What problem does Sonar solve?
Entity management has traditionally required manually extracting information from legal documents and organizing the data into a structured format. This process is time-consuming, error-prone, and repetitive. Sonar is our answer to this challenge.
What is Sonar?
Sonar is an AI-powered tool that automatically extracts data from legal documents and ownership charts, so you don’t have to. Upload your documents into ORCA, and Sonar will create and update Entity data, build your ownership structure, and file your documents. It transforms you from a "doer" to a "checker," saving you time and ensuring data accuracy.
Getting Started
How do I enable Sonar?
Sonar is enabled in Account Settings → AI features. See more information here. Once turned on, it works across all your Vaults and is available to all users.
Note: Sonar is currently in beta.
What information does Sonar extract?
Sonar analyses the content of documents to extract key legal and corporate information. This information currently includes:
- Ownership/Corporate structures and Transactions: purchases, ownership stakes, transfers, and share classes.
- Trust Roles & Details: Trust-related roles and attributes, such as trustees, settlors, protectors, and trust classifications.
- Executive & Governing Roles: Leadership and governance positions, such as directors, board members, managers, and executives.
- Addresses & Countries: Addresses, tax domiciles, and country information.
- Corporate Information: Formation and registration documents, such as articles of association and certificates of incorporation.
- Legal Powers: Legal authorities and signing rights, such as powers of attorney and signatory rights.
- Identifications & Permits: Identification documents and permits, such as passports, national IDs, visas, and work permits.
- Certifications & Documentation: Legal and supporting documents, such as certificates, deeds, operating agreements, and registrations.
- Loans.
What documents should I upload?
Sonar isn't limited to specific document types. If your document contains relevant information, Sonar will attempt to extract it.
Examples of document types:
- Ownership Charts (from PowerPoint, Visio, or napkin sketches; just export or capture it as a PDF or image before uploading).
- Cap Tables.
- Purchase Agreements.
- Trust Agreements.
- Limited Partnership Agreements.
- And many more.
Uploading & Processing
Which browsers are supported by Sonar?
Sonar requires Chrome or Edge.
What file formats work?
PDFs and images (e.g., JPEG and PNG). If your document is in another format, you can still upload it by either exporting it to PDF or taking a screenshot of the content and uploading that. All other file types can also be uploaded using our standard upload process.
Which languages does Sonar support?
Temporarily, Sonar supports English-language documents only. We expect to restore full language support within the coming weeks.
Is there a file size limit?
Yes, 10 MB per file, due to OpenAI’s processing constraints. For larger files, consider splitting them and uploading each file separately.
How many files can I upload at once?
Up to 10 at a time. You can repeat the upload process as many times as needed.
How long does processing take?
Processing takes about 1-2 minutes per file, but it varies depending on the file size, how clear the scan is, and the current system load. You’ll see a “processing” indicator while Sonar works. When your files are ready for review, a small notification bubble will appear on the button “Sonar - Files Upload”, showing the number of processed documents waiting for you.
Do I have to stay logged in?
Yes. Keep your browser window open while processing happens; closing the tab or losing your internet connection will stop the process. You can, however, keep working elsewhere in ORCA while Sonar runs in the background.
What happens if I close ORCA before reviewing my files?
Any Sonar results that haven’t been reviewed will be discarded, but the file itself will remain in your Vault. Sonar requires you to confirm results before saving them. Your file will stay in the “Misc” folder in your Vault, and you can manually link it to events or entities later.
Who can see my files and the AI results?
Files are visible to all users in your Vault as soon as they’re uploaded. The AI results (extracted events and entities) are visible only to you until you review and save them. Once saved, they become visible to all users.
AI & Technology
What AI provider or Large Language Model (LLM) is Sonar using?
Sonar is powered by OpenAI Enterprise, under a zero data retention agreement. This means OpenAI doesn’t store your data or use it for training, and all processing occurs within the EU. Sonar currently uses the GPT-5 mini model for analysis. See our Terms & Conditions for more details.
Who can see the content of my files during AI processing?
When you upload a file, its content is temporarily accessed by ORCA and OpenAI servers so that the model can analyze it. This access is secure, and once processing is complete, the file and its data are deleted and no longer visible to the model or ORCA.
Will you use my data to train your models?
No. Neither ORCA nor OpenAI trains on your documents.
Will Sonar learn from my corrections?
No; in order to protect your privacy, we don't train the model on your inputs. However, any feedback you share with our team helps us identify improve Sonar over time. You’ll also automatically benefit from improvements to Sonar and OpenAI’s base models.
How is Sonar different from standard Optical Character Recognition (OCR)?
OCRs convert images to text. Sonar goes deeper: it understands what the document is about and extracts structured information like events and entities.
Do you have a Data Processing Agreement?
Yes. ORCA’s Data Processing Agreement is available here.
Are there any usage limits?
No. There are no upload or token limits at the moment. However, each upload batch is limited to 10 files at a time. This is done to ensure you can review the files while in session.
Troubleshooting
Why didn't I get any results for my document?
Possible reasons include:
- Low-quality or blurry scans.
- Languages other than English.
- The content doesn’t include information Sonar can analyze.
- Temporary processing issues.
Try using the “Rerun Sonar” option in the “Review File” screen. As AI results are non- deterministic, re-running the analysis can yield more or different results.
You can also try uploading a clearer version of the document. If the issue persists, contact support.
My ownership chart didn't extract correctly. Why?
Possible reasons include:
- Low-quality or blurry scans.
- The chart contains more than 25 entities.
- The chart has complex lines or shapes that Sonar may not interpret correctly.
- Temporary processing issues.
Try using the “Rerun Sonar” option in the “Review File” screen. As AI results are non- deterministic, re-running the analysis can yield more or different results.
You can also try uploading a clearer version of the document, or break large charts into smaller sections and upload again,. If the issue persists, contact support.
How Sonar Maps Ownership
Sonar doesn’t just extract dates and names; it understands the "action" within a document and maps it to the correct relationship type in ORCA. This ensures your Structure Chart remains accurate and legally consistent.
When reviewing suggestions, you will notice two primary ways Sonar creates ownership connections:
- "Owns" (Structural Holdings): When Sonar analyzes documents like Structure Charts, it identifies permanent holding patterns. These are added to ORCA using the verb "owns," representing a current state of ownership between a parent and a subsidiary.
- "Bought" (Transaction-Based): When Sonar analyzes a Purchase Agreement, it identifies a specific point-in-time event. These are added using the verb "Bought," which allows ORCA to track the historical transaction and the specific acquisition date.
Find more information about these two verbs here.
Pricing
How much does Sonar cost?
Sonar is free while in beta. Additional charges will apply afterward.
Roadmap
What’s planned for future versions of Sonar?
Upcoming capabilities include:
- Deeper data extraction: Capture more detailed information from each document.
- Automatic entity updates: Keep your data accurate by automatically applying entity details found in documents.
- Analysis of existing files: Apply Sonar’s AI analysis to documents already stored in your Vault or received vis email.
Have feedback or questions?
We'd love to hear from you! Reach us at product@withorca.com.