adding note in how it works section

README.md

@@ -1,10 +1,10 @@
 # Fair Housing Guardrail
-Fair Housing Guardrail is a tool designed to help housing providers, real estate professionals, and related organizations review compliance of textual content with fair housing and fair lending laws in the United States, specifically the federal Fair Housing Act and the Equal Credit Opportunity Act. By leveraging a combination of stop list and fine-tuned BERT-based classifier which offers low latency and ease of fine-tuning compared to LLMs, Fair Housing Guardrail identifies content that may potentially violate these laws, allowing users to make necessary adjustments to help maintain compliance.
+Fair Housing Guardrail is a tool designed to help housing providers, real estate professionals, and related organizations review compliance of textual content with fair housing and fair lending laws in the United States, specifically the federal Fair Housing Act and the Equal Credit Opportunity Act. By leveraging a combination of stop list and fine-tuned BERT-based classifier which offers low latency and ease of fine-tuning compared to LLMs, Fair Housing Guardrail is intended to identify content that may potentially violate these laws, allowing users to make necessary adjustments to help maintain compliance.
 
 ## Overview
-For a comprehensive overview of Fair Housing Guardrail, including its features, usage instructions, and implementation details, please visit our [blog](https://www.zillow.com/tech/navigating-fair-housing-guardrails-in-llms/).
+For a comprehensive overview of Fair Housing Guardrail, including its features, usage instructions, and implementation details, please visit our [blog](https://www.zillow.com/tech/navigating-fair-housing-guardrails-in-llms/)
 
-In use cases which use an LLM, we recommend using Fair Housing Guardrail in combination with a fair housing specific prompt as we found that it yielded better coverage in violation detection and prevention. Refer to the ‘Prompt Engineering’ section [here](https://www.zillow.com/tech/navigating-fair-housing-guardrails-in-llms/) for more details.
+In use cases which use an LLM, we recommend using Fair Housing Guardrail in combination with a fair housing specific prompt as we found that it yielded better coverage in violation detection and prevention. Refer to the ‘Prompt Engineering’ section [here](https://www.zillow.com/tech/navigating-fair-housing-guardrails-in-llms/) for more details. 
 
 ## Features
 **Compliance Detection**: Fair Housing Guardrail utilizes advanced natural language processing techniques to identify content that may violate fair housing and fair lending laws.
@@ -15,11 +15,13 @@ In use cases which use an LLM, we recommend using Fair Housing Guardrail in comb
 
 **Fair Housing Policy**: Specifically, the model was trained to address the risk of illegal “steering” in real estate. In the traditional brick and mortar sense, this is when a real estate agent takes into account the legally-protected characteristics of their client when determining which listings to show them. A well publicized example of this was documented in a three-year investigation by [Newsday](https://www.google.com/url?q=https://projects.newsday.com/long-island/real-estate-agents-investigation/&sa=D&source=docs&ust=1712163974278248&usg=AOvVaw3U1PPg4BEXVJm_kQ17UB5f), published in 2019. Extensive fair housing testing or “secret shopping” of real estate agents on Long Island, New York was conducted, finding that many would direct different clients to different neighborhoods, depending on the race or ethnicity of the client, and the racial and ethnic demographics of the neighborhood.  
 
-**Fine-tuned model and labeled dataset**: Refer to ‘Contact Us’ section to request this.
+**Fine-tuned model and labeled dataset**: Refer to the ‘Contact Us’ section to request this.
 
 
 ## How It Works
 Fair Housing Guardrail works by taking input text and running it through both the stop list and the fine-tuned BERT-based classifier. If the text contains any flagged terms from the stop list or if the classifier determines it to be non-compliant with the fair housing and fair lending guidelines that it was trained on, the tool will label the content accordingly, providing an opportunity to review and revise the content as necessary.
+
+Note that this project was designed to provide low latency, standalone and easy-to-tune guardrails for Fair Housing and Fair Lending compliance of conversational or text based experiences. The fine-tuned model (provided upon request) is an ongoing work in progress and we will be iteratively improving its accuracy.  We invite you - the developers and AI practitioners - to help us identify gaps, submit improvements via pull requests as well as share your feedback directly with us via email at [[email protected]](mailto:[email protected]).
 
 ## Usage
 After cloning the repo to a local directory, you can install the necessary dependencies with poetry.