Skip to main content

← All writing

Contentstack launched Agent OS, AXP, and the dev tools I have been itching to talk about

Contentstack launched Agent OS, AXP, and the dev tools I have been itching to talk about

Yesterday Contentstack made Agent OS generally available, renamed the platform to AXP, and launched a pile of developer tooling I have been obsessing over for months. I helped build parts of the AI side, so I am not pretending to be neutral here. I am biased, happily so.

I do not think this is just another AI launch. I have been waiting to write about this one because it pulls together a lot of things I care about at once: content, APIs, developer experience, agents, and the foundational work that suddenly gets exciting the moment the market catches up.

This one feels different

A lot of AI product work right now feels like someone took an existing product, taped a chatbot to the side, and called it transformation. That is not what this is. Contentstack had a real advantage going in, because the platform was already API-first, composable, with automations, brand kit, and built around content and context long before agents became the word in every slide deck. Agent OS does not feel like something glued on after the fact, it feels like the logical next step.

This matters as agents are only useful when they understand the system they are working inside. Without that understanding they guess, and at enterprise scale guessing becomes a hallucination with a budget.

Agentic Experience Platform

AXP stands for Agentic Experience Platform. The way I think about it is simple: a general AI tool can write a decent paragraph, but it does not know your content model, it does not know who is on your site right now, and it has never read your brand guidelines. It does not understand your workflows, your permissions, or the difference between something that sounds right and something your company can actually ship, so it fills in the blanks on its own.

AXP is Contentstack’s answer to that, and it brings together three things. Content Cloud is where your content is structured, governed, and ready to be reused. Data Cloud is where real-time customer context comes in. Agent OS is where agents can actually take action inside the tools your teams already use. The combination is key, because content without context is limited, context without brand governance is risky, and agents without either create generic slop. Put them together and AI helps teams do real work with the right grounding underneath it.

Agent OS

Agent OS is the action layer of AXP. It is a workforce of AI agents that can use your content, customer data, brand rules, and permissions inside Contentstack. It operates in the heart of the product.

The first pieces people will notice are Polaris and Brand Kit. Polaris is the conversational companion inside Contentstack: it understands your content model, it can use live audience data, and it respects role-based permissions. Brand Kit is the other one I keep coming back to, because it turns your voice, visual rules, messaging, and facts into something agents can follow. Instead of asking AI to vaguely sound on brand, you hand it actual brand context. This moves you from “please write something and hopefully it is fine” to “here is the system, here are the rules, here is the context, now help me move faster without making a mess.” Agent Builder and Automations come next, which is where this gets even more interesting.

The new developer site

We launched developers.contentstack.com, and I wanted it to feel like a real home for engineers rather than a marketing page wearing a hoodie. The goal is to bridge the space between official docs and actual production engineering, so the site has guides, an engineering blog, livestreams, and kickstarts for Next.js, Nuxt, Astro, and all other meta frameworks. The point is not just “here is the API,” it is to show how this works in practice, why you would structure it a certain way, and how to get moving without spending your first day wiring everything together from scratch.

And because it is 2026 and agents are part of how many of us build now, the site is agent-friendly too. Every blog post, guide, and kickstart is available as markdown, there is an agents.md file that tells LLMs how to browse the site, and there is an MCP on top so you can learn Contentstack from your AI assistant instead of constantly bouncing between tabs. That detail makes me weirdly happy. It is a small developer experience thing, but it matters: if people are building with AI next to them, our docs and tools should be designed for that reality.

Official AI skills

Alongside the site we shipped two pieces of AI tooling I am especially proud of. The first is the Contentstack vibe docs. The official docs are detailed, which is good for humans but often rough for agents, since they can blow through a context window fast. The vibe docs are a condensed, token-efficient version of that knowledge, around 20k tokens total, designed to load when needed instead of dumping everything in at once. The second is the Contentstack agent skills bundle, which includes twenty ready-to-use tasks across CMS, Brand Kit, Launch, and developer experience. What I like is that these are not just prompts, they include rules around credentials and safe handling, because agents doing useful work need guardrails, not vibes alone.

You can install both with a single command each:

npx skills add contentstack/contentstack-vibe-docs
npx skills add contentstack/contentstack-agent-skills

MCP

And then there is the MCP. The current one is experimental, but it has become a lot more solid, and the architecture is smarter now. The tools live on a server, which means you do not have to wait for us to expose every possible version of the thing. You can take those tools, build a Developer Hub app for OAuth, and assemble your own MCP today.

The next version is the one I am very excited about, because it is URL-based and profile-based: you select the tools you actually want, connect to any Contentstack product, and get OAuth, audit logs, and enterprise controls built in from the start. That is coming soon, and I cannot wait for people to try it.

Rounding it up

I have spent a lot of my career arguing that API-first and developer-first are not just nice product principles, they are what let you move when the ground shifts. This launch feels like proof of that. The content foundation was already there, the APIs were already there, the developer experience was already there, and Agent OS turns all of it into action.

So yes, I am proud. Proud of the team, proud of the craft that went into this, and proud that we did so much of the unglamorous foundational work before it became the obvious thing to do.

On this page
Aug 22, 20263 min read

Buy the plumbing, vibe the rest

AI makes it tempting to cancel SaaS tools and prompt your own internal platforms into existence, but that often creates brittle systems that are “a mile wide and an inch deep.” Performance, security, and edge cases quickly become serious problems, and suddenly your team is doing database administration and security engineering instead of solving business problems. A better pattern is to buy robust, headless infrastructure for the hard, invisible parts (content, data, security, scaling) and then use AI to build bespoke experiences on top. With a solid SDK and stable backend services, AI tools like Claude can safely orchestrate UI and workflows instead of guessing at architecture. Pay for the plumbing, and vibe on the interface layer where your differentiation really lives.

  • Composable architecture
  • AI engineering
  • Frontend
Aug 17, 20265 min read

The Doer Economy - we are killing the translator class

This article argues that AI is collapsing the distance between vision and execution, ushering in a Doer Economy where the primary winners are those who can both imagine and build. The traditional corporate stack of translators (product managers, marketers, and multiple layers of management) is shrinking because executional tasks are increasingly handled by AI. Founders and CEOs are moving closer to product, validating ideas directly with AI-generated scaffolds and modern SaaS primitives. For individuals, the value has shifted from narrow, ticket-driven skills to end-to-end ownership, product thinking, and understanding users. The future belongs to people who ship, iterate quickly, and leverage AI and SaaS platforms to focus on business logic and user experience, rather than those who only manage the builders.

  • Composable architecture
  • AI engineering
  • API design