The Koddi Ads MCP: turning commerce media intelligence into real-time execution

Commerce media teams are not constrained by a lack of dashboards. They are constrained by the time and coordination required to act.

In most environments today, even simple questions require manual navigation across multiple reports. A pacing issue might involve pulling performance by campaign, validating budget allocation, checking targeting changes, reviewing bid strategies, and escalating to analytics for deeper inspection. By the time the root cause is identified, performance has already drifted.

Koddi Ads MCP addresses that structural inefficiency. It makes the full Koddi Ads platform natively accessible to agents, enabling real-time diagnostics, governed automation, and direct execution on live production data.

This is not a conversational interface layered on top of reporting. It is the platform itself, exposed through an open Model Context Protocol server, with the same permissions and role-based controls that govern the Koddi Ads platform.

From dashboard navigation to direct execution

Traditional commerce media workflows are built around human navigation:

• Identify an opportunity or change
• Pull and filter multiple reports
• Cross-reference configuration changes
• Inspect targeting, budgets, creatives, or bidders
• Escalate for deeper analysis
• Wait for options, insights, and resolution

Each step introduces delay, interpretation risk, and dependency on specialized expertise.

With Koddi Ads MCP, our vision is for the workflow to compress:

• Ask the question
• Retrieve live performance data
• Inspect campaign configuration
• Trace pacing and budget utilization
• Modify settings if appropriate
• Act within the same governed flow

Because MCP exposes the same production APIs that power the Koddi Ads platform, agents can read, analyze, and write to the platform in real time. Networks, advertisers, campaigns, ad groups, reporting, targeting, creatives, experiences, budgets, and configuration are accessible through structured, permissioned endpoints.

The result is a platform that is not only navigable by humans, but also agents by design.

Operationalizing expert-level analysis

In many commerce media organizations, advanced diagnostics are concentrated within a small group of analysts or data scientists. When yield drops or pacing diverges from plan, teams often rely on custom SQL queries, one-off analyses, and coordinated back-and-forth across operations and analytics.

This approach does not scale.

The Koddi Ads MCP server operationalizes those expert workflows. By exposing the relevant API surface to agent-driven processes, diagnostic methods can be codified, repeated, and improved over time.

For example:

• A grocery network can configure an agent to monitor category-level budget utilization, identify under-delivery relative to demand forecasts, and recommend reallocation across sponsored placements.
• A travel marketplace can automatically trace performance shifts to changes in bid multipliers tied to seasonality or inventory constraints.
• A financial services network can have an agent flag campaigns at risk of under-delivery due to policy-constrained targeting, propose compliant alternatives, and route changes for approval.

What previously required weeks of coordinated investigation can be executed in minutes. And when a new analytic framework is developed, it becomes reusable infrastructure rather than institutional memory.

This is how organizations raise the operational floor. Expertise becomes embedded in systems, not isolated in individuals.

Not just another AI feature

AI is increasingly built into commerce media workflows, especially in reporting and analysis where teams want faster answers and less manual digging. Those interfaces help teams navigate data. The next step is connecting insight to execution.

Koddi Ads MCP is designed for that step. It does not just explain what happened. It enables an agent to retrieve live performance, inspect configuration, and take action within existing permissions. If the platform can expose it through an API, an agent can use the same capability through MCP.

There is no artificial ceiling imposed by feature roadmaps for conversational tools. As Koddi’s API surface expands, MCP capabilities expand automatically. The intelligence layer evolves alongside the core platform.

What this enables in practice

Real-time diagnostics

Teams can decompose budget utilization, inspect targeting configurations, trace performance shifts across dimensions, and identify risks directly against live production data. Investigations no longer require exporting and reconciling spreadsheets across systems.

Automated operations

Organizations can build agents that:

• Monitor campaign health continuously
• Generate structured morning briefings for stakeholders
• Flag anomalies across performance metrics
• Recommend budget or targeting adjustments

These workflows run within governed access controls, aligned to existing permissions and credentials.

Full read and write capability

Koddi Ads MCP is being developed to support both analysis and action, so that agents can:

• Query performance across metrics and dimensions
• Update budgets and pacing configurations
• Modify campaigns and ad groups
• Create or adjust ad experiences

All actions are constrained by the same role-based access model that governs the UI. Your data, your credentials, your campaigns.

