Agent OS is now widely available. The difference is what it's grounded in.

AI promised transformation. Most teams got more work.

More tools to manage. More output to review. More ways for the brand to drift, the facts to slip, and the team to spend its day cleaning up after a model that never actually understood the business it was writing for.

The problem was never the intelligence—it was the grounding. A general-purpose AI tool can write a paragraph, but it doesn't know your content model, it can't see who is on your site right now, and it has never read your brand guidelines. So it guesses. And a guess at enterprise scale has a name your legal and brand teams already use for it: a hallucination.

Today that changes. Agent OS is now generally available across Contentstack.

The logic underneath everything

There’s a simple way to think about why most enterprise AI has underdelivered: 

Agents – Context = hallucination

We believe in a completely different thesis, and it’s behind everything we’re shipping:

(Content x Context x Brand) x Agents = Trusted Action

That’s the Contentstack Agentic Experience Platform. Three systems working as one:

• Content Cloud structures and governs your content, so AI works from a source of truth instead of a blank page.
• Data Cloud supplies real-time context: who the customer is and what they're looking for at the moment they engage.
• Agent OS is the layer of action, where that content and context get put to work inside the tools your team already uses.

Take any one of these three away and you’re back to generic output. Together, they produce work you can actually ship. 

With the wider availability of Agent OS, we want to show you the two parts of Agent OS your team will touch first: Polaris and Brand Kit. 

Polaris: Every user can be a power user

Polaris is the conversational companion built into Contentstack. Not a chatbot bolted onto the side, and not a separate window you paste things into. It lives exactly where the work happens. 

This is what it does, in practice:

• Turns multi-step operations into a sentence. Ask Polaris to draft, translate, restructure or find something, and it handles the interface work that used to require additional clicks or a support ticket.
• Answers from your stack, not from the internet. Because Polaris is grounded in your content model and your live audience data, the answer reflects what your customers are actually looking for, not a generic suggestion from a model that has never seen your business. 
• Respects what each person is allowed to do. Polaris operates strictly inside your role-based permissions. It will never surface data or take an action a user couldn’t perform manually. The human stays in control, and the audit trail stays intact. 

The shift is straightforward: the person who used to wait on a queue to get something done can now do it themselves, in seconds, without leaving the page. Your everyday users become power users. Your power users get their time back. 

Brand Kit: Your brand is the moat, so govern it like one

Here’s what most AI vendors aren’t saying out loud. The fastest way to dilute the one asset your can’t automate, your brand, is to deploy AI without governing it. Volume goes up, consistency goes down, and “generic slop” starts showing up in production under your logo. 

Brand Kit is how you scale output without making that trade. 

Most brand tools were built for humans. A brand portal tells a person where the logo lives. A writing assistant nudges a draft toward the right one. Neither of them can govern an AI. Your agents can’t read a PDF of your style guide, and a ten-page system prompt is too fragile to run an enterprise on. 

Brand Kit codifies your brand into something AI can actually use:

• Voice and tone, so generated content sounds like you by default, not like a model's idea of "professional."
• Visual standards, governed alongside the words, because brand is more than adjectives.
• A Knowledge Vault that separates your facts from your tone, so you update a product detail once and every agent reflects it everywhere.

The result is enforcement at the moment of creation and validation at the point of approval, not a manual review step you bolt on afterward and hope someone remembers to do. Brand Kit doesn’t tell your team what the brand is, but tells the AI how to represent the brand. 

Agent OS is already in production

Golfbreaks runs on accurate venue data, and there is no single reliable source for it, so keeping thousands of properties current used to be a constant manual effort. With Agent OS, the team cross-references their live content against multiple sources in the background and surfaces only the changes that matter.

"This shift from wholesale review to targeted, insight-driven updates is a meaningful change to how we manage content operations." — Jack Simkins, Digital Product Manager, Golfbreaks

That's the grounding thesis in one sentence. AI that knows your content and your standards doesn't create more work to check. It removes work you used to do by hand.

Where this goes next

Polaris and Brand Kit are the foundation. They make AI inside Contentstack grounded, on-brand and safe to use across your whole team. In July, we'll show you the next layer: Agent Builder and Automations, where those grounded, governed moments become agents that run on their own rules, on a schedule or on an event, with the human in the loop exactly where you want them.

Every AI capability in Agent OS draws from a shared pool of AI Credits included in your plan, with real-time visibility and protection on by default, so you can scale with predictability rather than surprises. We'll walk through how that works during the Jun 25 small group demo.

See it live on Jun 25

We're hosting a small group demo on June 25 to show how AXP brings Content Cloud, Data Cloud, and Agent OS together, with Polaris and Brand Kit running against real content.

Register for Jun 25 small group demo ->

Already know you want agents in production? Agent Accelerator pairs your team with our implementation engineers to design, build, and deploy your first agents in your live environment, in as little as two to three weeks. 

Learn more about Agent Accelerator ->

The teams that ground their AI now will build an advantage that's hard to close later. The work starts with content, context, and brand. It starts with Agent OS.