Generative AI Agent

Generative AI Agent offers various business capabilities with high degree of customization tailored to specific industry, business practice and domain knowledge.

Scenario:
An organization needs a quick integration of AI into their business process to minimize costs through increased employee productivity, improved customer service and automation.
Utilizing Generative AI Agent, the organization can quickly integrate customized business assistants into current technology stack with minimum disruption, maximum benefits and quick ROI.

Capabilities

• Integrate Natural Language Processing into Customer Service, Business Intelligence and Knowledge Mining.
• Generate content, used in customer outreach or marketing.
• Minimizing language limitation, via automated translators.
• Pre-training QnA bots tailored based on customer purchase history, product knowledge, or frequently asked questions.
• Image generation often utilized by internal departments i.e. HR, Marketing
• Offering RAG applications, offering Subject Matter Expertise trained on organization's collective knowledge.
• Offering out of the box companionate customer service, being able to handle impolite customers and de-escalation.
• Offering out of the box ethical behavior.

Prompt Engineering

Prompt Engineering is used to refine user's question in order to significantly improve the quality of model's response.

Scenario:
In a chatbot, a user asks a question that is ambiguous or unclear.
Utilizing Prompt Engineering, the chatbot can ask the user to clarify the question, thus providing a more accurate response. User's response can be channeled back into the model to improve future responses.

Scenario:
A customer calls a customer service line and asks a question about a product to be used in a specific way.
Utilizing Prompt Engineering, we can refine the question to provide a more accurate response, thus improving customer satisfaction, by generating a response specific to the product and the use case scenario provided by the customer.

Scenario:
An organization wants to improve chatbot customer service, by providing more accurate responses based on product knowledge and various applications.
Utilizing Prompt Engineering, we can pre-load the chatbot with products and their applications, then using is in the response generation process. Thus providing a very accurate and specific response to the customer's question, instead of an ambiguous response if the product knowledge was not pre-loaded.

Capabilities

• Chat-role playing scenario, generating chat completions
• Response tuning for higher/lower relevance
• Response tuning for higher/lower accuracy
• Response output formatting specific to i.e. report-like output, email
• Article based response i.e. provide summary, translate
• Grounding based response i.e. responses based on article's content
• Cues prompting i.e. "What are most common complaints?"
• System message setup i.e. "you are a customer service representative specializing in interior design"

Personally Identifiable Information (PII) Agent

PII Agent specializes in detecting private information which should be or sometimes must be protected.

Scenario:
A user submits a chat entry and includes his/her SSN.
Currently, the SSN will be included in the chat.
PII Agent can detect the SSN and notify the user that the SSN should not be included in the chat. Or the chatbot can automatically remove the SSN from the chat.

Detection Capability

• Email
• US Mailing Address
• IP Address
• Social security Number
• Phone Number

Video Agent

Video Agent is capable of extracting a number of information from videos.

Scenario:
An security personnel reviews a camera footage to identify that a non-employee has entered the facility.
Currently, the entire footage must be viewed.
Vision Agent can scan the footage and generate a summary of events with time stamps dramatically reducing the review process time.

Detection Capability

• Face Recognition
• Face Comparison
• Face Features Detection
• Text Detection (Optical Character Recognition)
• Speech Transcription (when sound available)
• Language Detection (when sound available)
• Adult and/or Violent Content
• Content Labeling
• Logo Recognition
• Scene Segmentation (with timestamps)

QnA Agent

QnA Agent specializes in understanding and answering specific questions.

Scenario:
An automated chatbot is created to answer frequently asked questions.
QnA Agent can be trained to answer questions that are specific to an industry, subject or a knowledge domain. It conducts a natural and conversational dialog with the user, thus providing a more engaging experience.

Capabilities

• Extracting utterance and actions.
• Interpretation of the utterance and matching with an intent and objects to which the action should be applied.
• Generating responses based on probability or based on pre-defined answer
• Generating responses based on pre-defined answer
• Apply specific actions based on the questions (i.e. reservation cancellation)

Business Intelligence (BI) Agent

BI Agent specializes in ingesting and understanding business data thus offering BI based on knowledge mining.

Scenario:
An organization wants to expand customer service to include previous purchase history.
Utilizing BI a customer service representative can quickly be provided options specific to current conversation, thus providing a more personalized experience.

Scenario:
An organization wants a quick access to business data, during a meeting.
Utilizing BI, the meeting participants can ask questions and get answers in real time, based on available data and content specific.

Capabilities

• Build custom skills specific to data and business process
• Protect business sensitive data or Intellectual Property (IP) by running on-prem with Knowledge Mining Capabilities
• Correlate on-prem data with external search results
• Provide search result statistics: ranking, relevance
• Data embedding
• Perform vector search to improve performance