ChatGPT is the reason that artificial intelligence and its potential are known to the general public. Although many have seen the light and no longer deny the value of artificial intelligence (read: ChatGPT), few really act on it. That is good news! There are still a lot of opportunities that your competitors are not seizing. And no, our golden advice is not to let your colleagues browse to chat.openai.com to talk to ChatGPT there. But what, how to start with AI and where?

Is a tailor-made AI solution relevant for your company?

We like to be completely transparent. At integrators.ai we do not start our free (yes, really free!) research process at companies smaller than ~10 FTE (with the exception of early stage start-ups). The fixed processes, amounts of data and infrastructure required for efficient AI implementation in companies smaller than 10 team members are often still too changeable or do not (yet) exist at all. Experience shows that a tailor-made (basic) AI solution starts with an investment of approximately €10,000. This type of investment is logically not included as standard in the annual financial budget of every company.

How to use AI if your company is larger than 10 employees?

Now it gets interesting. We see in practice that for every company with more than 10 employees, a custom AI solution can be devised that will pay for itself. But where to start? We often start by asking the following questions:

• Which business processes are currently slowing down the service or product you provide? For example, do you have enough leads and good conversion figures, but does it take forever before a lead actually signs the contract? Or does it sometimes take a day to prepare a quotation, causing your sales department to lose valuable time when they could have had more introductory conversations? Or is your product or service complicated and do you need many support staff to accommodate this, which means you can invest less in growth? In all these cases there is something that can be improved and accelerated using AI. Truly!

• Check with your colleagues whether they individually encounter repetitive tasks on a daily basis. These repetitive manual processes can often be automated with AI. This way you not only ensure higher productivity, but also a happier colleague who no longer has to perform that repetitive, annoying task! Another tip is to peek (if you have the rights to do so) in the calendar of colleagues, do you see the same appointments / tasks recurring there frequently? Then it might be interesting to ask about that!

• What data and data flows are there within the company? Am I collecting valuable data? Who has access to this? Does the data I collect provide real-time insight to make better decisions? What data am I still missing to make better decisions? For example, are you a store with multiple branches and have you had an online accounting program for 10 years? That data is extremely valuable! Through smart automation and analysis, you can use AI to make predictions about your turnover, for example based on the weather, the AEX price or the popularity of a hashtag on x.com. The data on which you want to send does not have to be numerical! Do you use ticketing software to assist your customers with support questions? Top! The questions and answers contained therein are now very useful for creating automations and analyses.

These are logically not the only questions you can ask yourself to determine where value can be gained from AI in your company. At integrators.ai, we have already spoken to more than 30 different companies about possible AI applications, which means we are getting better at asking the right questions.

Of course, it helps when spotting opportunities if you know what is possible! So why do this alone? Let us help you completely free of charge, without obligation, with an informal conversation about the possibilities of AI for your company! Easily book a time slot at https://calendly.com/integrators-ai/hi.

These are logically not the only questions you can ask yourself to determine where value can be gained from AI in your company. At integrators.ai, we have already spoken to more than 30 different companies about possible AI applications, which means we are getting better at asking the right questions.

What is a custom AI solution?

Okay, so we conclude that there is potentially an interesting use case for AI within your company. What now? At integrators.ai we build tailor-made AI solutions, completely tailored to the exact needs of your company. This means that we will not advise you to follow training in using ChatGPT with your team. There are already plenty of other providers! So what do we do?

Basic AI implementation via APIs of existing AI models

AI = booming. The rise of large AI models that perform fantastically is enormous. This means that in many cases we do not have to develop our own AI model at all! We simply use those from another party. For example, OpenAI, Google and Amazon all have fantastic models available for all kinds of different applications that we can call with a minimal amount of code and thus integrate into your infrastructure.

When to choose this solution?

We opt for an API implementation when your company has no or insufficient data of its own to develop a model with, or when the specific AI model we need has already been developed and performs well on the task we want to automate. For example, it is currently no longer necessary to always develop your own AI language model (LLM), as ChatGPT and Bard, among others, are much better at this than we will ever manage (unless you still have 50 billion to invest somewhere 😉 ).

