Box retweeted
Aaron Levie of @Box on how AI agents with wallets could create new business models for the internet:
"I've always wanted micropayments for things online... I've appreciated why it's a bad business model, but I want it."
"Now it's probably a good business model. Let's say you roprietary data. You're Bloomberg. You're a medical institution. You could imagine there's maybe a hundred million agents that would be tapping into that data."
"All of a sudden, a penny a transaction starts to make a lot more sense than it did when you were capped at 20 million [knowledge workers] that maybe were interested in your content."
"We probably have new business models for tools and APIs... We probably want an agent to be able to have a budget and make any API call it wants."
"All of a sudden you have this new revenue stream, which is these agents showing up, just transacting on your system."
@levie at Daytona Compute 2026
"I've always wanted micropayments for things online... I've appreciated why it's a bad business model, but I want it."
"Now it's probably a good business model. Let's say you roprietary data. You're Bloomberg. You're a medical institution. You could imagine there's maybe a hundred million agents that would be tapping into that data."
"All of a sudden, a penny a transaction starts to make a lot more sense than it did when you were capped at 20 million [knowledge workers] that maybe were interested in your content."
"We probably have new business models for tools and APIs... We probably want an agent to be able to have a budget and make any API call it wants."
"All of a sudden you have this new revenue stream, which is these agents showing up, just transacting on your system."
@levie at Daytona Compute 2026
Wallet Wars: Agents need payment credentials as they become the new UX for $7T in ecommerce. Enter the Wallet Wars and why the AOL-era of agentic commerce will give way to open standards. x.com/i/article/2110…
39
32
209
36,236
Box AI Extract pulls structured fields from documents, and @OpenAI Decisions API routes them to the right workflow.
In this demo, a COI arrives in Box through a File Request. Box AI extracts coverage fields from both the upload and the project requirements. The Decisions API and returns a typed route: request endorsement, specialist review, or ready for human review.
A Box webhook triggers the whole process on document upload. The app creates a review task with context, adds the right collaborator, and writes extracted values and decision confidence back to the file as metadata.
Box is where the documents arrive and where people review them.
Watch👇
In this demo, a COI arrives in Box through a File Request. Box AI extracts coverage fields from both the upload and the project requirements. The Decisions API and returns a typed route: request endorsement, specialist review, or ready for human review.
A Box webhook triggers the whole process on document upload. The app creates a review task with context, adds the right collaborator, and writes extracted values and decision confidence back to the file as metadata.
Box is where the documents arrive and where people review them.
Watch👇
4
3
15
179,490
Box retweeted
Counting down to BoxWorks DevSummit! ✨ Dev Zone, technical talks, masterclasses, cool swag and rooftop afterparty. See you there!
At the BoxWorks Developer Summit, join developers, AI engineers, and technical leaders for two days of talks and hands-on workshops exploring how to unlock AI with unstructured data.
Hear what’s next for the Box platform, build with Box APIs, SDKs, and AI agents, and explore se cases.
Speakers include technical leaders from Box, @nvidia, @OpenAI, @getpostman, and @daytonaio.
Registration includes access to both BoxWorks and the Dev Summit: boxworks.box.com/series/boxwork…
0
2
3
294
At the BoxWorks Developer Summit, join developers, AI engineers, and technical leaders for two days of talks and hands-on workshops exploring how to unlock AI with unstructured data.
Hear what’s next for the Box platform, build with Box APIs, SDKs, and AI agents, and explore se cases.
Speakers include technical leaders from Box, @nvidia, @OpenAI, @getpostman, and @daytonaio.
Registration includes access to both BoxWorks and the Dev Summit: boxworks.box.com/series/boxwork…
Hear what’s next for the Box platform, build with Box APIs, SDKs, and AI agents, and explore se cases.
Speakers include technical leaders from Box, @nvidia, @OpenAI, @getpostman, and @daytonaio.
Registration includes access to both BoxWorks and the Dev Summit: boxworks.box.com/series/boxwork…
2
5
13
65,034
Box retweeted
Yesterday, we began rolling out Gemini 4 Argon — our newest frontier model that excels at complex, real-world tasks across coding, finance, legal, cybersecurity, and science — more widely.
