The buzz around Artificial Intelligence is deafening. It’s like 2011 when “big data” became a thing, or 2006 when we started hearing about “the cloud.”
For many business leaders, AI shows promise, but they’re still trying to figure out how to use it profitably for their business. The CREATE Framework helps you work through the business problem, requirements, build, rollout, and ongoing evaluation of an AI agent.
Every day, there's a new tool, a new capability, a new headline promising revolutionary change. For marketing leaders and their teams, the pressure is immense. How do you leverage AI effectively while juggling strategic planning, team management, data analysis, and execution? How do you adopt AI without employees wondering if their jobs are at risk? How do you find the time, optimize resources, and stay ahead of the curve?
Other organizations jump headfirst into AI, tasking teams with "finding uses for AI" or building agents without a clear objective. This often leads to fascinating tech demos but little tangible business value, wasted resources, and ultimately, disillusionment.
The truth is, AI is a powerful tool, but it's not the starting point. The starting point should always be your most pressing business challenges.
At Whittington Consulting, we've created our own custom AI agents to use in our digital marketing agency, but we didn't start with the technology. We start with our specific business needs. To ensure AI delivers real results (for us, that's efficiency and strategic insight), we developed The CREATE Framework.
What is the CREATE Framework?
The CREATE Framework is our proven, collaborative process for developing custom AI agents that are perfectly aligned with business goals. It ensures we focus on solving specific business challenges first, determining if and how AI can be the most effective solution.
CREATE stands for:
- Challenges Identification
- Refine Use Cases
- Establish Requirements
- Assemble Solution
- Test and Deploy
- Evaluate
The CREATE Framework provides a structured approach to building impactful custom AI agents.
Let's break down each step:
C: Challenges Identification
This is the crucial starting point. Before we even think about AI features, you need to uncover your business challenges.
- Purpose: Identify the core business challenges, bottlenecks, skills gaps, or opportunities where improvement is needed.
- Key Questions To Explore:
- What are the most pressing business issues or challenges you and/or your team are facing today?
- What critical skills are absent or not highly developed on your team?
- What tasks do you wish you had more time for? What do you wish you could do less of?
- Where are you struggling to be effective or do enough?
R: Refine Use Cases
Once you understand the business challenges, it’s time to explore specific solutions. These solutions might include AI, but there are lots of ways to solve business challenges without AI. Not every challenge needs an AI solution, but many repetitive, data-intensive, thought-intensive or analysis-heavy tasks are prime candidates.
- Purpose: Determine specific, high-impact use cases for a potential AI agent that directly align with the identified business challenges.
- How We Prioritize: We often use a simple quadrant chart, plotting tasks based on the dislike or repetitive nature of a task or skill deficit versus the business impact. Tasks that are highly disliked (or where skills are lacking) AND have a high potential business impact are often the best candidates to tackle first with an AI agent.
- Output: A shortlist of well-defined, valuable use cases for AI.
E: Establish Requirements
With clear use cases defined, the next step is to map out exactly what the AI agent needs to do and what it needs to work with.
Think about where the data the AI will need actually lives. You don’t always have to pull reports, download files, and feed everything to the AI yourself. With the right connections in place, an AI agent can use Model Context Protocol (MCP) servers to fetch information from multiple data sources for you. Identify the sources it needs, whether your chosen platform can connect to them, and what access it should have before you build.
- Purpose: Establish the functional requirements, necessary data inputs, desired output formats, and any existing processes (like SOPs) the AI needs to understand.
- Key Questions To Explore:
- What specifically should the custom AI agent do?
- What format should the results be in (e.g., summary, bullet points, table)?
- Are there existing Standard Operating Procedures (SOPs) that can be used to train the AI?
- What data exists (documents, spreadsheets, website content, transcripts) that can provide context or be processed by the AI? Which sources can the agent connect to, and which will need to be provided manually?
A: Assemble Solution
This is where the technical build happens, leveraging powerful, existing AI platforms securely. We prefer using existing AI models like OpenAI's ChatGPT, Google's Gemini, or Anthropic's Claude to build AI agents.
- Purpose: Build and configure the custom AI agent based on the defined requirements and use cases.
- Technology: Use readily available AI platforms like OpenAI's ChatGPT, Google's Gemini, or Anthropic's Claude. Build within your own account, and review the platform’s data-use settings and permissions for connected services before sharing proprietary information. Privacy protections depend on the plan and configuration you choose.
- Key Considerations:
- What AI platform is best suited for the specific use case? (Some are better for analysis, others for creative tasks, etc.)
- How will the necessary data be accessed and potentially updated?
T: Test and Deploy
An AI agent is only useful if you and/or your team use it effectively.
- Purpose: Deploy the agent within your organization and ensure users are comfortable and proficient.
- Our Approach: This isn't just a hand-off. We typically will provide guided roll-out & training. We ensure you and your team know exactly how to use the agent effectively, providing clear instructions and best practices, even if you're new to AI.
- Key Questions To Address:
- Who needs access to the agent?
- Who needs to be trained?
- How long will any pilot or initial testing period last?
E: Evaluate
Continuous improvement is key. We establish ways to gather feedback and refine the agent over time. As models change, your AI agent will change as well.
- Purpose: Establish a feedback loop to iterate and refine the AI agent based on real-world usage and results.
- Key Considerations:
- How will user feedback be collected systematically?
- How will updates or future versions be communicated and rolled out?
What Can a Custom AI Agent Achieve? (Real-World Examples)
Following the CREATE Framework ensures we build agents that tackle real needs. Here are a few examples of custom agents we've built:
- "My Co-Founder": A ChatGPT plugin designed to be a trusted business partner, using company context, offering structured, empathetic advice for tough business decisions and helping leaders identify blind spots. (Solves: Need for strategic sounding board, identifying decision risks and blind spots).
- "Customer Persona Generator": Guides users through expert questions to generate comprehensive, best-practice ideal customer profiles (ICPs) based on inputs like website data, interview notes, or existing research. (Solves: Time-consuming persona creation, inconsistent ICP quality, need for deeper customer understanding for marketing/sales).
- "Meeting Notes Summarizer": Transforms transcripts, documents, or even photos of notes into concise, formatted meeting summaries. (Solves: Time spent summarizing meetings, extracting key actions/decisions).
- "Marketing Agency CFO": Analyzes Profit & Loss statements against industry benchmarks, identifying key ratios and suggesting areas for financial improvement. (Solves: Need for quick financial health checks, improving financial acumen).
- "Web Page Analyzer": Reviews a URL while acting as a specific ICP at a specific buyer journey stage, scoring it and recommending content/call to action improvements. (Solves: Ensuring content relevance, optimizing pages for conversion, lack of team expertise in detailed content audits).
These agents don't just perform tasks; they enhance team capabilities, sharpen strategic thinking, boost efficiency, unlock data insights, and ultimately drive better business outcomes.
Ready to Solve Your Challenges with AI?
Stop chasing AI hype and start solving your real business problems. If you're a marketing or business leader feeling overwhelmed and looking for ways to gain efficiency, unlock strategic insights, or fill skills gaps, a custom AI agent, built the right way using the CREATE Framework, could be the solution you need.
If you'd like to explore if a tailored AI agent can help you and your team achieve more, or you'd like to explore getting your own custom AI agents to solve challenges unique to you or your business, feel free to reach out here.
Last updated: October 2026