It’s a tale as old as time: one minute you're trying to hammer out a detailed plan for a project, the next you're scratching your head, unsure of the right steps. It’s time to say goodbye to confusion and hello to clarity - we’ve introduced generative AI to Prioritize to expedite your project planning like never before.
Introducing AI-Suggested Subtasks
It all starts with Task Templates – custom, repeatable workflows that ensure consistent intake every time. However, without a foundational understanding of what’s needed to round out the process, developing the right steps for your templates can be time-consuming and lead to process gaps.
With AI-Suggested Subtasks, generative AI cuts out the confusion. Simply tell the system what you need. For example, ask the AI to “create a list to review marketing collateral", and it will generate suggested steps you can incorporate into your Template. Accept, decline, or regenerate the suggestions to create a list of subtasks that fits your specific needs or processes. The next time a task is generated, your chosen subtasks will automatically populate, not only saving you time, but ensuring the appropriate steps are always followed.
Generative AI may be a new addition to the LinkSquares Cloud, but we're no stranger to AI for legal teams. AI has been core to our product since its founding, but we understand it can be overwhelming to understand the different approaches to AI and where you, the user, fits in. AI is the umbrella term for a wide range of techniques that allow computers to perform tasks previously done by humans. At LinkSquares, we employ both native, predictive AI as well as generative AI. Let’s dig into how these approaches differ and how each is trained.
Our approach to native AI
In the broadest terms, LinkSquares’ native, proprietary AI is supervised machine learning (ML), exclusively trained on billions of legal words and phrases. To train our models, AKA Smart Values, we begin by identifying the language we are looking to extract, such as a clause or date field. We then annotate agreements noting where the language appears in the document. Next, we train the model on the labeled data, so it learns to identify where that specific language appears given the surrounding context. Finally, we evaluate the accuracy of extractions before releasing them for customer use - we only release models that meet our stringent accuracy standards. To keep models current, we consistently audit outputs and boost models as needed by exposing the model to language it hasn’t seen yet.
Our approach to generative AI
In contrast to machine learning, generative AI has the ability to understand prompts and create new, original content based on a set of instructions. This is the result of large language models (LLMs) that are pre-trained on vast amounts of data from a wide variety of sources - essentially the entire internet. We aren’t training the LLM ourselves nor are we relaying any data to that model that is used in its training. Rather, we’ve built features across the LinkSquares Cloud that interface with a customized instance of the model that is tailored to legal use cases - some examples being text summarization, task suggestions, and contract review to name a few. Incorporating generative technology puts the user in control, offering a baseline to work from vs. starting from scratch every time.
Keep in mind that:
1. Generative AI is not a replacement for predictive AI and
2. AI is not a replacement for you.
Both approaches complement one another, ultimately helping legal professionals work more efficiently than ever before. AI is another tool in your belt and allows you to take control of your work vs. completely automating it away.
Interested in having AI-Suggested Subtasks activated in your account? Contact your CSM today!