We gave early access to enterprise customers and startups, who have already seen strong s what those teams — including those at @Deloitte and @Box — and more had to say ↓
We gave early access to enterprise customers and startups, who have already seen strong s what those teams — including those at @Deloitte and @Box — and more had to say ↓
26
26
258
36,870
Box retweeted
Our love language is gifts: you can now sign up for a FREE @Box developer account at developer.box.com 💙
Today is National I Love You Day.
2
5
9
467
3 weeks until BoxWorks brings the enterprise AI community together in San Francisco. We’re grateful to our Premier Sponsors @awscloud, @Deloitte, @googlecloud, @IBM, @TCS, @DataBank_Ltd, and @MicroStrategy, and to all our partners helping organizations put Content + AI to work.
o meet our partners, explore their solutions in the Expo Hall, and see what’s next for enterprise AI.
Register today: boxworks.box.com/series/boxwork…
o meet our partners, explore their solutions in the Expo Hall, and see what’s next for enterprise AI.
Register today: boxworks.box.com/series/boxwork…
2
0
3
1,392
Box retweeted
Another spin on running agents in the open source MXC sandbox from @Microsoft, this time with a Box Mount-backed workspace. Fun to see all the pieces working together 🤙
🚀@Microsoft Execution Containers (MXC) provide policy-driven, sandboxed execution across Windows, Linux, and macOS. Box Mount runs inside that sandbox and maps a Box folder to a standard POSIX path, with two-way sync and Box permissions and governance applying automatically. ent reconciles fictional invoices, purchase orders, and payments using deterministic tools. It identifies six exceptions including duplicate records, a missing PO, an over-limit invoice, and unpaid invoices. The finished Markdown review and Excel exception workbook sync back to Box through Box Mount.
MXC is the isolated execution layer, and Box Mount is the governed content layer. The result lands in Box, ready for human review.
Demo ↓
0
2
4
345
🚀@Microsoft Execution Containers (MXC) provide policy-driven, sandboxed execution across Windows, Linux, and macOS. Box Mount runs inside that sandbox and maps a Box folder to a standard POSIX path, with two-way sync and Box permissions and governance applying automatically. ent reconciles fictional invoices, purchase orders, and payments using deterministic tools. It identifies six exceptions including duplicate records, a missing PO, an over-limit invoice, and unpaid invoices. The finished Markdown review and Excel exception workbook sync back to Box through Box Mount.
MXC is the isolated execution layer, and Box Mount is the governed content layer. The result lands in Box, ready for human review.
Demo ↓
MXC is the isolated execution layer, and Box Mount is the governed content layer. The result lands in Box, ready for human review.
Demo ↓
3
2
8
120,452
Box retweeted
Exciting day! Gemini 4 Argon is here, and it’s a significant leap forward for complex knowledge work.
In @Box’s evals, which reflect the work our customers do, it reached 85% accuracy, up from 71% for Gemini 3.1 Pro. What personally stood out to me: an incredible 93% on both data fting.
In one task, it showed how a company could launch sooner using stock it already had. In another, it spotted that an investment memo needed two more approvals—something the previous model missed.
This is critical for our customers: one overlooked detail can delay a launch or leave a decision without the right approvals. Gemini 4 Argon checked the details and finished the job.
We’re excited to be a @google launch partner!
Gemini 4 Argon is now available in Box AI Studio and Box AI for our customers to use in their workflows.
You can read our full evaluations here: blog.box.com/gemini-4-argon…
In @Box’s evals, which reflect the work our customers do, it reached 85% accuracy, up from 71% for Gemini 3.1 Pro. What personally stood out to me: an incredible 93% on both data fting.
In one task, it showed how a company could launch sooner using stock it already had. In another, it spotted that an investment memo needed two more approvals—something the previous model missed.
This is critical for our customers: one overlooked detail can delay a launch or leave a decision without the right approvals. Gemini 4 Argon checked the details and finished the job.
We’re excited to be a @google launch partner!
