# Model in minutes. Trust, but verify. Source: https://aletyx.ai/ai-assistant/ AI Assistant for your business rules A general AI chat can write rules. AI Assistant also validates them, tests them, and shows you exactly what it changed. You approve every edit. Have a spreadsheet, a policy document, or an existing model? Start there. Try AI Assistant free for 14 days **$95** per user / month Free for 14 days, in your browser. Card required, nothing charged until the trial ends. [Explore what you can do](https://aletyx.ai/ai-assistant/#existing-models) ### Ask for the coverage you need. Ask the assistant to cover the policy’s branches and boundary conditions. ### AI works with your model. The assistant prepares the cases and checks the expected outcomes. ### One request. Seven test cases. Both customer types, small and large orders, and the 10-item boundary. Expected results are saved for later checks. Click a row to inspect a case. ### Change the policy. Keep the tests. The Business small-order discount changes from 15% to 20%. The saved expectations stay at 15%. ### The tests catch the changed behavior. Two cases fail: the model returns 20%, but the saved expectation is 15%. The other five still pass. Click a red result to inspect the mismatch. Why AI Assistant ## Why not a general AI chat? A general assistant writes plausible rules and leaves the checking to you. AI Assistant is built for decision modeling, so you can trust, but verify. > Aletyx is the main team behind Drools, building the most advanced AI decisioning for agents I've seen. **Paul Vincent** Independent analyst, ex-Gartner - **Validates the model** Finds errors, gaps, and overlaps before anything runs. - **Tests it** Generates cases for your rules and boundaries, runs them, and saves the expected results. - **Shows what changed** Compare every proposed edit with the current model, then accept or revert it. - **Keeps the source** Each rule stays linked to the policy or spreadsheet it came from. - **Knows the craft** Decision-modeling best practices are built in, so your models stay clean as they grow. Understand ## Ask how it works. Get a straight answer. Ask why your model returns a result. AI Assistant explains the rules in plain language, and nothing in your model changes. ### Start with a question. Ask why the model produces a result. You can explore its rules without editing them. ### AI works with your model. The assistant reads the model’s rules to explain its outcomes. ### See how your model reaches a result. Read the explanation, then explore the decision table and data types. Your model stays unchanged. Test ## Build test coverage. Spot mistakes early. Ask AI to generate cases for your rules and boundary conditions. Review the expected outcomes, run the suite, and see when a change affects the results. ### Ask for the coverage you need. Ask the assistant to cover the policy’s branches and boundary conditions. ### AI works with your model. The assistant prepares the cases and checks the expected outcomes. ### One request. Seven test cases. Both customer types, small and large orders, and the 10-item boundary. Expected results are saved for later checks. Click a row to inspect a case. ### Change the policy. Keep the tests. The Business small-order discount changes from 15% to 20%. The saved expectations stay at 15%. ### The tests catch the changed behavior. Two cases fail: the model returns 20%, but the saved expectation is 15%. The other five still pass. Click a red result to inspect the mismatch. Create ## Turn what you have into a model. Start with the policy or spreadsheet you already use. Get a decision model you can inspect and test, without entering every rule by hand. From a spreadsheet ### Keep the logic. Skip the transcription. Attach your spreadsheet and ask AI to build the model. Review the generated rules without copying each row by hand. ### Start from the spreadsheet you already use. Attach your source and ask AI to create or update the model. Open the attachment to inspect it. ### AI works with your model. The assistant turns the spreadsheet’s criteria into executable decision logic. ### Your spreadsheet becomes an executable model. Eight source rows become a decision table. Review the generated rules. From a document ### Start from the policy. Review the model. Attach your policy and ask AI to turn it into a decision model. Inspect the rules and compare them with the source. ### Start from your policy document. Attach your source and ask AI to create or update the model. Open the attachment to inspect it. ### AI works with your model. The assistant turns the document’s discount rules into a model with data types. ### Your policy becomes an executable model. Review the three discount rules, their input types, and the references back to your policy. Change ## Make the change. Keep control. Changing one rule can affect other parts of your model. A missed edit