Helping devs build the future of intelligent, content-driven apps using @Box. Check out our docs and sign-up for a free developer account on https://nitter.gq/t.co/oDLXCnem9S
Joined May 2008
- Tweets13.7K
- Following3.1K
- Followers80.4K
- Likes4.9K
An agent can find your contract. It might still be the wrong version.
@levie on why fragmented enterprise content makes reliable AI harder, and why knowing which information to use matters as much as access.
More from BoxWorks: blog.box.com/boxworks-2026-b…
Prepare content.
Keep indexes current.
Combine keyword and semantic retrieval.
Return relevant passages.
Our CTO Ben Kus explains how Box Context Retrieval will give agents focused business context while respecting permissions.
blog.box.com/box-partners-nv…
🆕@AnthropicAI's Claude Fable 5.2 checks the data before it trusts it. We ran it against Claude Fable 5.1 on real enterprise document tasks: pull the source files, reason across them, produce the actual deliverable.
→ Public sector: 92% vs 81% (+11 points)
→ Energy: 93% vs 85% (+8 points)
→ Consumer products: 95% vs 91% (+4 points)
→ Speed: about 10% faster
It catches the timestamp that's off by a fraction of a second, applies the correction filed after the report date, and reconciles conflicting figures before building on them.
Read the full evaluation here. 👇
blog.box.com/claude-fable-52…
Box retweeted
use muse ember to save millions
pay $0.10 / $0.50 / Mtok and save millions of dollars on legal clerical errors
We put Muse Ember from @AIatMeta to work untangling conflicting records in a $12.6 million private-fund capital call.
Across 13 documents in Box, Ember catches a $2.48 million misallocation and a missed $250,000 credit, recalculates each investor’s obligation, and produces a source-cited review for Fund Operations.
Customers can build custom agents for their own document-heavy financial workflows with Box AI Studio.
We put Muse Ember from @AIatMeta to work untangling conflicting records in a $12.6 million private-fund capital call.
Across 13 documents in Box, Ember catches a $2.48 million misallocation and a missed $250,000 credit, recalculates each investor’s obligation, and produces a source-cited review for Fund Operations.
Customers can build custom agents for their own document-heavy financial workflows with Box AI Studio.
Today we're introducing Muse Ember, our smallest and most efficient model, built for complex reasoning execution and everyday agent work.
Muse Ember can carry workflows across apps, call tools in sequence, run tests and make targeted code fixes. With a 1M-token context window and native understanding of text, images, audio and video, it can also work across large document sets and varied inputs.
A file’s responsibilities do not end when the work is finished.
At BoxWorks, Vaish Krishnamurthy explains how retention, legal holds and disposition fit alongside AI workflows and security in Box.
Learn more: blog.box.com/box-announces-n…
We were thrilled to have @AnthropicAI join us on stage with @BTSGroupAB yesterday at BoxWorks!
We’re deepening our partnership to expand how you work with Box content in @claudeai.
Coming soon: Search Box before prompting, @ mention files for context, and save Claude-created docs back to Box.
Demo 👇
✔️ Trusted content in Box.
✔️ Claude Design and Claude Skills on top.
At BoxWorks, @BTSGroupAB's Kevin Bronk explained the BTS brain, and says some RFP responses that took weeks now take days or hours.
More from BoxWorks: blog.box.com/boxworks-2026-b…
Vendor submissions → AI risk tiers → human review for higher-risk cases.
@SamsungDSGlobal's Kevin Shin explains the Box Automate workflow his team built without complex custom code or managing servers.
blog.box.com/boxworks-2026-b…
Build on Box without requiring the Box UI.
APIs, SDKs and the Box MCP Server connect apps and agents to enterprise content, with existing permissions and audit trails.
@FernandoCerenza explains the headless platform.
blog.box.com/boxworks-2026-b…
Box Agent understands content.
Box Extract structures it.
Box Automate puts it to work.
Box Apps brings it into view.
Nirmal Ganesh shares how Box Architect will build on that foundation to reimagine workflows.
Learn more from our announcement: blog.box.com/box-unveils-arc…
The Box Impact Fund is accepting applications for digital innovation in child welfare, crisis response and environmental protection.
6 nonprofits. $25,000 each. Apply by November 30.
box.org/impact-fund
Enterprise content is the critical context for agents. And they will create and consume 100X more unstructured data than today.
@JensenHuang: “every time the business process executes, it generates more data. It goes back into Box. It becomes the memory of the company."
Introducing Box for Industries: banking, public sector, and media & entertainment.
Our COO Olivia Nottebohm explains how industry-specific workflows and metadata templates bring Box AI, security and governance to the way your industry works.
Learn more: blog.box.com/box-unveils-all…
Box retweeted
Marketing has two audiences now: people and AI agents
Our CMO, Erin Oles, and @Box CMO Tricia Gellman use Palmata from @Contentful to see how AI reads their brands — what customers are asking, what content is missing, and where perception is shifting
See the tools and strategies they’re using: sforce.co/4A7aHp0
Box retweeted
Congrats to our friends at @Box on announcing Box Mount.
On Managed Agents, Box Mount puts a team’s Box folder in the agent’s sandbox. Edits sync back with Box permissions and version history in place, and you approve key steps from Cuttlefish.
Box retweeted
BREAKING!!!! Every AI keynote talks about models. BoxWorks 2026 talked about context. I was in the front row in San Francisco today as Aaron Levie opened the keynote. And yes Jensen Huang is here as well. Here is everything @Box announced, and my take –-
- Box Context Retrieval, powered by NVIDIA- A new engine built on NVIDIA Nemotron Retriever. It turns your contracts, policies and files, including scans, images, audio and video, into a knowledge layer agents can reason over. Box says it can deliver up to 20% better search accuracy, 90% fewer tokens and 4x faster answers. All permission-aware
- NVIDIA open weight models in Box AI, including Nemotron 3 Ultra- More model choice for high-volume work, with content staying in Box
- Box Architect- Describe a process in plain language. Architect designs the agentic workflow, lets you test it on real content, add human approvals and pick a model per step. Then you deploy
- Box for Industries- Starting with Banking, Public Sector and Media & Entertainment. The Norinchukin Bank cut contract review time by nearly 88%. HBO and the Texas DMV are on board too
- Security for agents- Agent security controls are live today: classification policies, MCP Guardrails and activity thresholds. New threat protection against bulk exfiltration and account takeover coming soon
My take: Context is the real bottleneck. AI pilots stall because agents don't know the business. Box securely holds the files that contain that knowledge. Permission-aware retrieval is the quiet win. It's what lets banks and governments say yes to agents. Box Architect is the one I'll watch closely. If business teams can really build agentic workflows in plain language, that changes who builds with AI.
I also sat down with Ben Kus, Heather Ceylan, Olivia Nottebohm and Vaibhav Shinde from Deloitte. Interviews drop soon on The Ravit Show.
What would you build first with @Box Architect?
#boxworks #boxpartner #boxai #enterpriseai #aiagents #futureofwork #data #theravitshow