Gemini 4 Argon is now available in Box AI Studio and Box AI for our customers to use in their workflows.
You can read our full evaluations here: blog.box.com/gemini-4-argon…
Today, Gemini 4 Argon is officially rolling out.
0
2
2
534
We put Gemini 4 Argon from @GoogleDeepMind to work untangling an insurance billing dispute across 12 documents and 2 policies.
In Box AI, Argon traces a payment credited to the wrong policy and catches a bank return that changes the balance. Argon also reconciles the customer’s with the carrier’s final amount.
The statement shows $6,000. The customer expects $4,200. Argon generates a cited review explaining why $6,800 remains due after recommended corrections, with a draft customer response for human review.
Build custom agents for your own insurance workflows with Box AI Studio.
In Box AI, Argon traces a payment credited to the wrong policy and catches a bank return that changes the balance. Argon also reconciles the customer’s with the carrier’s final amount.
The statement shows $6,000. The customer expects $4,200. Argon generates a cited review explaining why $6,800 remains due after recommended corrections, with a draft customer response for human review.
Build custom agents for your own insurance workflows with Box AI Studio.
2
2
10
70,524
Gemini 4 Argon from @GoogleDeepMind is built for complex, multi-step work. We ran it against Gemini 3.1 Pro on real enterprise document tasks: pull the source files, reason across them, produce the actual deliverable.
→ Overall accuracy: 85% vs 71% (+14 points)
→ Data 23 points)
→ Life sciences: 88% vs 61% (+27 points)
→ Healthcare: 87% vs 57% (+30 points)
The gains land where the work has to be carried all the way through: reading evidence out of figures, covering the whole batch, and acting on its own findings.
→ Overall accuracy: 85% vs 71% (+14 points)
→ Data 23 points)
→ Life sciences: 88% vs 61% (+27 points)
→ Healthcare: 87% vs 57% (+30 points)
The gains land where the work has to be carried all the way through: reading evidence out of figures, covering the whole batch, and acting on its own findings.
Gemini 4 Argon is now globally available.
Built to solve your hardest problems from start to finish, the model brings frontier capabilities across coding, enterprise workflows, and cyber defense.
4
3
51
1,419,256
Box retweeted
At Box, we've been testing Gemini 4 Argon on our Box AI Complex Work eval for the past week. It's a huge upgrade over Gemini 3.1 Pro, jumping by 14 percentage points across all our tests, and firmly at the frontier.
In our evaluation, we give the Box Agent (using Gemini 4 plex enterprise tasks involving enterprise documents in a range of industries. These tasks approximate the kind of hard, time-consuming work that knowledge workers do every day with their enterprise content. Gemini 4 is a huge leap on effectively every dimension of this work.
Gemini 4 is far more capable on math, logic, analytics, and deep knowledge of a variety of business domains. It also works far harder to complete tasks fully without stopping short of the desired result. This is incredibly important for complex enterprise workflows that depend on robust task execution.
Here are some example wins that we saw to put the improvements in perspective:
* Financial services (+18 points): a funding-gap analysis for a family-owned company, where the brief spelled out every step of the projected statements, down to the interest on the bank loan and the overdraft for each year. Gemini 4 Argon built them exactly as specified, every line broken out and carried through to the funding gap. Gemini 3.1 Pro folded the interest into a single figure.
* Healthcare (+25 points): an intensive care management plan where both models spotted a dangerous potassium level and a drug interaction. Only Gemini 4 Argon went back and pulled the offending drugs from the final regimen, so its plan actually matched its own diagnosis.
* Technology (+37 points): a product performance review deciding which of two regional markets should have its cloud infrastructure optimized first. Gemini 4 Argon pulled the right month's user, retention and per-user figures for both regions and correctly flagged North America. Gemini 3.1 Pro often pulled the wrong rows, at one point reporting retention above 100%, and pointed the effort at the wrong region.
* Energy (+16 points): an energy consumption report across two facilities that had to flag any gaps in the meter data. Gemini 4 Argon consistently caught that the data center's smart meter log stops in November. Gemini 3.1 Pro often missed it, at times declaring the logs complete.
