At BoxWorks 2026 on November 5, Box announced an expanded NVIDIA collaboration: Box Context Retrieval, a new engine built on NVIDIA Nemotron Retriever embedding models, combines keyword and semantic search to return the most relevant passages to AI agents while honoring existing file permissions, with general availability for Enterprise Advanced customers planned for early next year. Box's press release claims 20% better search accuracy (Recall@5), 90% fewer tokens and 4x faster answers. Open-weight Nemotron models, starting with Nemotron 3 Ultra, are available in Box AI now. Box also unveiled Box Architect, which designs workflows from a stated goal; Box Mount, which mounts Box folders inside agent sandboxes and works with NVIDIA OpenShell; and agent security controls including MCP Guardrails.
Key Takeaways
- ✓Box Context Retrieval uses NVIDIA Nemotron Retriever with hybrid keyword and semantic search, honoring existing file permissions
- ✓Vendor claims: +20% Recall@5 accuracy, 90% fewer tokens, 4x faster answers; GA for Enterprise Advanced early next year
- ✓Open-weight Nemotron models, starting with Nemotron 3 Ultra, are available in Box AI now
- ✓Box Mount puts Box folders inside agent sandboxes and works with the NVIDIA OpenShell secure runtime
- ✓New agent security controls include MCP Guardrails, classification-based access policies and audit trails
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Core Background & Industry Pain Points
Most enterprise knowledge sits in unstructured files such as contracts, proposals and meeting notes. Agents either pull the wrong file, burn tokens reading whole documents, or risk seeing content they should not. At BoxWorks 2026 on November 5, Box made trusted context for agents its main theme and announced an expanded collaboration with NVIDIA.
Architecture Highlights & Internals
Box Context Retrieval indexes files with NVIDIA Nemotron Retriever embedding models and combines exact keyword matching with semantic search, returning the passages most relevant to a task across scanned documents, images, audio and video while keeping Box permissions. Box Mount puts Box folders inside agent sandboxes so agents can read and write with normal file and shell commands and sync back to Box, and it works with the NVIDIA OpenShell runtime. Box Architect designs and builds workflows from a described goal.
Authoritative Benchmarks & Measured Scores
Box's press release claims 20% better retrieval accuracy (Recall@5), 90% fewer tokens and 4x faster answers. These are vendor figures with no published eval details or third-party replication.
Developer Hands-on Guide
Nemotron models such as Nemotron 3 Ultra and the agent security controls are available now for Enterprise Advanced and Shield customers. Context Retrieval is planned for GA early next year, and Box Mount is marked coming soon. The new MCP Guardrails limit which tools and actions external agents may use, so teams connecting MCP clients to Box should plan permission policies early.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.