Architecting Agentic AI: How SDKs, Scaffolding & Frameworks Are Different
Building with Agentic AI? Are You Using an SDK, Scaffolding, or a Framework?
Recently I spoke to a tech executive who stated that their organisation do not want to be tied a technology stack…
The explosion of AI related SDKs (OpenAI, Anthropic, X.ai, Cohere, etc.), Frameworks and scaffolding make building agentic experiences more accessible than ever.
But you’re inevitably going to be tied to their underlying abstractions and conventions.
Yes, you’re always tied to the abstractions and conventions of whatever SDK, scaffolding, or framework you use — just to different degrees:
SDKs
Even though SDKs are lightweight, you’re still aligning your code to API contracts, data formats and the mental model of that provider.
For Example, OpenAI’s function calling structure or built-in libraries shape how you design your prompts and workflows.
Choosing the right level of abstraction is less about avoiding “lock-in” and more about choosing your constraints wisely.
An SDK is a broader set of tools, libraries, APIs designed to help developers build applications for a specific platform, ecosystem, or technology.
Scaffolding
With scaffolding, you inherit structure and patterns that influence your development decisions. For example, how to manage AI Agents, thier tools, routing, or UI state.
You can change it, but there’s still an initial frame of thinking you start with.
Scaffolding refers to a set of pre-generated code, templates, or project structures that provide a starting point for building an application.
It’s typically used to automate the setup of repetitive or boilerplate code.
Frameworks
I guess here, you’re most tightly coupled — because the framework often dictates the entire app’s architecture, lifecycle, and how logic flows.
LangChain AI Agents use a specific chain-of-thought design, memory handling and even a declarative way to add tools.
Whether it’s SDKs, scaffolding or full-blown frameworks, you’re always buying into a set of assumptions.
The trade-off is:
Flexibility vs Speed
Control vs Productivity
Custom design vs Convention-based development
Prototyping / demo vs Production roadmap
A framework is a reusable, structured set of libraries, conventions and tools that provide a foundation for building applications. It enforces a specific architecture or design patterns.
Frameworks can range from pro-code to no-code primitives which serves as an abstraction layer between underlying foundational technology and the builder UI.
A key challenge for framework creators is to provide a continuous and smooth technology roadmap for their customers, as the technology landscape unfolds.
In Conclusion
In principle, SDKs, scaffolding and frameworks differ in their scope, purpose and role in the development process.
SDKs enable integration with specific technologies.
Scaffolding automates project setups.
Frameworks provide a structured foundation for building apps.
When building generative apps, you’re likely to use all three in complementary ways, but understanding their distinct roles helps you choose the right tool for each stage of development.
Chief Evangelist @ Kore.ai | I’m passionate about exploring the intersection of AI and language. From Language Models, AI Agents to Agentic Applications, Development Frameworks & Data-Centric Productivity Tools, I share insights and ideas on how these technologies are shaping the future.