Ok, so you've learned how to use ChatGPT. What's next?

Red and black circuit board

AI (or machine learning for pedants) has become ubiquitous. In just a few months, AI has gone from being the domain of large technology companies with astronomical budgets to being available to the smallest bicycle repairman on top of the narrowest mountain in Lom municipality, as long as he has internet access and is not busy racing cars (That was a reference to the Norwegian movie The Pinchcliffe Grand Prix, which you should watch if you haven't already).

In addition to powering ChatGPT, AI can now, for example, retrieve information from proprietary data systems such as accounting software and intranets, create illustrations for landing pages, or make information accessible that would otherwise be buried in public archives.

Some of the examples in this article already exist, while others could quickly become a reality with a little help from us. In other words, they are not far-fetched dreams of something you might be able to accomplish in the future, but concrete examples of something that is fully possible to implement today.

Training AI on your own data

If you have an accounting system, you have probably sifted through piles of reports and graphs to find answers to something the board is wondering about. But what if you could just ask questions like "What were our top five budget exceedances last year?" and get answers in seconds? An AI algorithm can be trained to understand the structure and data of your own systems, so you never have to spend valuable hours digging up information again.

Or what if a new employee could ask your intranet how to fill out the project proposal template instead of asking you? The time savings would be enormous in many cases. AI becomes like a personal assistant for the new employee who is able to search through vast amounts of data quickly, so they don't have to click through page after page of company directives to find what they're looking for.

Training AI on traffic data

AI can be trained on analytics data (in Google Analytics, Matomo, or other alternatives) to provide insights into website performance. Wondering which blog posts work best among 25-30-year-olds in Cambridge and what characterizes them? Without knowing anything about analytics tools, you can quickly get answers to such questions. The answers you get can then be used to create targeted content that resonates with specific target groups, increase engagement, and increase sales.

Training AI on laws and regulations

Communicating information about decisions that affect specific demographics is a big challenge for public agencies. After all, there is a limited amount of space on the front page. On the consumer side, it is often difficult to find the information they are looking for, even when they know exactly what they are looking for. AI enables businesses and individuals to ask specific questions, such as "What are the requirements for building a new garage in Birmingham?" An AI tool can quickly provide answers, saving time and reducing errors.

Example: Microsoft 365 Copilot

The suggestions above require some development before they can exist, because the algorithms need to be trained on the organization's own data. We can program them for your business, but if none of the examples appeal to you and you are on the Microsoft platform, Microsoft 365 Copilot is a good choice. It is an AI-driven tool that already helps users perform common tasks such as scheduling meetings, formatting documents, and even writing emails. The tool learns from the user's behavior and provides personalized suggestions to improve productivity. This saves time and allows users to focus on more complex tasks.

Example: Stable Diffusion

Many have heard of Midjourney and DALL-E, which can generate new illustrations and images based on existing art. However, these are closed solutions where it is difficult to generate consistent illustrations for a brand over time.

As we are fond of open source solutions, we instead implement Stable Diffusion for our clients. This is the open-source version of image generators, which can be downloaded onto your own servers, tweaked, and trained. This is a great solution when you need images for blogs, landing pages, or other marketing materials that require a certain look that is difficult to achieve with Midjourney and DALL-E.

Future example: AutoGPT

The most hyped version of tools that can put ChatGPT into "anything" is called AutoGPT. It requires some technical expertise to set up, but from there, you have virtually endless possibilities. The developers write that it is designed to develop business ideas and generate revenue completely on its own. Testers report that it quickly programs infinite loops it cannot get out of, but when it becomes more refined, it will be able to function as an autonomous assistant that can accomplish almost anything.

It is conceivable that you could eventually give it a goal, such as increasing social media followers by 10,000 within a year. It will then create "prompts" the sentences we currently manually enter into ChatGPT- to try to reach that goal. Provided it does not go and buy followers or start generating content unrelated to your business just to generate clicks, many marketing departments will have a powerful tool in the future.

Bonus: Personal Coaching with ChatGPT

Do an experiment. Log in to ChatGPT and type something like:

"You are a coach with 30 years of experience helping people reach personal goals. You are an active listener, and you motivate and challenge your clients."

The robot will likely ask you a question about what goal you want to achieve. If you feed it a personal goal, based on the skill development goals you have set in your company, it can, with a little tweaking, provide each employee with a weekly plan to achieve their own goals.

If you test it out a bit yourself, you can eventually set up standardized questions that employees can ask the robot. This can help employees develop the necessary skills and confidence to achieve their career goals. For example, an employee can request personal coaching to become a master at sales pitches, and ChatGPT will offer guidance and exercises tailored to help them achieve this goal.

If you need help setting up these or other AI solutions for your business, feel free to contact perandre@frontkom.com or submit a request via our contact form.



Per André is a Co-founder at Frontkom with 14+ years of experience as CTO and CIO. He has extensive experience with people and technology for both private and public sector. He is also co-founder of Dignio Health Tech, SMSpay, Web3 enthusiast and Co-pastor of 3:16. Per André writes about CMS, headless, awesome tech and team composition and efficiency.

Subscribe to our blogs