Generative AI and Your Tech Company

Sep 20, 2023


Back in 2019, Akshay Shrimanker, our Founder and CEO, joined a panel about artificial intelligence (AI). Just three years ago, the panelists were excited about the promise of AI but said it had a long way to go. Specifically, they identified the need for more inputs alongside more willingness among people to buy into AI.

Fast forward three years and all of that has happened. AI is here in full force, and companies are already far down the road of exploring its potential uses. In fact, 35% already report using AI in their businesses.  

Clearly, there’s an open opportunity here. And that’s particularly true when it comes to generative AI. With that in mind, let’s look at this tool and what it can do for your tech company.


The possibilities of generative AI 

Generative AI hinges on models — like large language models (LLMs) and foundation models — that use neural networks. Through these networks, the AI models can identify patterns and trends. Then, with that information, they can extrapolate to generative just about anything. In other words, generative AI takes the data given to it and uses it to create something new. 

That something could be anything from text to an image. Generative AI can even make audio, video, and 3D models. 

And, as ChatGPT demonstrates, generative AI can make analyzing data and synthesizing complex topics simpler. It can help your tech company by analyzing both market trends and your historical data to help you predict demand, forecast your finances, create marketing content, and more. 

Long story short, there’s a lot of opportunity here. But there’s also a lot of risk. Tech companies should explore using generative AI, but they should do so with these best practices in mind:


#1: See AI as a co-pilot

Generative AI should support your team, not replace them. In fact, keeping a human in the loop when creating with generative AI can help you avoid a lot of the issues that can crop up, from bias to outright inaccuracy. View generative AI as an assistant to work alongside employees — one that has the ability to know everything about your business via the data you feed it. 

As you invest in developing and training generative AI tools, invest just as much in training your team how to use them. Everyone should get a base level training in how to use AI. You might even invest in your team by bringing on new headcount with competencies like AI engineering or enterprise architectures. 

You should see returns on this investment. As your team learns how to use AI, they will find ways to integrate it into their workflows. This drives productivity and can help your tech company grow without adding employee hours. 


#2: Tailor the tools you use

While there are a lot of out-of-the-box generative AI tools out there — like the aforementioned ChatGPT or Google’s Bard — don’t stop there. You can use these options to get started with AI, learn its capabilities, and see quick returns. 

Ideally, though, you want to tailor the generative AI you use. If you have the bandwidth, developing your own internal generative AI protects you from the security concerns that come with using a publicly available tool. Alternatively, your AI engineers might start with an API and tailor the model to your company (just be mindful of the data privacy ramifications). 

In either case, optimizing generative AI models for your company’s specific use cases helps you unlock the most potential here.  


#3: Clean up your data 

As we mentioned up top, generative AI generates by extrapolating from the data it was given. For the AI model to function its best, you want that data to be clean and curated. If you feed it outdated information, for example, the AI won’t know to set that to the side. That data will influence what it generates just as much as anything else you use to train it.

If you plan to use generative AI at your company — and you should — data cleaning matters now more than ever. This can be a tedious task, but it can help you avoid inaccuracies as you deploy AI tools. And we’re likely only seeing the beginning of these models’ capabilities. In the next few years, a cohesive pool of clean data will be paramount for every business that wants to tap into AI. 

Once you get your data to a state you trust, build data policies to ensure that pool of data is safeguarded. Using an enterprise data platform with checks built in can help. You likely also want to distribute a data policy for your team. 


#4: Address security concerns

That IBM report we linked earlier showed some pretty worrisome statistics. In companies already using AI, 74% had made no effort to reduce bias and 68% weren’t tracking performance variations and model drift. In other words, early AI adopters are often letting their AI models run wild. 

This opens you up to serious risk. If your team starts relying on AI and you don’t have measures in place to ensure the accuracy of what it generates, you could be in trouble. Say you’re using the generative AI model for trend identification and forecasting. Your business could be led astray. Or if you’re using it to generate marketing content, you could ruin your brand reputation. 

Remember, generative AI is smart, but it has its limitations. It struggles with context. It can even hallucinate, fabricating information and reporting it as if it was fact. This brings us back to the first best practice: see AI as a co-pilot. Make sure you have a human involved in any AI processes at your business so that they can review the output for inaccuracy, bias, and other red flags. 

You should also be extremely cognizant of data privacy if you’re using publicly available AI tools. Anything you put into a prompt could be seen by outside eyes. To protect proprietary and otherwise private information, your company may benefit from creating a guiding policy around what can go into AI prompts and what can’t. 

Ultimately, as you scale up your AI use, scale up your company governance with it. Make sure you’re building in the right review processes, policies, and controls to protect your company against potential risks.

To free yourself and your team up to focus not just on developing and deploying generative AI, but also on using it safely, let us handle your accounting needs. To work with accountants who specialize in serving tech companies, get in touch today.