “no change” is appropriate). 3. **Phase 3 — Integration:** Configure the secure document intake, processing, and output workflow within the raia platform. 4. **Phase 4 — Test Agent:** Backtest the agent against a corpus of previously reviewed NDAs to validate accuracy, consistency, and edge-case handling. 5. **Phase 5 — Launch:** Deploy the production-ready agent for the M\&A and legal teams, including user training and feedback collection. ### 5 | Expected Impact (Targets) CORA’s target outcomes for the engagement include: * **Accelerated deal velocity:** Reduce time spent on initial NDA review so the M\&A team can engage counterparties faster. * **Improved consistency & risk control:** Ensure every NDA review uniformly applies CORA’s approved legal positions and risk tolerances. * **Reduced legal team load:** Free up the legal team from repetitive drafting to focus on complex and strategic M\&A work. * **Scalable M\&A operations:** Handle increasing deal flow without requiring a proportional increase in legal headcount. ### 6 | Lessons for Prospective Clients 1. **For legal work, human-in-the-loop is a feature:** The safest path to value is using AI to assist—not replace—human experts. 2. **Your playbook is your moat:** The AI is only as good as the knowledge it’s trained on. A clear playbook and good precedent examples are the critical inputs. 3. **Avoid unnecessary technical risk:** Live-editing integrations can introduce security and compliance risk. A document-in, document-out workflow is safer and faster to deploy. 4. **Backtest before rollout:** Comparing AI drafts to historical finals creates a clear quality bar and reveals edge cases early. ### 7 | What’s Next Following deployment of the NDA Review & Redlining Agent, raia and CORA Group can expand the reusable legal knowledge foundation to additional M\&A document types (e.g., LOIs, MSAs) and adjacent M\&A enablement workflows. ### 8 | Key Takeaway CORA Group is building a **secure, human-in-the-loop AI agent** to automate first-pass NDA review and redlining. By training an AI on its internal legal playbook and historical data, CORA is creating a scalable system to **accelerate deal cycles, enforce consistency, and reduce manual legal effort**—establishing a reusable foundation for broader M\&A automation.