Headless marketing.
Let AI operate the tools.

Stop babysitting your MarTech stack. Let your team do the marketing.

From familiar tools like Claude, ChatGPT, Slack, or Teams, ask AI to get work done across your connected marketing tools. It queries customer data, builds audiences, and prepares journeys—so your team spends less time clicking through screens and configuring every step by hand.

You direct the work. You approve customer-facing actions.

Treasure AI Studio

The shift

From operating the UI to directing AI.

Same objective: build an audience and a re-engagement journey. A different way to get there.

Before

UI

Drag. Drop. Repeat. 1,000 times.

Every filter. Every branch. Every setting.
You operate the tools, one step at a time.

After

AI

Just tell AI what you need.

Your AI of choice

AI StudioCopilotSlackTeamsClaudeOpenAI

“Find customers who abandoned their cart in the last 7 days and prepare a re-engagement journey for my review.”

Audience draftedJourney prepared
Preview. Refine. Approve.

Audience Studio

Cart abandoners

Added to cart
In the last 7 days

No purchase
Since adding to cart

Review the audience and its conditions.

Edit draftApprove

Engage Studio

Re-engagement journey

Review the journey before it goes live.

Edit draftApprove

AI builds. You preview, refine, and approve.
In Audience Studio and Engage Studio.

The UI does not disappear. Its role changes: from where you build everything by hand to where you preview, refine, and approve what AI prepares.

How Headless Works

Different ways in. One governed platform.

Connect your AI through Agent Gateway or the CLI. Work on the same governed customer context.

Collaboration

  • Slack
  • Microsoft Teams
via Agent Gateway

AI Assistant

  • Treasure AI Studio
  • ChatGPT Enterprise
  • Claude.ai
  • Microsoft Copilot
MCP

Agent SDK

  • Your app
  • Vercel AI SDK
via Agent Gateway
Agent Gateway

AI Coding

  • OpenAI Codex
  • Claude Code
  • GitHub Copilot
  • Cursor
via CLI
CLI (TDX)

Treasure AI platform

Governed customer context

  • Customer data
  • Segments
  • Journeys
  • Activation
  • Same permissions
  • Same audit trail
  • Human approval for customer-facing actions

More ways to operate. Not a second platform. Your team can still preview and refine the work in UI.

Work where you already work

Your AI. Your workspace. Your choice.

Use Treasure AI Studio, your AI assistant, your coding agent, or your team’s workspace. Or build your own application. Choose your starting point below to explore the relevant setup and documentation.

Same customer context. Same permissions. Same audit trail. Setup and availability vary by connection.

Case study

The power of the Intelligent CDP and marketing automation together

Brother Sales in Japan faced a data crisis: their legacy systems in Japan couldn't scale with the surge of IoT data from its connected printers.

By adopting Treasure Data, Brother Sales moved marketing automation capabilities natively into their Intelligent CDP. This eliminated costly data transfers and streamlined operations.

  • Fast ROI: Cost-effective switch delivers positive return in about two years.
  • Proactive CX: Targeted, real-time emails reduce support costs and improve loyalty.
image-brother case study@2x

Built-in skills

Connected is not enough. Give your AI the know-how.

50+Skills

400+CDP implementations
informing the skills

Skills come with Treasure AI. Each skill is a SKILL.md folder with the instructions, examples, and reference material for one job, in the same format Claude Code uses.

Audiences & journeys

Define segments, build journeys, and validate configurations before deployment.

SegmentsJourneys

Workflows

Build data workflows with Digdag and transformations with dbt.

Digdagdbt

Real-time experiences

Set up real-time personalization, event-triggered journeys, and identity stitching.

Real-time skills

Open any skill to read exactly what your AI is told.

Explore the skills on GitHub

FAQs

Headless gives the AI tools your team already uses direct, governed access to your customer data, segments, journeys, and activation. Ask an AI assistant or coding agent for a win-back audience or journey, and it is drafted and validated in Treasure AI for your review. Agents draw on 50+ skills built from 400+ CDP implementations, as described in our Headless announcement, so they follow proven patterns instead of guessing.

Treasure Code CLI is available today, with documented setup for Claude Code, OpenAI Codex, VS Code with GitHub Copilot, and Cursor. The Agent Gateway and remote MCP server are also available now, including Slack and Teams connectors. Remote MCP is reachable from Claude and ChatGPT Enterprise. You can also build Treasure AI into your own applications with the Agent SDK or Vercel AI SDK.

No. Treasure AI Studio remains the UI. Headless is another way in to the same data, with the same permissions and the same audit trail. Your team can work in Studio, in the AI tools they already use, or in both.

No. Customer-facing sends require human approval. Agents run in isolated conversation containers, their network and file access is traced, and their activity is recorded in an audit log.

The gap isn’t AI vs no AI. It’s speed vs irrelevance.

Don’t fall behind. Start capturing revenue in minutes.