top of page
  • Rachel Barton

Getting the Best Service from Generative AIs: The SERVE Prompt Framework

Updated: Jun 22

Rachel Barton, Speech and Language Therapist, 2nd June 2024

man and robot sitting in a field chatting


The rapid advancement of generative AI technology has unlocked extraordinary possibilities for speech and language therapists (SLTs) to elevate their practice. AI models like ChatGPT can assist with a myriad of tasks, from generating personalised therapy materials, streamlining administrative processes to conducting insightful analysis. But to truly harness the transformative power of these AI assistants, a structured approach to prompting is essential.

Effective prompting is the key to unlocking the full potential of AI language models. Well-crafted prompts lead to accurate, relevant, and tailored outputs that align with our professional goals and standards. Conversely, vague or ambiguous prompts can result in suboptimal or even misleading responses. The good news is that effective prompting lies in effective communication—an area where SLTs excel. Speaking to the AI with the same clarity and detail you’d use for a new employee generally results in excellent outputs.

To support SLTs in this endeavour, I have developed the SERVE Prompt Framework. This framework provides a systematic approach to crafting prompts, ensuring clear communication with AI assistants, and enabling you to get the best service from these powerful tools. By breaking down the prompting process into clear steps, the framework helps you convey your requirements precisely, minimising ambiguity and enhancing the relevance of AI-generated responses.

The SERVE Prompt Framework

SERVE is an acronym that outlines the five essential steps in the prompting process:

1. Set the scene

The first step is to clearly establish the context for the AI assistant. Define your role as an SLT and clarify your target audience or explain a client's specific needs. Provide relevant background information necessary for the AI to fully understand the situation. Imagine you are briefing a colleague on a collaborative task —what would you say to help them grasp the context? Setting the scene with comprehensive information upfront helps the AI generate more contextually appropriate and tailored responses. Remember, it's essential to exclude any client-identifiable information.

2. Explain the task

Once the scene is set, provide detailed instructions to the AI about the task. Clearly articulate the desired outcome, whether it's creating therapy materials, drafting a letter, or planning a training session. Use specific language and provide examples to illustrate your requirements. Using profession-specific terminology rather than lay terms can be more effective e.g. saying ‘phoneme’ rather than ‘sound’ generates better speech resources. Follow up your instructions by asking the AI, "what else do you need to know before starting this task?" This encourages the AI to ask clarifying questions or request further details, ensuring a more precise output.

3. Refine the response

After receiving the initial response from the AI, it's crucial to review it critically. Provide the AI with feedback on areas that require clarification, correction, or additional details. Ask the AI how it can improve the response further. This back-and-forth iterative process of refinement, adjusting the tone, fine-tuning the wording, and adding details allows you to shape the output until it meets your standards and aligns with your goals.

4. Verify the facts

Generative AI models can sometimes produce outputs that seem plausible but may contain factual inaccuracies or hallucinations. To ensure the integrity of the information, it's crucial to verify key facts and claims by cross-checking them against reputable and independent sources. Consult authoritative resources, research studies, or professional guidelines where necessary. This step helps you catch and correct any potential errors or misinformation.

5. Evaluate professional integrity

 As an SLT, it's essential to ensure that the AI-generated content adheres to professional standards and aligns with your values and ethical principles. Evaluate the output critically, considering potential biases, cultural sensitivity, and appropriateness for your client's needs. If the content does not meet these standards, provide feedback and refine it further or consider generating a new response.

woman and robot walking towards a beautiful view surrounded by animals

Next Steps

By following the SERVE Prompt Framework, you can effectively communicate with AI language models, iteratively refine the outputs, and ensure that the final product meets the highest professional standards. Working with AI tools in this way can streamline your workflow, enhance your outputs, and ultimately improve the outcomes for your clients. As the field of generative AI evolves, I hope this framework will be a valuable resource, helping you navigate this landscape in a safe, ethical and creative way.

Want to learn more?

Look out for further blogs in the coming weeks or book onto one of my AI workshops:

or for more information visit:

Acknowledgement: This blog was developed with AI assistance from Open AI’s ChatGPT4-o &  Anthropic’s Claude AI, images were created by DALLE in ChatGPT.

254 views0 comments

Recent Posts

See All


bottom of page