In today's era of relentless AI advancement, firms are recognizing AI's potential to transform workflows, enhance client service and streamline operations. However, with countless options and overwhelming market noise, many ask: Where do we start? What's relevant? How do we determine what's worth our time and investment?
To help navigate these questions, I present the AI Adopt, Test, Monitor Framework, developed from my extensive conversations and work with Top 500 CPA firms on AI and innovation.
This framework is designed to guide CPA firms through the practical implementation of AI technologies. By categorizing opportunities into what you can adopt now for immediate value, test to explore potential, and monitor as emerging opportunities, the AI ATM Framework will put more cash in your pocket by improving productivity and accelerating revenue growth. (As an aside, this may be my first and last attempt at a dad joke in an Accounting Today article).
Adopt: Proven technologies you can implement now
1. Microsoft Copilot
Microsoft Copilot has come a long way and while it still has a way to go, it is now a viable option for firms looking to leverage generative AI. At current pricing, it offers a break-even investment by saving time equivalent to a fraction of an hour per employee per month, with potential productivity gains far exceeding costs. Use cases include:
• Drafting emails: Helps junior staff create polished, professional emails in Outlook, enhancing client communication.
• Meeting summaries: Generates detailed recaps for missed meetings via Teams integration, making it easier to catch up.
• Presentation creation: Simplifies PowerPoint creation, from templates to design.
Both the general chat assistant and embedded Office365 Copilot versions are a good way to get your team to adopt generative AI technology at work. With proper training and engagement, your team members will be well on their way to incorporating AI assistant into their daily work.
2. ChatGPT Plus for non-sensitive data for power users
OpenAI's ChatGPT remains a leader in generative AI technologies. While other models like Anthropic's Claude, Google's Gemini and open-source Llama excel and sometimes are superior in specific domains, ChatGPT offers an accessible gateway into existing cutting-edge AI and provides direct insight into future Microsoft Copilot and generative AI implementations.
Some reasons to consider premium ChatGPT licenses include:
• Advanced models: Access to the latest foundation model, like ChatGPT's o1 model when it first came out and introduced advanced reasoning.
• Different UI/interaction methods and superior performance: Beyond text, ChatGPT offers its Canvas UI and voice chat mode. You may find duplicate features in Copilot, and ChatGPT performs differently (i.e., I find voice transcription to be more accurate in ChatGPT vs. Copilot).
Combining Microsoft Copilot for general employee use with ChatGPT premium licenses for tech savvy team members positions your firm to take full advantage of today's AI tools while understanding where the technology is headed.
3. AI cybersecurity training
While this is strictly not a "technology", cybersecurity training specifically around generative AI risks is a key component of successful firm adoption of AI tools. Training ensures your firm uses AI safely while protecting sensitive information. Equip your team with the knowledge to recognize AI-driven phishing scams and other sophisticated attacks will go a long way in empowering your team to utilize AI to its fullest extent.
TEST: Technologies worth experimenting with
The "Test" category includes technologies with great promise that require thoughtful experimentation to ensure they're a fit for your firm.
1. Internal AI assistants
Firms today are deploying internal AI assistants and reporting success. For example, at Digital CPA in 2024, Schellman CEO Avani Desai discussed AskShelly, the firm's internal HR AI assistant powered by GPT technology. It references specific internal documents to answer HR-related questions, improving employee efficiency. Other use cases include:
• Answering internal operations questions (i.e., IT, and onboarding); and,
• Providing guidance on firm-specific stances regarding regulatory questions.
With Microsoft's Copilot Studio, a low-code custom GPT builder, creating internal AI assistants are now more accessible than ever and, from my testing, hold promise. However, successful implementation requires a robust foundation of:
• Standardized data: Consistent and well-organized information;
• Correct document storage: Stored in easily accessible repositories like SharePoint;
• Ongoing maintenance and process: Regular maintenance of the underlying documents and data to remain accurate and trustworthy.
Without these foundational elements, trust in AI assistants can erode, reducing adoption and value. Given the required investment of not just technology but potentially of firm processes and systems transformation, this may not make sense for you today.
Monitor: Keeping an eye on emerging trends
1. External-facing AI assistants
Imagine enabling clients to interact directly with an AI assistant for simple questions about their accounts, tax filings or financial planning. While it seems promising, this technology involves more complex data integrations, advanced security measures and systematic updates to maintain accuracy. Most firms should monitor this trend until solutions become more accessible to successfully implement.
2. AI agents
AI agents represent the next frontier in generative AI, capable of autonomously handling tasks like drafting proposals, managing schedules or coordinating workflows. While they look promising, these tools are still in early development.
Adopt, Test and Monitor AI for success
Adopting AI doesn't have to be overwhelming. By adopting proven tools, testing emerging solutions, and monitoring trends, you can create a clear roadmap for your AI journey.
• Adopt: Implement Microsoft Copilot and ChatGPT Plus to improve employee productivity and kickstart a culture of AI adoption.
• Test: Explore custom GPTs via Copilot Studio or Azure deployment for narrow, internal AI assistant solutions.
• Monitor: Stay informed about technologies like client-facing AI assistants and AI agents, ready to act when the time is right.
The question isn't whether to start with AI, it's how. The AI ATM framework equips you to chart a confident path forward and position your firm for success.