* Consumer products (+33 points): a product launch evaluation deciding which production expansion to fund. Gemini 4 Argon saw that existing inventory could bridge an early capacity gap, which made the smaller expansion viable with an earlier launch and a faster break-even. Gemini 3.1 Pro defaulted to the larger expansion and the later timeline.
Gemini 4 Argon will be available in the Box AI Studio shortly for anyone to build custom Box Agents with.
In our evaluation, we give the Box Agent (using Gemini 4 plex enterprise tasks involving enterprise documents in a range of industries. These tasks approximate the kind of hard, time-consuming work that knowledge workers do every day with their enterprise content. Gemini 4 is a huge leap on effectively every dimension of this work.
Gemini 4 is far more capable on math, logic, analytics, and deep knowledge of a variety of business domains. It also works far harder to complete tasks fully without stopping short of the desired result. This is incredibly important for complex enterprise workflows that depend on robust task execution.
Here are some example wins that we saw to put the improvements in perspective:
* Financial services (+18 points): a funding-gap analysis for a family-owned company, where the brief spelled out every step of the projected statements, down to the interest on the bank loan and the overdraft for each year. Gemini 4 Argon built them exactly as specified, every line broken out and carried through to the funding gap. Gemini 3.1 Pro folded the interest into a single figure.
* Healthcare (+25 points): an intensive care management plan where both models spotted a dangerous potassium level and a drug interaction. Only Gemini 4 Argon went back and pulled the offending drugs from the final regimen, so its plan actually matched its own diagnosis.
* Technology (+37 points): a product performance review deciding which of two regional markets should have its cloud infrastructure optimized first. Gemini 4 Argon pulled the right month's user, retention and per-user figures for both regions and correctly flagged North America. Gemini 3.1 Pro often pulled the wrong rows, at one point reporting retention above 100%, and pointed the effort at the wrong region.
* Energy (+16 points): an energy consumption report across two facilities that had to flag any gaps in the meter data. Gemini 4 Argon consistently caught that the data center's smart meter log stops in November. Gemini 3.1 Pro often missed it, at times declaring the logs complete.
* Consumer products (+33 points): a product launch evaluation deciding which production expansion to fund. Gemini 4 Argon saw that existing inventory could bridge an early capacity gap, which made the smaller expansion viable with an earlier launch and a faster break-even. Gemini 3.1 Pro defaulted to the larger expansion and the later timeline.
Gemini 4 Argon will be available in the Box AI Studio shortly for anyone to build custom Box Agents with.
Gemini 4 Argon is here.
It’s available globally today for developers and enterprises in the paid tier of the Gemini API and the Gemini platform, for Ultra subscribers in Google Antigravity, and Ultra and Pro subscribers in the Gemini app and Search.
Hope you all enjoy it!
72
26
425
73,228
Box retweeted
We are 3 weeks out from the first ever BoxWorks Dev Summit!
We'll be exploring how AI is unlocked by unstructured data with:
- @ivanburazin / @daytonaio
- @charlierguo / @OpenAIDevs
- Joey Conway / @nvidia
- @sobtiankit / @getpostman
Plus, hands-on workshops from @awscloud ocker.
If you're in the SF Bay Area, we have a limited number of free passes for devs, use the link below to RSVP!
luma.com/box-dev-summit…
We'll be exploring how AI is unlocked by unstructured data with:
- @ivanburazin / @daytonaio
- @charlierguo / @OpenAIDevs
- Joey Conway / @nvidia
- @sobtiankit / @getpostman
Plus, hands-on workshops from @awscloud ocker.
If you're in the SF Bay Area, we have a limited number of free passes for devs, use the link below to RSVP!
luma.com/box-dev-summit…
1
3
18
6,003
Box Mount lets developers mount Box as a filesystem directly in any agent sandbox, so agents can work with files and data just like a person would.
In this article, we show how it works with @digitalocean, @e2b, @OpenAI, and @vercel. x.com/i/article/2109…
In this article, we show how it works with @digitalocean, @e2b, @OpenAI, and @vercel. x.com/i/article/2109…
9
10
43
194,139