Understanding the Agentic Real-Time Framework
Programmatic is advancing rapidly as our industry adapts to the growing influence of AI and automation across the broader tech ecosystem.
Real-time intelligence, like models and specialized logic, is moving directly into the auction, enabling faster, smarter decisioning.
But realizing that potential requires a level of cooperation the industry hasn’t yet seen at scale. It requires shared infrastructure. And it requires standards that allow companies to participate inside each other’s environments in a predictable, governed way.
That’s exactly why the IAB Tech Lab introduced the Agentic Real-Time Framework (ARTF). ARTF creates the foundation for interoperable, high-performance agentic workflows inside programmatic, allowing intelligent systems to operate securely and efficiently within the auction itself.
This is the groundwork for near-instantaneous media trading and a new way of applying and enabling the advancement of AI in programmatic.
What is ARTF?
Index is proud to be one of the pioneering companies behind the ARTF. We worked within the IAB Tech Lab to create a standard that modernizes how OpenRTB operates. ARTF combines a containerized execution environment with a standardized API. This allows ad tech companies to run their logic—whether that’s models, enrichment, verification, or optimization—directly inside the infrastructure where ad transactions occur.
It moves intelligence closer to the impression and creates the potential to reshape performance, efficiency, and control across the open internet. Instead of sending bid requests across multiple external systems and public clouds, participants can run their logic locally—in the same data center as the auction—using secure, portable containers.
This eliminates unnecessary outbound calls, making decisioning faster, lowering compute costs, and simplifying the system itself.
Why standardization matters
Standardizing this model through the ARTF allows for scale while managing infrastructure costs. It helps programmatic keep pace with the ever-growing open internet.
In ad tech, flexibility matters. If you’re building a solution, you don’t want to rebuild your integration every time you work with a new exchange. And host platforms don’t want to manage dozens of one-off integrations.
As new agentic participants enter programmatic, shared standards become essential to keeping the ecosystem interoperable, transparent, and high-performing. Without a shared standard, innovation becomes fragmented, compromising scale.
ARTF ensures containers can plug into different host environments in a consistent, scalable, and secure way so everyone operates on the same foundation.
How ARTF works in practice
Imagine you’re an ad tech company—maybe an identity provider, an ad verification platform, or you have a custom algorithm. If your logic operates on a per-impression basis, needs to run in milliseconds, and benefits from massive global internet scale, ARTF is designed to help.
In the legacy programmatic model, an impression opportunity triggers a bid request. The SSP then sends the request to you and other partners, and waits for a response. Each call could add hundreds of milliseconds of latency.
With ARTF, your code is instead localized into a container. The host platform, typically an SSP, then runs that container right inside its infrastructure.
When an impression opportunity triggers a bid request, the host platform makes an API call to applicable containers. Essentially asking each, “Here’s the request. What would you like to do with it?”
Each container executes and applies its own unique and proprietary logic and value, then sends an API response that includes a suggested mutation to the bid request. There are many use cases here, with many more being built as time goes on, but for example, that might mean:
- Adding an audience segment
- Activating a deal ID
- Adjusting a floor
- Attaching a signal
- Applying other custom logic
The host reviews these proposed mutations, applies its policies, and decides which changes to merge into the live bid request. This all happens in under 10 milliseconds.
The enriched and validated bid request then continues along its normal path to DSPs.
DSPs still bid the way they always have. But now, the request they receive can be informed, enriched, and optimized in real time inside the auction environment.
The mechanics of RTB don’t change dramatically. What changes is that we now have a standardized interface that allows specialized partners to contribute logic directly inside the flow. And because all of this happens within the host’s environment, compute is substantially reduced while the host maintains control over data and outcomes.
How ARTF benefits the programmatic ecosystem
This shared framework and interoperable infrastructure is essential to realizing the full potential of agentic systems. It benefits all parties—buyers, media owners, platforms, and specialist technology partners—in three key ways:
- It standardizes execution. Partners integrate once to a containerized environment instead of maintaining many bespoke and costly integrations.
- It reduces latency and variability. ARTF is designed to reduce bid request and response times by up to 95%, dramatically increasing opportunities for bid enrichment and real-time decisioning, while making performance more predictable.
- It improves data governance. Host platforms remain gatekeepers of what data is shared and which mutations are accepted, while still enabling richer collaboration.
Above all, ARTF lowers barriers to entry and makes it much easier for new companies to bring innovative solutions to market, accelerating progress across the industry.
If you’re considering ARTF, your solution should:
- Operate in real time
- Work on a per-impression basis
- Deliver meaningful decisioning or enrichment
- Tie back to an OpenRTB request
Solutions could be identity resolution, outcome-based optimization, fraud detection, deal logic, creative enrichment, or something entirely new. If your value can be expressed as a change to a bid request, ARTF provides a standardized way to participate.
ARTF modernizes collaboration within the existing programmatic framework. By running logic locally, we reduce unnecessary cloud hops, lower latency, and reduce infrastructure costs while enabling a new class of real-time, intelligent decisioning across the open internet.
Most importantly, we do it in a standardized and interoperable way. Because innovation scales faster when everyone speaks the same language.
See what’s possible when you bring intelligence closer to the impression.