can leave you with the wrong result. Ask AI to prepare the related edits together. Compare before and after, then choose what to apply. ### Ask for one change that touches two tables. One request covers two decision tables, the shared data type, and the diagram. ### AI works with your model. The assistant prepares the related table, data type, and diagram edits as one proposal. ### Review the Business discount changes. One proposal covers both order-size branches: 10% → 15% and 15% → 20%. ### Review the Private discount changes. The same proposal includes 5% → 10% for Private customers. No second request. ### Check the shared data type. The proposed values are 0.10, 0.15, and 0.20. This type is shared across the connected decisions. ### Review the diagram change. The proposed name is Approved Discount Rate. The connected decisions and their references stay together. Change from a document ## Your policy changed. See what needs updating. Attach the revised policy to your existing model. Review the proposed changes against the document before applying them. ### Start from the revised policy. Attach your source and ask AI to create or update the model. Open the attachment to inspect it. ### AI works with your model. The assistant checks the request against the model and prepares a reviewable proposal. ### Compare before and after. See the current values beside the proposed edits. You choose which changes to apply. Get started ## Start in your browser. Nothing to install. Use AI Assistant in Aletyx Playground or inside Aletyx Platform. The Playground workspace is free; your subscription adds the assistant. AI Assistant ### Start with the model you already have. Ask questions and generate tests first. Create or change models as you build confidence. **$95** per user / month Try AI Assistant free for 14 days Cancel anytime. You won’t be charged during your trial. #### What you can do: - Ask questions about your model’s logic - Generate and run multiple test cases - Create models from policies, spreadsheets, or prompts - Review proposed edits across an existing model - Use AI Assistant in Aletyx Playground - Download your DMN models and run them on compatible runtimes Credit card required. Usage limits apply. Prices shown don’t include applicable tax. Prices and plans are subject to change at Aletyx’s discretion. Keep what you build. Download your models and choose where they run. Frequently asked questions ## What to know before you try AI Assistant. 01 Do I need Aletyx Platform to use AI Assistant? No. You can use AI Assistant directly in Aletyx Playground, the free hosted modeling workspace. The Playground costs nothing; AI Assistant itself needs a subscription. Aletyx Platform users can also use the assistant in the model editor, and developers can use it through the Aletyx Extension for [VS Code](https://marketplace.visualstudio.com/items?itemName=Aletyx.automation-design), [IntelliJ](https://plugins.jetbrains.com/plugin/33190-aletyx-automation-design), and [Cursor](https://open-vsx.org/extension/aletyx/automation-design). 02 Can I run the resulting models outside Aletyx? Yes. AI Assistant works with the DMN standard. You can download your models and run them in any compatible DMN runtime. 03 How does AI-assisted validation work? Ask AI Assistant to generate test cases for the model’s rules and boundary conditions. Review and save the expected results, then run the cases against the model. Saved tests help you spot changed behavior when the model is updated; generated tests still need your review. 04 How is my content handled? It is used to produce your answer, nothing else. We don’t train on it, and our model providers can’t either. Your models and the output are yours. Keep personal data out of it, and keep your own copies of what you build. 05 How does additional usage work? Your subscription includes AI usage, subject to limits. If you need more, you can purchase additional usage. Current rates are shown before purchase. 06 What happens after the free trial? A credit card is required to begin the 14-day trial, but you will not be charged during it. Cancel anytime before the trial ends. Otherwise, the subscription continues at $95 per user per month, plus applicable tax. 07 Can I use AI Assistant in my IDE? Yes. Developers can access AI Assistant using the Aletyx Extension from their preferred IDE: [VS Code](https://marketplace.visualstudio.com/items?itemName=Aletyx.automation-design), [IntelliJ](https://plugins.jetbrains.com/plugin/33190-aletyx-automation-design), and [Cursor](https://open-vsx.org/extension/aletyx/automation-design). It is also available in Aletyx Playground and Aletyx Platform. ## Start with a question. Understand your rules, test the results, and review changes before applying them. Bring a model to Aletyx Playground and start with a question. Try AI Assistant free for 14 days $95 per user/month after the trial. Cancel anytime.