Implementing Generative AI Safely and Successfully

Implementing Generative AI (GenAI) involves balancing analysis, cross-functional teamwork, and clear guidance to secure its success, alongside training and outreach to align stakeholder expectations.

‍

What is Shadow AI?

Shadow AI is when GenAI enters an enterprise via employees who want to use it to improve their productivity without IT approval and before any official plans for implementing GenAI, bringing with it data security and other risks. It’s often rampant in the early stages of GenAI maturity and even after enterprises are still maturing in their AI governance.

‍

Implementing GenAI

Here’s an overview of how to implement GenAI inside your enterprise for the first time:

‍

1. Build a business case for GenAI

One of the most painful realities of GenAI is that the hype can be intoxicating. Decision-makers and stakeholders may settle on GenAI as part of their project portfolio, even if there’s no business case. There may also be managers and employees who are a bit more pessimistic and need to realise how AI fits into the business, which serves as another audience for your case.

‍

The business case is one of the best antidotes to GenAI hype. Your GenAI business case should include the following:

The business problem you intend to solve with GenAI
A step-by-step breakdown of how GenAI will help solve the business problem
Potential benefits and outcomes for the business that come from implementing GenAI

This step isn’t about delivering an MBA-level analysis. Instead, it’s about telling a verifiable story with facts that business stakeholders and employees alike can read hype-free about the benefits GenAI could deliver for your organisation.

‍

2. Form your GenAI team

Rolling out GenAI requires a cross-functional project team, including:

Security and compliance responsible for securing employee access to a GenAI solution
IT staff who’ll work with security and compliance on the technical implementation steps
Executive sponsor for the initiative who controls the budget
User representation from the business side

‍

3. Rescope key workflows to include GenAI

Whether it’s an editorial workflow or a process your company uses for customer outreach, you want to document the current process. After that, look for elements in the process that can benefit from the application of GenAI. For example, your field employees to write trip reports as part of a process. It’s time-consuming for some of your busier staff. Rescoping the process for GenAI means inserting GenAI touchpoints into the process, such as a custom prompt for drafting and editing the report for grammar, impact, and consistency.

4. Do your due diligence and testing on GenAI solutions

Performing your due diligence on AI vendors and tools is a must. You want no AI vendor to define the role of their technology in your enterprise. Pilot testing or a “bake-off” between two to three potential GenAI tools on a business use case such as editing or drafting documents.

‍

5. Analyse and clean your data

Your corporate data is foundational to the success of your GenAI implementation. You’ll want to spend the upfront time analysing and cleaning the data you’ll be feeding into your new GenAI to prevent AI hallucinations and speed up information processing and retrieval.

‍

6.Over-communicate to your internal users

Make every effort to communicate, or even over-communicate, with your users about your GenAI strategy and the status of your implementation. Take the extra steps to create job aids and fact sheets outlining the benefits of introducing GenAI into your organisation and its workflows.

‍

7. Go live with GenAI

Taking GenAI to live for your users can be as simple as adding it to your single sign-on (SSO) as an option for authorised users and leaving it that. You should also provide your users with job aids, cheat sheets and best practices during this step to ensure their success.

‍

8. Practice continuous improvement

Implementing GenAI tools requires ongoing attention, focusing on continuous improvement as new GenAI models and cybersecurity challenges emerge. There’s also the prospect that your business users may find new and intriguing uses for GenAI, especially when interacting with corporate data implementing GenAI as their non-programmatic tool of choice.

‍