Kant AI Solutions LLC

Kant AI intends to transform emergency response through applications of AI.  The Kant AI Team has extensive experience in public safety communications and homeland security missions.  We have applied this experience to develop patented-AI solutions (US Patent Nos.: 11,395,124 and 11,889,398) empowering PSAPs to better serve the public, without requiring changes to existing technology and only requiring fundamental training.

Our solution contains the following functionality:

• Assists call takers operating in PSAPs by providing accurate assessments of emergencies and offering recommended actions that can be taken during a specific 112 call, and
• Enhances situational awareness by bridging PSAPs with the HSE detection of large-scale incidents.

This dual functionality of our AI solution not only improves the way in which PSAPs can process and respond to specific emergencies, but also breaks down the silos between 112 and the jurisdiction’s emergency management organization.  Some of the most serious threats facing Americans today may initially be invisible, and the first signs that a chemical or biological threat agent has been released – either deliberately or accidentally, may be affected citizens calling 112, and that Emergency Managers need to activate major response operations.

Building on this overarching solution, our patented-AI capability is cloud-based, and does not require any hardware or installation.  Specifically, our solution:

• Listens in the background to provide recommended questions to ask and actions that can be taken during an emergency,
• Is trained to make accurate assessments of developing emergencies and offering recommendations for immediate actions which may be taken based on trusted sources of data and protocols,
• Improves upon protocol systems currently available by eliminating manual input and second guessing,
• Improves capabilities and protocols through real life experience,
• Understands and speaks multiple languages and can immediately translate for both the caller and call taker in a conversation where the two parties do not share a language,
• Can anonymously aggregate caller information in the background to predict potential emerging larger-scale incidents.  This could range from an accidental release of toxic chemicals or a biological outbreak - blind to the human eye but visible to Carla AI - based on symptoms being reported to emergency call centres across a geographic region,
• Is capable of migrating e into the next generation 112 environment, and operating e with multi-media communications such as video, photos, and text, and
• Is consistently retrained through human feedback to improve quality of assessment and response.

We call our solution “Carla” has the following characteristics:

• Cloud-based solution built upon Microsoft Azure.
• Access to the full suite of Azure AI solutions, including language translation, call recording and transcription, and cognitive machine learning services.
• Reliable security to preserve data integrity for the PSAP.
• Leverage any input sources the PSAP can receive (voice calls, photos, videos, text, etc.)
• Allows for integration of data from existing emergency response tools and other existing AI solutions.

Carla is non-intrusive and can be easily deployed and operated in any PSAP and EOC.  Carla does not interrupt or stand between any existing 112 infrastructure or process. All protocols for permissions remain under the control of the PSAP and EOC.

Carla is a machine learning tool trained on trusted databases and real-life emergency calls with human feedback.  Carla does not replace a call taker or any other human role in emergency response.  Rather, Carla is a virtual assistant that does not require integration or change in existing hardware or software used in PSAPs.

Carla provides the following functionality:

Call Diagnosis and Processing. Carla, as an AI engine, is structured initially to “listen in” on 112 emergency calls received by a PSAP.  Carla appears as a chat box to the call taker, configured to the call taker’s preference, and set to the call taker’s language preference.  When a 112 call is received through PSAP’s existing call taking system and the call taker inputs data into PSAP’s existing computer aided dispatch (CAD) system, Carla begins automatically assessing the nature of the emergency.  If the call is regarding a health emergency, Carla will suggest questions the call taker can ask of the caller to gather more information, and as Carla gathers data from the conversation, Carla will make recommendations to the call taker on immediate response actions, while first responders are dispatched.  Carla is trained in trusted and reliable medical protocols and databases and is perfected from human feedback to improve accuracy.

Language Translation. In addition to diagnosing and assisting on the nature of the emergency call, Carla also will provide automatic language translation when the caller is speaking a different language than the call taker’s language preference.  The call taker sees in text the translation from the caller’s language to the call taker’s language in the chat box.  The call taker then communicates back to the caller by typing instructions to Carla about what to say to the caller or by verbally speaking to Carla. Carla then speaks to the caller in their native language.

Automatic Call Recording and Transcription.  All calls and text communication between the caller and call taker are automatically recorded, transcribed, and saved into the cloud in a time and date stamped record that can be retrieved by the PSAP at any time.

Threat Detection Capability.  Carla is also designed to detect large-scale threats such as chemical, biological, health outbreaks, and suspicious activity.  Carla is trained to detect key words, patterns, anomalies as well as connecting-the-dots among all emergency calls emanating from the same area or over a region.  As the call or calls continue, Carla is constantly querying the trusted databases that enable her to estimate a specific threat.  When a threat is detected, Carla will automatically send an alert to an EOC and other Emergency Management personnel outside the PSAP about the potential threat.  No personally identifiable information from any caller is ever compromised or sent out of PSAP.  Carla only sends alerts which only contain specific information prescribed by the PSAP and the jurisdiction’s EOC.

Benefits of a Cloud Solution

Hosting Carla, our virtual machine learning tool in the Azure cloud allows us to have complete control of the application.  Azure cloud hosting enables access to Carla on any device, allowing control of the data that “educates” Carla.  Kant AI works with our databases using GPT-4 as a foundation.  Our data adds both language and speech processing trained using our specific cases of emergency response and awareness databases.  Responses are updated from testing and real-world experience with human input to correct incorrect responses and further refine our system’s effectiveness in emergency response.