Harvey AI has overhauled the upload system for its Vault product, switching to presigned URLs that cut transfer times by up to 57% for large document batches. The change, aimed at legal teams that regularly move thousands of files, also improves reliability by eliminating common failure points in traditional upload methods.
How presigned URLs speed up the process
Presigned URLs generate a temporary, secure link that allows a client to upload a file directly to cloud storage without routing through the application server. For Harvey AI's Vault, this means large batches—often hundreds of documents at once—no longer bottleneck at a single server. Instead, each file gets its own direct path, and the system can handle multiple uploads in parallel. The result: a 57% reduction in total upload time for the heaviest batches, according to the company.
Why reliability matters for legal work
Legal professionals using Vault to store and organize case materials frequently deal with time-sensitive filings and massive discovery documents. A failed upload mid-batch can mean re-uploading everything, wasting hours. By using presigned URLs, Harvey AI reduces the chance of a single point of failure. If one file's upload stalls, the rest continue unaffected. The system also retries failed files automatically, something the previous architecture could not do efficiently.
What the update means for existing users
The revamp is live now for all Vault customers. No additional configuration is needed—the change happens on the backend. Users should notice faster completion times for large uploads, especially when working with PDFs, scanned images, or other common legal file types. Harvey AI has not disclosed whether the same technology will roll out to other parts of its platform, but the company continues to invest in infrastructure improvements for its AI-powered legal tools.
