From Manual Coding to AI Agents: A Step-by-Step Guide to GAITA

If you’ve ever spent weeks manually coding transcripts for qualitative research, you know how exhausting and time-consuming it can be. Thematic analysis is essential for uncovering patterns and insights, but traditional methods often involve repetitive tasks that slow down your progress.

Enter GAITA (Guided AI Thematic Analysis)—a structured approach that combines the rigor of qualitative research with the efficiency of AI. This method doesn’t replace the researcher; it empowers you to work smarter, not harder.

What is GAITA and Why It Matters

GAITA is an AI-assisted framework designed to streamline thematic analysis while maintaining academic integrity. Developed from recent research and published in reputable journals, GAITA uses AI tools like Copilot to assist with coding, clustering, and theme development.

Why GAITA is a game-changer for PhD students and researchers:

• Saves time: Automates repetitive coding tasks.
• Maintains rigor: Follows structured steps aligned with qualitative best practices.
• Supports triangulation: Works with multiple data sources like interviews, field notes, and reflective journals.

Keywords to remember: GAITA framework, qualitative research AI, thematic coding automation.

The 4-Step GAITA Workflow

GAITA organizes thematic analysis into four clear steps:

Step 1: Data Familiarization

• Summarize interviews.
• Highlight key ideas in bullet points.

Step 2: Preliminary Coding

• Identify and refine codes.
• Organize in a table format (Code, Description, Quotation).

Step 3: Template Formation & Refinement

• Cluster codes into broader categories.
• Develop subcodes and merge overlapping tables.

Step 4: Theme Development

• Generate themes from clusters.
• Revise theme names and descriptions for contextual relevance.

This structured approach ensures consistency and reduces cognitive overload.

Practical AI Prompts and How to Use Them

One of GAITA’s strengths is its prompt-driven workflow. Here are examples you can use in Copilot or similar tools:

Step 1: Data Familiarization

Prompt 1.1:
“Can you please provide a summary of this interview [Interview number 1 with Barbara]?”
Prompt 1.2:
“Highlight the key ideas of the interview in three key bullet points.”
Prompt 1.3:
“Can you please provide me with a 400-word summary of key ideas from all the interviews in the uploaded file?”

Step 2: Preliminary Coding

Prompt 2.1:
“You’re a qualitative research assistant. You will help me identify the relevant codes from the following transcripts in response to this question: ‘How do these people connect to Down East?’…”
(Includes instructions for table format: Code, Description, Quotation)
Prompt 2.2:
“Please study the following list of initial codes developed based on the interview with Barbara. We will use this list to code relevant information…”
Prompt 2.3:
“Now, apply the revised coding framework based on the interview with Barbara to the following interview in response to ‘How do these people connect to Down East?’…”
(Includes detailed instructions for table structure and editing)

Step 3: Template Formation and Refinement

Prompt 3.1:
“From the above refined list of codes, please group codes into clusters (i.e., more abstract codes) in response to the question: ‘How do these people connect to Down East?’…”
Prompt 3.2:
“Now I need you to further develop the coding table above by generating subcodes for each code based on the quotations I give you…”
Prompt 3.3:
“The two tables are still somewhat overlapping. Please create one new table combining two coding tables. In this table, please build the coding structure as follows…”

Step 4: Theme Development

Prompt 4.1:
“From the above table of clusters, please generate themes across the clusters and codes in response to the question: ‘How do these people connect to Down East?’…”
Prompt 4.2:
“Can you revise the themes’ names and descriptions by: (1) extending the theme names so that they are meaningfully connected to the social context of Down East; (2) ensuring that the themes and their descriptions answer the question…”
Prompt 4.3:
“From the themes above, please produce an overarching theme in response to the question ‘How do these people connect to Down East?’”

These prompts help you automate coding while keeping control over interpretation.

---

5. Tips, Best Practices, and Next Steps

• Keep prompts general so they can be reused for different projects.
• Combine multiple transcripts for triangulation (e.g., interviews, reflective journals, field notes).
• Maintain ethical standards: AI assists, but researcher judgment is essential.
• Next step: Download the GAITA prompt library or create your own AI agent for thematic analysis.

Sample GAITA Agent Prompt:

Goal:
You are a qualitative research assistant helping me conduct a full thematic analysis of interview transcripts using the GAITA (Guided AI Thematic Analysis) framework adapted from ChatGPT in thematic analysis: Can AI become a research assistant in qualitative research? | Quality & Quantity

Context:
We are analyzing interview data to answer a specific research question. You will follow a four-step process:

Data Familiarization
Preliminary Coding
Template Formation and Refinement
Theme Development

Use the ACTOR prompting structure:

Actor: You are a qualitative research assistant.
Context: I will provide the research question and transcript(s).
Task: You will perform the assigned step in the GAITA process.
Output: Always return results in a clear table or bullet-point format.
Reference: Use the transcript(s) I provide—either pasted or uploaded.

Source:
You will work with either pasted interview text or uploaded transcript files. I will specify the format each time.

Expectations:

Provide summaries, codes, clusters, and themes in structured tables.
Explain your reasoning briefly for each output.
Ask for clarification if any input is unclear or incomplete.
Do not hallucinate or fabricate content—only use the data I provide.