Esquire, the AutoParalegal

• 

  DRAFT workflow/API architecture diagram in Zapier

Inspiration

What if the lawyer worked for the paralegal? This disruptive notion presents an exciting use case for GenAI assistants.

What it does

Esquire allows you to upload a job description, conduct a matching/screening/discovery-type search and choose a LinkedIn profile to create a bot-/agent-type virtual assistant with client-/matter-specific skills, including grounding data from your company, a choice of fine-tuned legal industry-specific models, and validated prompt templates. It is an end-to-end RAG solution.

How we built it

We used Copilot Studio, including Generative Answers and Azure AI On Your Data with Zapier APIs to connect to Google Drive and LinkedIn. The Azure Cognitive Search LangChain integration, built in Python, provides the ability to chunk the uploaded job description documents, seamlessly connect an embedding model for document vectorization, store the vectorized contents in a predefined index, perform similarity search (pure vector), hybrid search and hybrid with semantic search. It also provides configurability to create your own index and apply scoring profiles to achieve better search accuracy. With LangChain, you can combine native workflows (indexing and querying) with non-native workflows (like chunking and embedding) to create an end-to-end similarity search solution. Cassidy.ai has a LinkedIn connector that we emulated. RAG in Azure AI Search: https://learn.microsoft.com/en-us/azure/search/retrieval-augmented-generation-overview Hybrid search using vectors and full text in Azure AI Search: https://learn.microsoft.com/en-us/azure/search/hybrid-search-overview https://techcommunity.microsoft.com/t5/ai-azure-ai-services-blog/azure-cognitive-search-and-langchain-a-seamless-integration-for/ba-p/3901448 https://python.langchain.com/docs/integrations/vectorstores/azuresearch/ https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/search/azure-search-documents/azure/search/documents https://learn.microsoft.com/en-us/linkedin/sales/analytics-services/

Challenges we ran into

Monetizing the idea. Adoption.

Accomplishments that we're proud of

Saving time. I am SO impatient.

What we learned

Users spend too much time on the character-building screen.

What's next for Esquire, the AutoParalegal

The Virtual Staffing Agency. This is the idea everyone is afraid of - will I lose my job to AI? Yes. We are building virtual skilled deepfake versions of legal professionals. The idea is to make it easy enough for an attorney to spin up one of these matter-scoped agents for an hour... a day... a week. Without having to call a hiring agency for a document review clerk or without hiring a developer to custom code. It is a no-code solution easy enough to use that an attorney could do it. Some of the skills belonging to the bot are AI-powered tasks themselves. So you might have a special paralegalBot that knows how to connect to FileVine, for example. But not all matters need that special API, so why pay for what you're not going to use?

Built With

• javascript
• langchain
• lucene
• openai
• python