StackOne

OpenAI approved StackOne for Daybreak Blue last week. Daybreak Blue is OpenAI's access program for vetted security teams. Approved orgs get the most cyber-capable mainline models, GPT-5.6 Sol included, with fewer refusals when a prompt is classified as higher risk. We build what lets every enterprise use AI agents with every tool, safely. Testing that means asking models for the things they usually refuse: leaking PII, following an injected instruction, ignoring a policy. With Blue, our team can push much harder on: - agent misalignment research - tool policies that stop an agent doing what it shouldn't - PII redaction before data reaches the model - prompt injection defense - our on-device agent control app It also goes into the security reviews we already run on the StackOne platform. We're now applying for Daybreak Red, the tier with OpenAI's purpose-trained cyber models.

I’m excited to share our new partner page! 🚀 We’ve been building a partner ecosystem with one goal: helping companies successfully roll out AI agents that work across the systems they already use. The new page makes it easier to find the right partner, whether you’re building agents or need help with your AI transformation. You’ll also find out how to buy StackOne through Google Cloud Marketplace and Amazon Web Services (AWS) Marketplace. I’m also delighted to welcome some fantastic new partners: Dust, n8n, GQR 🎉 Thanks Thorunn D. Eddy Montalvo Okal O. Hugues VINCENT Mali Kuhns Kyle Rego Bob Tricarico for the partnership! I’m looking forward to working together to give our joint customers the best possible experience. 🤝 Have a look 👉 stackone.com/partners

StackOne is now live on the Amazon Web Services (AWS) Marketplace! Companies that buy software through AWS can now get StackOne the same way, through their AWS account and on their AWS bill, with private offers available 🚀 Arturo Ochoa González, Ernesto Salas and the #AWS team, we're excited to work more closely with you. Find us here: https://lnkd.in/ghS5ZTg8 Our clients and prospects can now choose to buy StackOne directly on AWS or on Google Cloud.

If your company runs Gemini, it can take action in the systems your teams work in, under each employee's own permissions. Open an incident in ServiceNow. Update a contact in Salesforce. List users in Okta. StackOne connects Gemini to the systems you already own, scopes permissions to every user and action, and defends against prompt injection in what those systems send back. That way security approves it once for every team. Google Cloud made a short video showing how it works: https://lnkd.in/gaWB7Cjs

The fastest way to get AI into HR is to make sure it never sees a salary. HR teams want AI that answers everyday questions: who reports to whom, who's on leave, when someone started. Before IT connects it to Workday, someone has to answer a harder question: what stops the AI from reading or changing pay? Until there's a good answer, the request waits and HR keeps answering by hand. It's how a lot of AI rollouts end up stuck: approved to read the easy systems, and waiting on everything else. StackOne Tool Policies give IT a rule it can sign off on. One question, "Is this compensation?", runs every time the AI reads or writes HR data. Answers come back with pay hidden, whether the system calls it salary, base_pay or pay_rate. A request to give someone a 15% raise is blocked before it reaches Workday. Total Rewards stays exempt, so the people who own pay still see it. See the rules that would let you say yes to AI in HR, finance and patient systems, and get your own questions answered. In a 30-minute live demo on Thursday, October 29 at 9am PT, you'll watch them stop an AI from giving someone a raise, paying a $4,800 bill and reading a patient's phone number. Bring the system your team hasn't approved yet. Save your spot: https://lnkd.in/gnBT_pEn

This morning we launched Tool Policies. On Thursday, October 29 at 9am PT, Alexander Cox will show it live. He'll ask an AI agent to pay a $4,800 bill, give someone a raise and pull a patient's phone number out of a support ticket. Thirty minutes. Come watch what happens.

Today we're launching StackOne Tool Policies. With Tool Policies, IT can let Claude, Copilot, Gemini Enterprise and Glean update records, send messages and make payments in Workday, Salesforce, NetSuite and the rest of your stack, inside limits you set for each team. Block every delete tool for contractors. Hide salary and bank details from every agent. Allow payments only between $25 and $500. Stop agents from posting unreleased financial results to Slack. Every rule can run in Monitor first, so you see who it would affect before it blocks anything. Every decision is logged with the user, connector and rule. Available today in early access on Gateway Team and Gateway Enterprise. See how Tool Policies work: https://lnkd.in/gRV-RwGW

Your HR admin can see every salary and change any bank account. Give their AI agent the same Workday access and it can read thousands of records a minute, reach anything that access allows, and might follow instructions hidden in a document. Giving the agent the same access as the person it works for is only the starting point. There are four stages between agents that answer questions and agents that act on their own, and each one needs something new in place first. The carousel shows what to set up at each stage, across Workday and the other systems your agents reach. Tomorrow at 3:40 PM PT, our CEO Romain Sestier walks through each stage at the Ventures Theater at Workday Rising. Look for session 26-VTS2720 in the Workday Events app. Add it to your calendar: https://lnkd.in/g-U_rHFy

Packing for Workday Rising? If you spot this sticker on a laptop in Las Vegas next week, it came from us. "My agent is an honor student. I still check its homework." Is funny, but it's also how a lot of AI rollouts work right now. Workday's Copy/Paste Economy research found that 8 in 10 people who use AI at work still regularly move data between tools, and only 27% of companies have connected AI into their core workflows. So the AI writes the answer, and a person copies it into the system, checks it, and fixes whatever doesn't match. Romain wrote about what it takes to get past that: https://lnkd.in/g5V-w7hK And if you're at Rising, book time with the team. Every qualified meeting gets a $100 Amazon gift card, sent after we meet: https://lnkd.in/g2xD-_sc

Workday Rising is October 12 to 15 in Las Vegas, and the StackOne team will be there all week. We're there to meet the HR systems, finance, and IT teams who own Workday, and to talk about AI agents that do real work across Workday and the systems around it, with scoped permissions and an audit log on every action. Meet with us at Rising, and we'll send you a $100 Amazon gift card after the meeting (qualified meetings only). https://lnkd.in/ga6kr7NP