Applied AI & AI in business /

AI Unchained: Delivering Your Model Through Integrations

June 6/4 min read
  • Korey Stegared-Pace
    Korey Stegared-PaceDeveloper Advocate

The path towards a successful AI project can involve many steps but the journey should never stop after building a model. The question of how you will deploy the model and get it in the hands of your users and customers should be a priority to be answered.

At Peltarion, we truly believe that AI can solve real world problems. In order for that to be more than just a company slogan, we have made it easy to integrate your models from the Peltarion Platform to popular app and workflow building tools like Zapier. Here are some benefits to using these integrations for deployment of your models:

02/ Meeting Problems Where They Are

Most likely when you began your AI adventure, you had a vision of fixing some sort of problem or enhancing the experience of your users - either yourself or someone else. The ultimate goal for developing your AI solution could be to save time, money or even lives. Even more likely, these users are will not be working with your model directly but with the results of it. This could be by providing sentiment analysis on customer reviews so your sales team have better direction or classifying car damage images to ease the load of your customer service team.

The best way to solve your user’s problems is to meet them where they work. For example, you have created a model to forecast sales that you have created for people working in finance. What better way to deliver these predictions than directly into every finance person’s favorite tools of choice like Excel or Google Sheets? Incorporating predictions into their workflow ensure smooth and effective delivery of your AI solution.

03/ Producing Real Results

You have invested a lot of energy and resources in collecting your data, then time into building and fine-tuning your model and you are finally happy with the results. Now that you are at the “finish line”, you realise that you have less resources than expected to commit to building a full application that integrates the model. Instead of labelling this project a "failure to launch", you can be efficient in your resource planning at the application building stage. A quick way you can share your model inside of an application is by using a no-code app builders like Bubble and Microsoft PowerApps. By developing applications within these platforms, you can have a quicker time-to-market and make the results of your hard work a reality. There will always be time to create a a flashier or more sophisticated application once you have proved the value of the AI solution.

04/ Avoiding the POC Graveyard

Many organisations are interested in how AI can solve their business problems. Even with a high interest and some times an even higher investment, AI projects have a bad habit of never leaving the ground. But why is that? While performance may vary on the deployed model, not approaching releasing your AI model using product development best practices will hamper its success. By integrating the model into the existing workflows of your users or building no-code solutions, you can get fast and useful feedback on how well it is performing. Once you have this feedback, you can learn how to improve the model through techniques like collecting more data. Having real results and improvement over time is great evidence to anyone outside of the project that AI can deliver tangible improvements in their work.

  • Korey Stegared-Pace

    Korey Stegared-Pace

    Developer Advocate

    Korey Stegared-Pace is a Developer Advocate at Peltarion. After working several years within the E-commerce and Payments industry, he decided to pursue his passion of bringing AI to all. His favorite AI topic is anything to do with Generative Networks and how AI can not only learn but create. When not working with the community of users at Peltarion, he is also the organizer of the Stockholm Machine Learning Club.

02/ More on Business & Applied AI