3 Questions are on the table for AI
In the year and a half that I’ve been working with organizations to get benefits of AI’s potential, I’ve seen these three questions repeatedly arise:
- How do I get started with AI?
- Where can AI deliver the most value?
- How do I scale and operationalize AI?
These questions are crucial because they address the core challenges of implementing AI in business. Starting correctly ensures a solid foundation, understanding value helps focus resources effectively, and scaling is necessary for realizing the long-term benefits across the organization.
I’ve seen projects succeed by concentrating on practical, actionable steps that aren’t just about technology but about creating a meaningful impact for your business.
Today, I’ll share insights from my experiences working with various industries, from retail to finance, to make your AI journey more manageable and effective.
It’s a collective effort. I am sharing my experience and learning, and you share yours as an AI tech community.
1 - How to Start with AI in Your Business:
Getting started with AI involves laying a solid foundation.
In my experience, successful projects focus on practical, actionable steps that aren’t just about technology but about creating a meaningful impact for the business.
Here are actionable steps to take:
Identify Business Challenges: Define specific problems that AI can address. Don’t deploy AI for technology’s sake; use it to solve real issues. Example: A customer support center wants to reduce response times. Large Language Models (LLMs) like Oracle Generative AI can automate answers to common customer inquiries, providing instant responses and freeing up human agents for complex cases.
Assemble a Cross-Functional Team: Include members from IT, operations, and the impacted department to ensure holistic planning and execution.
Action: Create a small CoE and a task force, including representatives from data management, IT, and business units, to ensure collaboration and buy-in for AI integration. An earlier article, "Chief AI Officer’s Playbook: Executing the First 90 Days," provides more in-depth coverage of establishing a team.
Select a Practical Use Case: Use available data to select a feasible use case. You can get an in-depth framework from 'How to validate your first Generative AI Use Case'
Examples include automating routine processes, improving demand forecasting, or enhancing customer support.
Action: Start using an LLM to draft internal reports or answer routine customer questions. Later, the system can be scaled to handle more complex queries.
Perform a Gap Analysis: Assess your current data infrastructure and determine what needs to be improved or added.
Action: Conduct a data quality audit to identify gaps and ensure data is ready for AI implementation. For instance, if using LLMs, ensure the dataset includes relevant and high-quality training information.
Develop a Proof of Concept (PoC): Collaborate with AI experts or technology partners to create a PoC and follow the framework outlined in 'A framework for selecting an AI use case'.
Action: Limit the PoC to a 3-month timeframe to demonstrate tangible benefits, such as using an LLM to automate document drafting, thereby saving hours of manual effort weekly.
2 - Identifying Where AI Unlocks Value From:
In my early experiences working with industries like healthcare, retail, public sector, media, and finance, I’ve learned that the value of AI is unlocked by making meaningful improvements to key business areas.
Here’s how to pinpoint where AI can deliver maximum value, you can also refer earlier article Lost in AI use cases, don't forget to put your AI Strategy first and align it with the business:
Enhance Decision-Making: Use AI for data-driven decisions.
Real-World Example: A retail company uses LLMs to analyze customer reviews and extract critical insights about product performance, helping the marketing team adapt campaigns based on honest customer feedback.
Reduce Costs: Identify operational inefficiencies AI can address.
Action: Implement AI-based customer service AI assistants/chatbots powered by LLMs. These bots can handle up to 80% of routine inquiries.
Boost Productivity: Automate repetitive tasks to free up human resources.
Example: A legal firm uses LLMs to review contracts, highlighting key terms and flagging risks, which reduces the time spent on manual contract reviews by 50%.
Improve Customer Experience: Personalize customer interactions using AI-driven insights.
Action: Use LLMs to provide personalized product recommendations based on customer history and preferences, increasing engagement and conversion rates.
Establish clear KPIs to measure value.
For example, Customer Satisfaction: Increase satisfaction scores by 15% within six months using AI-powered chat support.
Operational Efficiency: Reduce average contract review times by 50%.
3 - Scaling and Operationalizing AI:
Scaling AI involves moving beyond isolated pilots to organizational-wide adoption. I’ve discovered that scaling AI successfully, even in these early stages, requires strong leadership support and a structured approach. Follow these actions:
Leadership Buy-In: Ensure AI initiatives are strategic priorities supported by top management.
Action: Present PoC results to executives to get formal approval and funding. Highlight how LLMs improved productivity by reducing report generation time by 40%.
Train Staff for AI Adoption: Employees must understand AI tools and outputs.
Action: Develop training workshops to educate staff on how to work with AI-driven insights.
For example, train customer service reps to handle escalations effectively when LLM-powered chatbots pass on complex queries.
Integrate AI into Existing Systems: Instead of separate deployments, integrate AI into core systems for seamless use.
Example: Integrate a Generative AI into the existing ecosystem, get simplified architecture for your Oracle SaaS applications, 'Simplified Architecture to take up Generative AI in the Cloud Applications'
Establish Governance: Set up rules for data usage, model validation, and ethical considerations to ensure responsible scaling.
Action: Form a governance committee responsible for overseeing AI practices, ensuring the ethical use of LLMs, and mitigating risks such as bias.
Iterative Scaling: Expand AI use cases one step at a time, incorporating feedback from each rollout.
Real-World Example: An insurance company uses LLMs to automate policy summaries in one branch, refine the model based on user feedback, and then scale it across all branches.
By following these actionable steps and focusing on specific, measurable outcomes, executives can ensure that their AI journey is routine and genuinely transformative for their business.
Leveraging the latest advancements in Large Language Models can give you a competitive edge, just as I’ve started to see in organizations I’ve worked with over the past year and a half.
My journey so far has taught me that the key to successful AI adoption is not about chasing trends but understanding your business and where AI can make an impact.
