Considerations and discussion
Where does SaaS fit?
Industry experts actively debate on how agents influence the software as a service (SaaS) landscape. While it's true that agents are changing software for many systems, it's a stretch to suggest that agents make delivery models obsolete. Some SaaS providers will likely be disrupted by adopting agents, and some may entirely rethink their value proposition, by leaning into an agent as a service (AaaS) model. Others may strike a balance by selectively introducing agents to address specific needs.
This topic is interesting because adopting the best SaaS principles may represent the next evolution of SaaS. This might mean that SaaS is going way, or it might mean that the foundational principles of SaaS are being packaged and realized in an agent-based model. It's probably less important to decide where the terminology ultimately lands, but it seems unlikely that SaaS as a concept will disappear. It's more likely that agents will shape the SaaS footprint.
Ultimately, we must decide which strategies can be applied to AaaS, which means enabling organizations to adopt agentic architectures and business strategies so that providers can maximize the efficiency, value, and impact of their agentic systems. Agents aren't black boxes. Agents consume resources, scale operations, depend on data, and generate costs—all factors that providers must address. Agent providers must evaluate how multi-tenant principles can shape service offerings and optimize operational models.
Discussion
The agentic landscape continues to evolve with designs varying based on domains, intended use cases, and target industries. Part of this evolution includes further refining our view of strategies, patterns, and tradeoffs that architects consider as they design and build agents.
A comprehensive agent strategy must align with both business and technical objectives. This includes defining target markets and personas, establishing pricing and resource management strategies, and determining how agents fit into larger systems. These considerations are particularly important when delivering AaaS, where scale, cost efficiency, and innovation are primary goals.
Operational capabilities are equally important. The environment must support monitoring of agent activity, health metrics, and usage patterns. This becomes more complex in multi-agent systems, where operations must be coordinated across independent agents.
Overall, this discussion of agents only scratches the surface of the various architectural considerations that could be part of agentic systems. Beyond selecting appropriate tools, frameworks, and LLMs, success depends on creating an architecture that meets business requirements for scalability, efficiency, deployment, and multi-tenancy.
