Frequently asked questions

  • Amatrium was born by a group of materials & manufacturing experts with +100 years of combined experience. Material design was too tedious as time is consumed by referencing books, patents, spreadsheets & process data…there had to be a better way.   Machine learning compliments tools like ICME, yet it is much easier to use and accessible by many people across the organization. 

    When GenAI was released by OpenAI, our team recognized that it is a special capability, yet organizations would only use it if there were security.  So, we built security around ChatGPT, named it AmatriumGPT, and we have been adding more value since.   

  • AmatriumGPT uses ChatGPT-4o, yet there are three distinct differences:  

    1. ChatGPT cannot learn from our customer’s data.

    2. Our customers upload files into their respective server space, and it remains there for employee access as needed.  One great opportunity here is to make the engineering dataset available for customer service, new employee training, and technical marketing!  

    3. There are two additional tools beyond the CHAT tool for Search & Discovery.

      - SUMMARY allows uploading 1 to 5 documents to create press releases, summaries, or other content creation.

      - TRANSLATOR allows word-for-word language translation.  There are 100 languages to select, and up to 5 files can be translated simultaneously.

  • AmatriumGPT is a language model, so the typical file formats that are uploaded include PDF, Microsoft Word, Microsoft Excel, .csv, and .txt files.  Amatrium is investigating other file forms, so please Reach out to us if you need another file format. 

  • There are multiple applications of Amatrium, from solving an intermittent issue on the plant floor to new material composition or a process recipe. In Media, see the white paper will show how Amatrium successfully predicted material properties. Also, recent studies of QC data brought production solutions that the metallurgists and operations personnel did not determine on their own using tools like Excel or Minitab.  Reach out, and let’s explore the possibilities.

  • There are multiple applications of Amatrium, from solving an intermittent issue on the plant floor to new material and process design. A coming white paper will show how Amatrium successfully predicted material properties. Also, recent studies of QC data brought production solutions that the metallurgists and operations personnel did not determine on their own using tools like Excel or Minitab. Let’s explore the possibilities. 

  • Amatrium Process takes in the data that you have, typically in a spreadsheet, and includes chemistry ranges, process metrics, properties, and pass/fail metrics. The more data the better. For problem-solving in production, the quality control (QC) data provides a good start. 

  • If it is a focused dataset, fewer lines are needed.  We have had success with 70 lines of data across five variables.  If there are 30 or 120 variables, then the minimum dataset should have hundreds or a thousand lines of data.  Ideally, we have a conversation and work it out together. Reach out, and let’s discuss. 

  • Yes. Customization is important to our customers, and with our agile and capable team, we develop custom software to meet the need.  To date, successes in machine learning come from a variety of use cases that include a machine shop to predict machining time; a mill to predict the causes of scrap; an aluminum foundry to predict mechanical properties, and a material development product to reduce the number of iterations to reach target performance. 

  • We take data security very seriously. We implement the best practices of data security according to three dimensions: Confidentiality, Integrity, and Availability. Some examples are: quarantining sensitive files, behavior-based permissions, preparing for cyber threats, deleting unused data, and implementing a good backup strategy. Reach out, and let’s discuss data security in more detail. 

  • There is good content on our Media page. Otherwise, Reach out and let’s get the conversation started. 

  • Reach out, and we will schedule a live demonstration via Teams.  You can share documents that you would want us to work with in that demonstration, or we will use our documents.  After the demonstration, we offer a free beta test for multiple people in your organization to try before you buy.

  • Reach out to us via the website, or direct to your sales contact.  We typically resolve any technical issue within hours, if not sooner.