In this bot you will learn to create your own small Machine Learning Model to predict selections of options in the contract, so your client or customer gets AI guidance using your NDA generation tool. You can use this function to show the most probable options based on historical data. Let me show you how this works in your frontend based on an NDA example.
In our example, we have set up the tool in a way that user needs to fill in a brief questionnaire about information about the parties which take part in the NDA, such as group entity, the industry, their name, address or HRB number.
Next, we are asked to briefly describe the project for preamble and if we want to provide a more detailed description of the project and add it as an annex.
Now it is time to select which options we want to include in the NDA and here is where the Predictive AI comes into action by predicting which options would fit best according to our situation. The options are predicted on historical data, and you can decide which inputs are going to predict which options.
Next, let’s select from when the NDA should apply and its expiry date.
And finally, here we have our completed NDA, which is now available for download.
Now that we’ve experienced the potential benefits of this tool, let’s master the art of crafting our own NDA with the help of your own trained AI.
– Let’s build this bot step by step! –
Step 1: Initializing the bot
We are starting by giving our bot a memorable name and a short description.
This first section purpose is to gather information about the different parties which take part in the NDA.
To do so we start with a “dropdown” node from which the user can select between “Lexemo GmbH” and “Mustermann AG”.
Step 2: Party Selection
Afterwards, we need to include a “logical condition” with two different “logic steps”.
The first “logic step” is related to the “Lexemo GmbH” selection, while the second “logic step” with “Musterman AG”.
Following each logic step let’s add a “boilerplate” node with the address of each company. So, for example, if the user picks “Lexemo GmbH” its address will be added to the NDA.
Step 3: Choose the industry
Also, we need the user to select the industry of the contracting party via “dropdown”. In this case we offer three options: IT/Developers, advisory and freelancer, but you can add options adapted to your precise use case.
Step 5: Enter party information
To end with this section, we introduced three different “text” nodes for the user to enter the following information:
- Contracting party’s name.
- Contracting party’s address.
- Their HRB number.
Step 6: Outline the project
Now, to maintain organization, let’s add a new section where we request a description of the project.
To do so, we need to include a “text” node.
Step 7: Annex option
Next, include a checkbox node. Here you can decide if you want a detailed description of the project and the information exchanged to be added as an annex.
Add a new “logical condition” and in case user’s selection is “yes” from the “checkbox-annex” they will be able to enter a detailed description of the project thanks to a “text area” node.
Also, add a “boilerplate” node in which we want to include the “annex-yes” variable to introduce that annex on the NDA. By adding the Annex variable, we are adding the text that user enters in the frontend.
Step 8: Key section
In this NDA example, we have included a new section which contains two differentiated parts, one for “reverse engineering” and the other for “contractual penalty”. Each of them, have an “AI Prediction” node that will predict the result for each of the questions in this section.
Step 9: Predictions setup
We start with the “reverse engineering” and the “AI Prediction” node we have configured for it. This node requires certain specifications, but remember that you can always modify this parameters according to the type of contract you would like to automate.
Click “Edit Input Nodes” and choose “dropdown- industry and dropdown party”. This is the basis for our prediction, so in other words, every input node we select here will affect the prediction of the selected output node. Now let’s click “Edit output Nodes” and select the checkbox node we actually want to predict. This is the one that follows immediately after, in this case “checkbox-reverse engineering”.
From the “Model Settings” dropdowns choose “random forest” and “f1-score”. If you want to know more about what these settings are, you can click on the relevant link and learn more about these models.
Now we want to train the model. As we do not have sufficient runs from the past, we use a CSV with historical data to train the model.
Double click the “.CSV” button to upload the training data. A modal window will appear, guiding you through the required format and uploading your .CSV file. You can watch the CSV requirements tutorial for further information about this kind of format. Now you can upload the .CSV file, which prepares your dataset for training the AI model.
Finally, hit “Evaluate Data” to confirm your dataset readiness and get some statistics of your own model.
If you have any questions about this node, you can always watch the AI Prediction tutorial.
Step 10: Clause implementation
Now include a new “checkbox” node. The special thing about this node is that the answer of the first question of this section “explicit exclusion of reverse engineering?” will be predicted.
Afterwards, introduce a new “logical condition”. If the user’s answer is “yes”, we are adding a boilerplate node with a clause about the “ownership of confidential information”.
Step 11: Contractual Penalty Implementation
Here, on the “contractual penalty” part of the section we are adding a new “AI Prediction” node, following the same steps from the first one. On the input nodes select the ones we want to base the prediction on. Then select the checkbox node that follows immediately after, in this case “checkbox-contractual-penalty”, which is the one we want to predict.
Step 12: Penalty Inclusion Check
It is the turn of a new “checkbox” node to be predicted. This one relates to whether a contractual penalty should be added.
Now add a new “logical condition”. If the answer to the previous question ends up being “yes” add another “AI Prediction” node (with the same settings from the previous but modifying the input and output nodes).
Right after this node, add “checkbox” node to define whether the structure of the contractual penalty should be “variabel” or “defined”.
Step 13: Specify Penalty Terms
Continue by adding a new “logical condition” and if the answer is “variable” include a “boilerplate” node with a “contractual penalty clause”.
However, if the answer is “defined” include an “AI Prediction” node. In this case, unlike in previous “model settings” the option are “ridge regression” and “R2 Score”.
Following that, add a “number” node to indicate the “contractual penalty amount”. This amount will be predicted thanks to the previous “AI prediction” node.
The last step of this section is to add a “boilerplate node” with a contractual penalty clause which should include the “number-contractual-penalty” variable.
Step 14: NDA Timeline
Next, we create a new section. This section’s purpose is to select the duration of our NDA. To do so, include two “date” nodes, one to select from when the NDA should apply and the other to select its expiry date.
Following that, include a “boilerplate master” in which you build the NDA by adding every variable needed. This boilerplate can be integrated into documents, text fields, etc. And serve as flexible variables for your Bot.
Afterwards, introduce both a DOCX and a PDF and edit them by adding the “boilerplate master NDA” variable. These nodes’ purpose is to allow users to create stunning DOCX and PDFs with ease.
Step 15: Finalize and Review your NDA
We have finally come to the last section. This one’s scope is to preview the NDA.
Include a “text field” node with the “boilerplate-master-NDA” variable. This node will make the preview possible.
The last step is to save inputs and display results add the “send & result” node.
I hope this video has served as a source of inspiration and guidance for you to automate any kind of standard contracts and provide AI guidance to your colleagues or clients.
Happy automating with e!