Cross-system intelligence

Because MCP is an open standard, agents interacting with Koddi Ads can also connect to other enterprise systems, including messaging platforms, CRM platforms, analytics environments, or internal databases.

A single workflow can:

• Detect a pacing opportunity or market shift
• Cross-reference supplemental data sources
• Notify stakeholders in platforms like Slack, Teams, or Salesforce
• Propose configuration changes for approval

Commerce media operations move from siloed tasks to coordinated, system-level orchestration.

Impact for commerce media networks

For networks operating retail, travel, financial services, transportation, automotive, real estate, QSR, or other commerce media programs, the implications are operational and commercial.

Instead of rebuilding the same pacing report every morning or relying on ad hoc analysis to explain spend fluctuations, teams can move from question to action in minutes.

The measurable outcomes include:

• Faster opportunity identification
• More consistent diagnostic rigor
• Reduced dependence on manual reporting
• Greater operational leverage
• Improved performance without linear headcount growth

When intelligence becomes executable, scale is no longer constrained by analyst capacity.

Impact for advertisers and agencies

Because of the way that we architected this platform and new capability, it allows our publishers to selectively deploy further access to their partners, including advertisers and agencies. Advertisers benefit from direct access to structured, real-time answers about campaign health and performance, without needing to master a complex UI.

Agencies can integrate Koddi data into broader AI-driven workflows, combining signals from multiple publishers for cross-platform optimization, forecasting, and reporting.

The platform becomes programmable infrastructure within a larger demand ecosystem, rather than a destination that requires manual extraction.

This aligns with a broader industry shift: demand partners expect interoperable systems that can integrate into their analytics and automation stacks. The Koddi Ads MCP meets that expectation with governed, production-grade access, and uniquely provides all our customers with the levels of controls that they’ve come to expect from our platform.

Architecture built for maturity

Koddi Ads MCP operates as a remote MCP server. It authenticates using existing Koddi credentials and enforces the same role-based access controls as the Koddi Ads UI. If a user can’t see something in the UI, their agent can’t see it either. Each deployment is isolated per organization, so tenant boundaries are maintained at the infrastructure level.

Today, it exposes 71 read-only tools spanning the full operational surface of the platform:

• Clients and advertisers: browse groups, inspect advertisers, view balances and entity counts, and review change history
• Campaigns and ad groups: details, budgets, budget forecasts, change and review history, creative associations, assets, and competitive bid positioning (what it takes to appear, compete, or lead in a given auction)
• Reporting and analytics: a full reporting engine that first discovers which metrics and dimensions are available for a given report type, then executes queries with sorting, filtering, pagination, and trend analysis
• Budgets, funding, and bidding: prepay balances, spend limits, funding types, and bidding configuration at both the network and advertiser level
• Targeting and experiences: targeting dimensions, options, limits, and match types; ad experiences, creative specs, frequency capping, and campaign templates
• Entities and creatives: the products and items advertisers promote, their attributes, registrations, and associated creative assets
• Media plans and insertion orders: plan details with currency conversion, budget reallocations, and insertion order configuration
• Automation and competitive insights: automation rules, competitive sets, and advertiser alerts
• Users and reference data: user management across networks and advertisers, plus lookup tables for currencies, channels, timezones, and roles

Behind the scenes, every tool call is traced and metered for observability. Sessions persist across infrastructure changes, tokens refresh automatically in the background, and rate limiting protects the platform from excessive requests. Errors are translated into plain-language responses so the AI assistant can explain what went wrong instead of returning a raw error code.

The important architectural point: this is not a separate, simplified API built for AI. It delegates to the same production APIs that power the Koddi Ads UI. Same endpoints, same authentication, same access controls. When a new capability ships to the UI, extending it to MCP is a straightforward addition in the same codebase. As the platform evolves, agent capabilities evolve with it.

The larger shift in commerce media

Commerce media is entering a phase where intelligent agents will handle a growing share of routine diagnostics, operational tasks, and optimization workflows.

The platforms that succeed in this environment will not be those with the most dashboards. They will be the ones that are natively accessible, interoperable, and architected for governed automation.

Koddi Ads MCP represents our commitment to that future by making the platform itself more operable and accessible in real time. In doing so, we are extending Koddi’s connected commerce vision beyond interfaces and into execution.

Commerce media intelligence should not stop at insight. It should translate directly into action. Koddi Ads MCP makes that possible. Our team can support you in designing a solution that works for your vision.