Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Mastering the power of generative AI: Ask these eight questions before your enterprise takes the leap

Just because the road to new tech implementation is littered with potholes and speed bumps doesn’t mean enterprise leaders shouldn’t make the journey.

Chatbot on a smartphone

[Images: panuwat / Adobe Stock]

Profile photo of Louis Têtu

BY Louis Tetu

Nothing tells the story of the unbridled enthusiasm for generative AI quite like the user adoption stats for ChatGPT. The AI-powered chatbot reached one million users just five short days after launch.

Given its meteoric growth, it’s no wonder enterprise leaders are excited about the promise of GenAI to deliver highly personalized, conversational, and advisory experiences to their customers, employees, and other stakeholders. But if we’ve learned anything from past tech hype cycles, it’s that once the dust settles, the enterprise is often left with a host of more complicated realities to navigate.  

EIGHT QUESTIONS EVERY CIO SHOULD BE ASKING

Just because the road to new tech implementation is littered with potholes and speed bumps doesn’t mean enterprise leaders shouldn’t make the journey. A sound strategy involves driving GenAI adoption with complete visibility into where the hazards lie. Answering the eight questions that follow is a great place to start the journey.

1. CAN MY BRAND AFFORD TO HALLUCINATE?  

ChatGPT doesn’t have much notion of veracity. At its foundation, it’s connecting themes together to generate an answer. Sometimes those answers turn out to be factually incorrect or even entirely made up. When the stakes are low, such hallucinations may not be that consequential. In some situations, consumers can’t tell the difference between content generated by AI versus a human representative.

The stakes are much higher for many enterprises. If you’re manufacturing commercial airplanes and an engineer is seeking expert answers to get a grounded aircraft back up and running, there is zero room for hallucinations. 

2. HOW IMPORTANT IS SECURITY TO MY ENTERPRISE?  

In the world of enterprise GenAI, you’re receiving answers based on multiple pieces of content from several secure repositories and data records. Enterprises cannot use a piece of content that compromises existing permissions.

In order to protect valuable assets and keep data secure, any enterprise-level GenAI needs to understand which employees, customers, or stakeholders can securely access which specific pieces of content before the answer generation.

3. HOW DO I KEEP MY PROPRIETARY CONTENT OUT OF THE PUBLIC REALM?  

Your employees should never be uploading your IP or customer data into a public GenAI engine. In fact, many large enterprises—including Apple, Verizon, Amazon, and JPMorgan Chase—have banned or limited the use of ChatGPT at work to prevent the release of confidential information.

Your GenAI solution should keep your proprietary data, well… proprietary and secure while unlocking the full power of GenAI technology.

4. DO I NEED MY CONTENT TO BE CUSTOMIZED AND UP TO DATE?

ChatGPT has a knowledge cut-off of September 2021, which means it doesn’t know about the spy balloons spotted over the U.S. in 2023 or current tax laws. ChatGPT also doesn’t know the specific lexicon of a particular enterprise, which can be problematic for both users and brand reputation.

GenAI that taps into real-time data sources will serve up answers that are consistent with the content in the enterprise and spoken in a language that is contextually relevant to the users it’s targeting.  

5. DO I NEED TO VERIFY THE TRUTH OF AI-GENERATED ANSWERS?  

ChatGPT will apologize if you call it out on giving you an incorrect answer, but your enterprise doesn’t need good manners—it needs verifiable information and, in many cases, compliance and traceability.

Maintaining the linkage to the source of truth is key so you can quote sources and prove with citations how you came up with an answer. GenAI needs to establish seamless links to sources along with the generated answers, providing peace of mind around validating the credibility of any information that surfaced. 

6. HOW MANY CONTENT SOURCES DO I NEED TO DRAW FROM?  

When using GenAI in the enterprise, you aren’t generating answers from a content silo or just surfacing links, paragraphs, or documents. Enterprises have a large number of disparate yet relevant content sources, from public websites to proprietary databases and more. The ability to securely integrate these content sources is what drives exponential value.

7. AM I SUPPORTING STAKEHOLDERS THROUGH SEARCH, CHAT, OR BOTH?

GenAI offers support through chat channels, but that doesn’t mean search is going away. In fact, the search box just got bigger, but it doesn’t mean that every user will type in a long-form question versus a simple query.

Enterprises need search and chat to be in step with one another. If you power separate channels with different relevance engines, your users will get a different set of answers from each of them. A unified index feeding both search and GenAI channels can help avoid the information silos and inconsistencies that can confuse customers.

8. CAN MY ENTERPRISE AFFORD TO DEVELOP PROPRIETARY GENERATIVE AI?

The cost of a generative chat can add up to 100 or even 1,000 times the cost of a query-based search experience. OpenAI has spent billions of dollars training ChatGPT, and when it comes time to retrain the model, it will cost billions more. A large language model (LLM) service can be expensive. Companies like Amazon and Wayfair can afford to have thousands of data scientists working on their own proprietary GenAI models, but the majority of enterprises need to seek out existing solutions that won’t introduce unnecessary costs or risks.  

By leveraging advanced semantic search and relevance AI and combining them with LLMs using advanced prompt engineering, enterprises can get GenAI results for a fraction of the cost.

HIT THE GROUND RUNNING BY WORKING WITH A TRUSTED GENAI PARTNER  

The new expectation is for enterprises to deliver modern, intelligent, and conversational experiences at scale, without compromising security and factuality. The businesses that don’t adopt GenAI will be forced to compete against it.

Seeking out a partner with extensive AI expertise and an in-depth understanding of LLMs is a smart way to proceed. By choosing technology that overcomes the core issues of hallucinations, privacy, security, veracity, siloes, timeliness, and cost, you can tap into all the business-boosting potential of GenAI in a brand-safe way. 


Louis Têtu is Chairman and CEO of Coveo (TSX:CVO), a global leader in applied AI.