Hybrid AI solution with its own data, but third-party AI model

A step further than a basic API implementation is fine-tuning an existing AI model with your own company data. With this solution we fine-tune an existing model until it performs almost perfectly for your specific business case. In this way you make use of good basic performance of a model, but at the same time you ensure that the model becomes stronger in dealing with your own data.

When to choose this solution?

This solution is only possible if a company has relevant data of its own (does not have to be a huge amount, but think of at least 500 high-quality data points). Again, the data does not necessarily have to be numerical data. For example, it could also be a very large database with its own contracts or manuals. In the example of contracts, we can implement an AI solution that is trained specifically on your database with all contracts (and associated conditions) concluded with suppliers.

With such a solution you can let your (purchasing) team ask very specific questions about the contracts via chat functionality. This not only means asking for specific details, but also abstract questions such as: which supplier contract should we renegotiate first, taking into account that our goal is to be maximally flexible in deliveries within area X?

Custom AI development, a completely own AI model

The most complex and extensive solution is to create an AI model completely specific to your situation and data. This means that we write, train, test, evaluate and optimize our own algorithms until we reach a jointly determined threshold for performance. With this solution we create one or more niche algorithms that become very strong in a specific prediction or classification.

For example, consider developing an algorithm that predicts the risk of a certain lifestyle disease based on patients' blood biomarkers. If you were to ask ChatGPT this question, you would quickly discover that the model was not trained on such deep, niche predictions.

When to choose this solution?

This custom route is only interesting to consider if there is sufficient (own) relevant data to train a model on. Without at least thirty thousand relevant data points, this solution is not worth considering. Of course, it is only interesting to develop your own model if there is not already a model available that performs well for the specific use case. It is important to remain pragmatic and look carefully at returns vs. costs involved in AI development.

What does a free exploration of my company's AI potential look like?

Okay, your organization is larger than 10 FTE and you are somewhat convinced that AI can add value, what now? At integrators.ai we notice that a lot of companies are currently struggling with this question: I feel like I need to do something with AI, but I have no idea what and where to start. To meet this need, we at integrators.ai offer a unique, free and non-binding AI research process. What does this process look like?

Broadly speaking, the process is divided into three phases, all free of charge and completely no-strings-attached:

Phase 1: Informal introduction via video call

We founded integrators.ai to get as many companies as possible excited about the possibilities of AI and wanting to create extreme value with AI innovation & implementation. This not only means that we first calmly explain to you what artificial intelligence is and what it can potentially do, but also that we investigate whether there is a personal connection. Whichever way you look at it, the success of our research process is entirely dependent on the people working on it. This requires an investment of time from both parties. We already know that we want to make that investment, and we would like to discuss with you whether this is also possible for you. We understand that your company never stands still and that other matters sometimes have more priority. That is why it is important to be immediately transparent about expectations.

Phase 2: Further exploration on location

The click and willingness in terms of time investment are there: yeah! At that moment we will come to you (don't worry, not immediately, we will set a date together 😉). This means that we travel anywhere throughout Europe to discuss your company face-to-face. With the information from the first introduction, we prepare ourselves as best as possible to immediately provide valuable insight into artificial intelligence for your industry and specific challenges. We will ask you a number of questions to better understand your business (model) and to encourage you to see the potential of artificial intelligence even more clearly.

We reserve this session 2 hours in advance as standard (longer is always nice). It is a requirement that we sit at the table with at least one decision-maker. This is not because other team members do not have relevant information (on the contrary), but because we want minimal noise in the information we share. This not only saves on translation in the event that a colleague informs the decision maker, but also creates clear expectations. We cannot afford the risk of noise with the time investment we make.

Phase 3: In-depth research of potential AI case including ROI estimation

If all goes well (so far it always has), a picture will emerge from phase 2 of one or more interesting challenges that AI can help with. Together we determine which opportunity we want to explore first. For such an investigation, we need a permanent point of contact who not only knows the company inside and out and has access to all relevant systems, but can also act quickly during this phase. In general, we try to gain in-depth insight through, among other things, the following (non-exhaustive) questionnaire:

• What is the infrastructure underlying the process?

