Why Workflows and SOPs Make or Break AI in Public Affairs

Everyone wants the benefits of AI in public affairs, but very few want to talk about the boring bit: workflows and SOPs. Yet without them, the promise of transformation never materializes.

I’ve been running a number of workshops recently to share my insights on using SOPs. Personally, I like them. I think I’m naturally inclined to mapping out processes and structures and working with them systematically. But I’m also very conscious of their limits: the potential for errors, unnecessary steps, or misunderstandings. That’s why SOPs need to be iterated and adjusted — to improve efficiency, to avoid rigidity, and to continually refine them for real-world use.

Why workflows matter

Public affairs is not just about issues, messaging, or relationships. It is a process and a system. And processes and systems need structure.

That’s what a workflow provides: a map of the entire process from beginning to end. In public affairs, workflows can take many forms, for example:

• Identifying issues as they emerge, evaluating their impact, and cascading through to choices about resource allocation.

• Gathering and collating intelligence, then analyzing and communicating it in a structured way.

• Mapping stakeholders, understanding their influence and alignment, and deciding how to engage them.

• Formulating messages, testing them, and refining them based on feedback.

Without a workflow, AI use tends to be opportunistic: one person trying ChatGPT here, another experimenting with Gemini there, but no shared structure for how these experiments fit into the larger process.

Where SOPs fit in

A standard operating procedure (SOP) zooms in on one step within the workflow and defines it in detail.

Take drafting communication materials as an example. The workflow may cover the whole chain — monitoring, sourcing, drafting, validation, dissemination. An SOP could zoom in on just one of those steps — for example, how to use NotebookLM to move from a set of sources to a structured first draft.

But SOPs don’t always stop at a single step. In many cases, they cover a sequence of related steps within a workflow. For instance, an SOP might guide a professional through identifying relevant sources, analyzing them systematically, and then drafting them into a communication material. In that sense, SOPs can be both granular and modular, depending on the level of guidance a team needs.

SOPs make sure the work is repeatable, consistent, and transparent — whether done by an early adopter, a cautious experimenter, or a newcomer to the team.

One participant in a recent workshop made an interesting observation: SOPs don’t only serve the organization. They can also be tailored to the individual. Each professional brings different strengths and weaknesses — some excel at research, others at drafting, others at analysis. An SOP can act as a kind of assistant, guiding people through steps that don’t come naturally while reinforcing areas where they are already strong. In this sense, SOPs are not just about consistency; they are also about professional development and confidence-building.

SOPs as organizational frameworks

On the flip side, SOPs also serve the organization. Any organization that wants to adopt technology first needs to reflect on its own practice: what it does, how it does it, and what really matters. Only then can it identify where technology — AI or otherwise — fits into the process.

At the same time, SOPs are essential for meeting client expectations. If a client requires a specific output, the SOP becomes the framework that ensures the team can collate data, analyze it systematically, and then tailor the output into a format — often a template — that reflects the client’s unique needs and preferences.

The right SOP also depends on the conditions: what information and analysis the client already has, and where the process truly needs to begin.

Where technology fits

AI can certainly help in the preparation of SOPs themselves — for example, by drafting the first outline of a procedure or suggesting steps that can later be refined by professionals. But its real power lies in being embedded within SOPs, supporting specific tasks inside a workflow.

Each choice about which tool to use should be task-dependent. For instance:

• Perplexity might be chosen for sourcing and discovery, because of its ability to find and classify documents.

• NotebookLM might be used for structuring and drafting from a set of sources.

• Claude might be applied for validation, proofreading, and refinement.

In this way, SOPs don’t just structure the process — they also guide decision-making about which tools are suitable for which tasks. That ensures AI isn’t used in a scattershot way, but is matched deliberately to the needs of each step.

From sprint to practice

Workshops are useful for introducing SOPs and showing how they run from beginning to end. In that setting, the exercise feels like a sprint: fast, focused, and illustrative. But to really internalize an SOP, professionals need to apply it.

That means experimenting with the steps, iterating and adjusting them, and seeing how the outputs change. It also means reviewing and challenging those outputs — asking what works, what doesn’t, and what contribution each person can make to further refine the process.

Only through that cycle of application, adjustment, and reflection does the SOP move from being a generic framework to becoming part of the team’s actual practice. That’s when workflows and SOPs stop feeling rigid and start becoming tools for both confidence and creativity.

Final reflection

No, workflows and SOPs will ever be the most glamorous part of public affairs. But they are the hidden infrastructure that makes everything else possible.

If we want AI to support — rather than disrupt — public affairs practice, we need to think systematically:

Workflows map the process.

SOPs structure the steps.

Technology supports the SOPs.

It may not sound exciting, but it is transformative.