Customers

/

Adspirer

CUSTOMER CASE STUDY

How Adspirer secures the full lifecycle of AI advertising operations

Adspirer's AI agents plan, launch, and optimize campaigns across Google Ads, Meta, LinkedIn, TikTok, and Klaviyo, driven from conversational AI tools like Claude and ChatGPT. BlueRock gives them visibility, policy enforcement, and operational control across their MCP infrastructure, from build-time validation to runtime.

Logo lockup

Adspirer's customers run advertising campaigns by talking to AI. Behind each conversation, autonomous agents plan, launch, and optimize across Google Ads, Meta, LinkedIn, TikTok, and Klaviyo, making thousands of decisions a day across MCP infrastructure that handles roughly 50 million tool calls daily.

That autonomy is the product. It is also a new security surface that traditional SAST and logging were never built to cover.

CHALLENGE

Autonomous agents acting across sensitive ad platforms created a security surface SAST and logging could not cover.

APPROACH

BlueRock secured the full MCP lifecycle: scan before deploy, observe at runtime, enforce at execution.

OUTCOME

Visibility and control across the action path, with no change to how Adspirer's agents work.

Industry:

AI-native advertising platform ·

Infrastructure:

MCP servers on GCP ·

Products:

MCP Trust Registry, Observability + Guardrails ·

Scale:

~3,000 customer environments, ~50M MCP tool calls/day ·

Deployment:

alongside MCP workloads, observe-only option

3,000

Customer environments

50M

MCP tool calls per day

6 Stages

6 Stages

BUILD

SCAN

DEPLOY

OBSERVE

ENFORCE

SCALE

Operational scale discussed at the time of collaboration: approximately 3,000 customer environments and 50 million MCP tool calls per day.

As an agentic AI company, Adspirer powers AI performance marketing agents that plan, launch, and optimize campaigns across Meta, Google, and more. With thousands of decisions flowing through our MCP infrastructure, our customers expect reliability and security. BlueRock gives us the visibility and control we need to run agentic systems with confidence.
A
Abhi MekalaCEO, Adspirer

What changed for Adspirer

BlueRock helped Adspirer strengthen operational confidence and enterprise readiness during a period of rapid platform growth.

Improved Visibility

Into MCP activity and execution behavior.

Reduced Complexity

During investigation workflows across agentic systems.

Strengthened Security

Across development and runtime environments.

Supported Assurance

For enterprise customer security requirements.

Scaled with Confidence

Across agentic marketing workflows.

Improved Visibility

Into MCP activity and execution behavior.

Reduced Complexity

During investigation workflows across agentic systems.

Strengthened Security

Across development and runtime environments.

Supported Assurance

For enterprise customer security requirements.

Scaled with Confidence

Across agentic marketing workflows.

THE CHALLENGE

Why traditional security wasn't enough

Adspirer's platform is built around public-facing MCP infrastructure connecting AI agents directly to downstream advertising and analytics systems. As the company scaled, traditional security and logging approaches no longer provided sufficient visibility or operational confidence for autonomous execution environments. Securing agentic systems required controls that extended beyond production runtime and into the software delivery lifecycle itself.

Key Challenges

Continuously evaluate MCP security posture before deployment

Understand how agents interacted with tools and downstream systems

Monitor runtime behavior across agentic workflows

Reduce exposure to prompt injection, tool abuse, and unsafe execution paths

Improve operational investigations during enterprise security evaluations

Before
AI Agent
Marketing
After: Expanding Attack Surface
AI Agent
MCP Servers
Advertising Platforms
Analytics & Data Systems
Customer Data & External APIs

THE APPROACH

How Adspirer secures its MCP infrastructure with BlueRock

Lifecycle coverage beyond production runtime

MCP-specific scanning traditional SAST misses

Observe-only deployment alongside existing workloads

Scan Before Deploy

BlueRock's MCP Trust Registry continuously scans Adspirer's server builds prior to deployment, identifying MCP-specific vulnerabilities traditional SAST scans do not detect.

OWASP MCP Top 10

MITRE CWE

MAESTRO

Observe Runtime Behavior

Visibility across agent decisions, MCP tool calls, downstream execution behavior, runtime anomalies, and behavioral drift.

Agent Decisions

MCP Tool Calls

Behavioral Drift

Enforce At Execution

Policy controls across agent-to-tool interactions, agent-to-data movement, and execution boundaries.

Agent-to-Tool

Agent-to-Data

Execution Boundaries

TECHNICAL INTEGRATION

Deployed without disrupting existing workflows

BlueRock integrated directly alongside Adspirer's MCP workloads running on GCP infrastructure, providing runtime telemetry, observe-only deployment, policy enforcement, downstream execution visibility, and compatibility with existing conversational AI workflows.

1
Build
2
Scan
3
Deploy
4
Observe
5
Enforce
6
Scale
Continuous feedback loop

Who is Adspirer?

Adspirer is an AI-native advertising platform that lets marketing teams execute campaigns, analytics, and optimization workflows across Google Ads, Meta, LinkedIn, TikTok, Klaviyo, and other systems directly from conversational AI interfaces such as Claude and ChatGPT.

The platform runs MCP infrastructure at scale and enables users to:

Create and optimize campaigns

Pull analytics and reporting

Adjust targeting and budgets

Coordinate cross-platform workflows

Automate operational marketing tasks

You interact
Claude
ChatGPT
Other LLMs
Adspirer Platform
AI agents orchestrate
workflows via MCP
Advertising & marketing systems
Google Ads
Meta
LinkedIn
TikTok
Klaviyo

The speed of agentic marketing, with guardrails that keep it safe at scale.

What is BlueRock's MCP Trust Registry?

The MCP Trust Registry continuously scans MCP server builds prior to deployment, identifying MCP-specific vulnerabilities that traditional SAST tools do not detect, mapped against frameworks including the OWASP MCP Top 10, MITRE CWE, and MAESTRO. The MCP Trust Registry is a free community resource. www.mcp-trust.com

Does BlueRock require changing how Adspirer's MCP infrastructure runs?

No. BlueRock integrated alongside Adspirer's existing MCP workloads on GCP infrastructure, including an observe-only deployment option, without disrupting existing conversational AI workflows.

What runtime visibility does BlueRock provide?

BlueRock provides visibility across the full agentic action path, including agent decisions, MCP tool calls, downstream execution behavior, runtime anomalies, and behavioral drift.

What kind of policy controls does BlueRock enforce?

BlueRock enforces policy controls across agent-to-tool interactions, agent-to-data movement, and execution boundaries, applying protections directly at the moment of execution.

Who is Adspirer?

Adspirer is an AI-native advertising platform that lets marketing teams execute campaigns, analytics, and optimization workflows across Google Ads, Meta, LinkedIn, TikTok, Klaviyo, and other systems directly from conversational AI interfaces such as Claude and ChatGPT.

Run your agents with the same confidence

BlueRock provides observability and guardrails across the full Agentic Action Path, from model decision through MCP tool calls to execution and outcome.

What to read next

MCP Trust Registry

Adspirer case study: blog post

Agentic AI breaches and visibility

About Adspirer

Adspirer is an AI-powered advertising platform that enables marketing teams to operate campaigns through conversational AI systems using MCP infrastructure. The platform connects autonomous workflows across major advertising and analytics systems including Google Ads, Meta, LinkedIn, TikTok, and Klaviyo.

About BlueRock

BlueRock.io provides runtime observability, guardrails, and trust infrastructure for agentic systems and MCP environments. Its platform helps organizations securely operate, understand, and control autonomous execution across AI-native infrastructure.

Last updated June 16, 2026