• What data is created during the process?

• Which data is stored automatically and which manually?

• What data is currently missing that could be relevant to the process

• Which data flows go in and out (input/output)?

• Who is involved in the process?

• Who processes or creates data in the process?

Once we have a clear picture of all steps within the process, the relevant data (flows), people involved and the degree of automation, we make a global assessment of the possible AI solution(s) including ROI calculation. An AI implementation is divided into two phases as standard:

1. Prove effectiveness of solution with a minimum viable product (MVP)

2. Company-wide implementation and further development

Ultimately, phase 3 results in a collaboration proposal in which we create the MVP for a fixed amount and after delivery we have built in an evaluation moment to decide whether you want to invest with your company in company-wide AI implementation and further development.

Why do you offer AI implementation in this way?

Practice shows that it is much more efficient to create a working basic version of the intended end product before we realize the actual implementation. In this way we eliminate certain teething problems at an early stage without them being in the production version. In addition, this method offers a lot of flexibility to iterate towards the optimal result and we are therefore not “stuck” to a previously devised solution.

For example, re-imagine the use case of a hybrid AI solution for supplier contracts. Of course, we have an idea in advance of what the final solution will look like. Yet it makes much more sense to test these assumptions at the earliest possible stage and first create a stripped-down version that works on one type of supplier contract, for example. Then, at this stage, we may immediately discover that the existing infrastructure is not sufficient or, by having employees work directly with the demo product, we may discover that the value does not lie in asking abstract questions about supplier contracts, but much more in specific ask for details.

Once the MVP has been deemed successful, we move on to phase 2: full implementation and further development. This means that we develop a production-ready solution that not only works well, but also integrates perfectly into the existing process. The advantage of this 2-stage rocket is that you can still do without phase 2. Business is complex and changeable, and it may be the case that what you thought would be valuable for your company, is not so valuable in practice. Fortunately, you have only made the investment for phase 1, instead of the entire implementation immediately!

Why and how to start an AI journey with integrators.ai?

Hopefully we have been able to convince you of the value of our research process. If this is the case, start today by booking a 30-minute online introduction via https://calendly.com/integrators-ai/hi. Once again: completely without obligation, free of charge and no-strings-attached! Need even more reasons to talk to us? Here they come:

• Did you know that we provide 100% ownership of our projects? In other words, the AI ​​solution we build is really yours. No fine print or copyright hassle. You will have exclusive ownership of all infrastructure and code, including a confidentiality agreement.

• At integrators.ai we do not do off-shoring. In other words, the price you pay is actually in line with the market for the technical talent working on your solution. No outsourcing to teams in low-wage countries. At integrators.ai only the highest quality and full control over the code produced in-house in the Netherlands.

• We like to see ourselves as a new generation of tech developers. This means that we only work with the latest frameworks and modern solutions. In this way, we are faster and more inventive than many established “tech” companies (read: often consultancy companies with external development partners). We have all the skills in-house to actually integrate your AI implementation into existing code and, if necessary, design and build the sleekly designed interface that makes the AI ​​model user-friendly and practically valuable.

We hope that the above is sufficient reason to book that informal first meeting of 30 minutes with us. Would you rather read again in a highly condensed format why you should book that introduction? Do you have another AI-related question or would you like to take a look at one of our slide decks about AI? Then check: https://integrators.ai/nl/start-now. Hopefully we will see or speak to you in some form soon!

What to do if your company is smaller than 10 employees and you want to use AI?

For companies smaller than 10 team members, we recommend using one or more of the thousands of AI tools that already exist. There are tons of different tools available that use artificial intelligence to do specific work smarter and faster. We won't bore you with another list of “the best AI tools for recruitment/marketing/ICT/etc.”, just Google this and you will find a sea of ​​​​AI possibilities. Still have a question? Feel free to ask us via [email protected] and we may be able to help you out!