AI agents are rapidly emerging across enterprise stacks, but most remain isolated from use cases rather than integrated into critical workflows. Although 90.3% of companies report using AI agents, only 23.3% have them in production and only 6.3% have fully integrated AI into their marketing pipeline.
Adoption is high because AI is easy to use for isolated tasks. Integration is delayed because tailoring that output to a controlled, record-based workflow is more complex. In martech, the real limit isn’t access to AI – it’s combining possible outcomes and deterministic systems without breaching regulation, compliance or consistency.
Data shows that organizations are not replacing SaaS with AI. They put AI at the top of the SaaS systems that still run the business. The challenge is to make these systems work together without creating fragmentation or loss of control. The agent stack provides that model, and it varies greatly by company size.
Deterministic SaaS and probabilistic AI play different roles, but should work in the same stack. Record programs are always basic. They store data, apply rules and answer one question: What is the truth?
AI agents interpret situations and decide what action to take. They answer a different question: What should happen next?
At its simplest, the agent stack works like this.
- Context = guardrails: Pricing rules, product availability, legal rules and products, define what is allowed.
- Purpose = condition: What the customer wants and what they are trying to do defines what happens.
- Agents = decisions: Ask them both to decide what to do
It enables AI to work across SaaS. Integration becomes more critical, but also more difficult to control, because decisions now depend on organizing data, rules and context across multiple systems in real time.
Your customers are searching everywhere. Make sure it’s your product he appears.
The SEO toolkit you know, and the AI visibility data you need.
Start a Free Trial
Start with
How the agent stack works in practice
Here is a simple example. The customer requests the price of the product in conversation.
In a normal stack, this opens a lookup. The system determines the price based on predefined rules. The answer is correct, but not relevant to the customer.
In the agent stack, the same request becomes a combined decision. The agent retrieves pricing rules, product limits and contractual agreements from record systems while analyzing customer behavior such as behavior, time, channel and profile.
- Total customer explains whose answer it is. It shows the current state of the customer, not just their saved properties.
- Content context explains what can be said. This includes pricing, product availability, product tone and regional or legal restrictions.
The agent combines the two, making a response that fits the company’s rules and the customer’s time. The result is accurate and consistent. The right value becomes the right message, delivered in the right way.
How the agent stack changes with company size
We scale the agent stack by changes in the way intelligence is defined, integrated and managed, not by adding more tools or agents.
Small companies and scaleups are often the most ardent followers of martech and AI. They rely on tools to drive growth, which is reflected in both the high use of martech related to their integration approach.
More than half of SMBs (53.6%) rely on iPaaS solutions such as Zapier, Make or n8n to connect systems, compared to only 20% in enterprise environments. They are also using AI for more accessible entry points, with 32.1% of agents integrating with iPaaS or automation platforms, compared to only 8% of enterprises. This enables rapid testing, but distributes business logic across tools and workflows.
As complexity increases, the limitations of this approach become apparent. Mid-market companies are starting to formalize their stack, including iPaaS, pre-built integrations and select custom functionality. The understanding of the decision begins to move beyond the individual tools and a clear layer of purpose begins to emerge.
In business contexts, integration shifts to control and ownership. Nearly three-quarters (72%) rely on custom-built integrations, compared to 53.6% of SMBs. Businesses are also deeply embedding AI in assistants and social media (52% compared to 46.4% in SMBs), while also facing higher challenges. Coordination conflicts amount to 68% (compared to 41.1% for SMBs), governance barriers 48% (compared to 26.8%) and cost recognition 44% (compared to 17.9%).
Agent maturity is defined as how effectively organizations integrate systems and manage decision making across them. As companies grow, the challenge shifts from enabling intelligence to managing where and how decisions are made within a highly interconnected mass.
Sales for example
Marketing provides a useful example of how the agent stack evolves as organizations grow. This example also plays clearly within one stance.
Let’s look at two perspectives: overall stack maturity and size, and, in particular, one category: integration and tag management.
Overall maturity increases with company size. Small traders rated maturity 2.6, medium traders 2.8 and large traders 2.9. The size of the stack is also increasing, from about 60% of the largest stacks for small companies to full scale in enterprise environments.
Syndication tells a different story. This phase enables companies to collect customer data and connect systems, allowing data to flow across platforms, build custom workflows (AI) and make agent-driven decisions across the stack.
However, as stacks grow, interconnecting systems, managing data flow and maintaining consistency become more difficult, widening the gap between power and connectivity.
Small traders build tightly connected stacks focused on direct revenue impact. Ecommerce, CMS, CRM, customer service and marketing tools are often integrated into iPaaS solutions. Agents already support use cases such as product content generation, ad optimization and customer interaction. But the decision logic is distributed across devices, making it difficult to scale.
Intermediate sellers increase communication. As the volume of the campaign increases and there are more channels, the programs are more deliberately integrated. Agents become active in all workflows and decision logic becomes more transparent.
Great sellers they operate at a different level and build their stack around integrated systems of record, including CDP, CDW, PIM and MRM, which support large volumes of data and campaigns. Agents coordinate decisions across these systems, from pricing and promotion to personalization. At the same time, increased complexity makes it difficult to maintain control over decision-making.
In all three, the pattern is consistent. The stack not only grows, but also becomes difficult to manage. The shift is from allowing action to controlling decisions. That is the real change that the agent stack brings.
