Prompts

This module is for creating and managing Prompts. They pair parameters into wholes that are then used for processing audio (transcriptions) and text inputs.

Configuration

On-screen elements:

• Details:

  • Name: input field of the prompt

  • Language: drop-down field for processing the prompt

  • Prompt state: dropdown field for changing the state to one of these: "Enabled", "Disabled for all" (completely disabled), or "Disabled for rerun" (disabled only for any of the following reruns) states

  • Multi-segment analysis (input longer than 110k tokens, ie >10 hours of speech or 40k words in an email): radio buttons for choosing which part of a long, multi-chunk record this prompt should read: use "Each" chunk (for sentiment analysis), only the "First" chunk (for greetings), or only the last chunk (for farewell).

Why the Multi-segment analysis setting exists

In the past, the LLM models used to have smaller "context" - ie the length of audio or text which we sent for analysis had to be broken into several chunks before processing, analysed in sequence and summarized.

However!

Nowadays models have no problem with long context. The chunking is only applied in these situations:

• recordings longer than cca 10 hours

• emails longer than 30-50k words (depends on the content, images, email structure)

• texts longer than 80k-90k words

In case you are using such long inputs, the Multi-segment analysis control gives you a say in how a prompt reacts to that split: it lets you decide whether the prompt should read every chunk, just the opening segment, or only the final one.

• Parameters:

  • Predefined parameter group: input field for selecting already predefined hardcoded parameter groups. Once any of them is selected, its predefined, hardcoded parameters are then listed in the "Parameters" field below.

  • Parameters: input field for selection of parameters (custom-made or predefined)

Predefined (hardcoded) parameters list and their description

addressing	Value YES if the Agent used the Client’s first name or last name in the conversation. Value NO otherwise.		YES, NO	string	NO	WEIGHTED_VALUE
benefit	Did the agent mention the product benefit?		YES, NO	string	N/A	FIRST
call_script_summarization	Agent summarized Client's request to check if he understood correctly		NO, YES	string	NOT_FOUND	WEIGHTED_VALUE
call_script_telco_offering_accepted	Client accepted mobile phone service, package or tariff that Agent offered to him			string	NOT_FOUND	WEIGHTED_VALUE
call_script_telco_offering_value	Name of the offering, service, package or tariff that Agent offered to Client to buy, add or activate			string	NOT_FOUND	LAST
category	Most dominant topic Client wants to discuss			string	NOT_FOUND	FREQUENT
category_product	categorize a product of selling		insurance, savings, loans, investment, credit card	string	other	FIRST
client_behaviour_goodbye	Client's goodbye	Value YES if client said goodbye, else value NO	NO, YES	string	NOT_FOUND	LAST
client_behaviour_thanks	Client's client_behaviour_thanks	Value YES if client thanked for help, else value NO	NO, YES	string	NOT_FOUND	LAST
cross_sell	Value YES, if the Agent tried to sell more than just 1 product, or tried to cross sell, upsell another product. Value NO otherwise.		YES, NO	string	NO	FIRST
empathetic	Value YES if the Advisor responds empathetically to the Clients concerns and questions. Value NO otherwise.			string		FIRST
extract_client_customer_contract_number	Client's customer number or customer contract number			string	NOT_FOUND	FIRST
extract_client_email	Client's email			string	NOT_FOUND	FIRST
extract_client_first_name	Client's first name			string	NOT_FOUND	FIRST
extract_client_insurance_number	Client's insurance number			string	NOT_FOUND	FIRST
extract_client_last_name	Client's surname			string	NOT_FOUND	FIRST
extract_client_phone_number	Client's phone number			string	NOT_FOUND	FIRST
extract_client_service_number	Client's service number			string	NOT_FOUND	FIRST
farewell	Value YES if the Agent said goodbye or wished the Client a nice day. Value NO otherwise.		YES, NO	string	NO	WEIGHTED_VALUE
future_interest	Value YES if the Client has expressed interest in a certain banking product in a certain time frame, such as another day, later, another time, sometimes in a few weeks, months, in half a year etc. Value NO if the Client has no interest in any banking product at all and when client does not mention interest then value No interest mentioned		another day, later, another time, sometimes in a few weeks, months, in half a year, No interest	string	No interest mentioned	FIRST
future_interest_ii	Value YES if the Client has expressed interest in a certain banking product in a certain time frame, such as another day, later, another time, sometimes in a few weeks, months, in half a year etc. Value NO if the Client has no interest in any banking product at all and when client does not mention interest then value No interest mentioned		YES, NO	string	NO	FIRST
gathered_category	Choose category that fit this conversation the most			string	NOT_FOUND	FIRST
greeted	The value YES, if the agent greeted, introduced himself, said the name of the company. The value NO otherwise.		YES, NO	string	NO	FIRST
greeted_end_of_the_call	Yes , if agent greeted at the end of the call, otherwise NO		YES, NO	string	NO	FIRST
handle_it	How well did the agent handle these objections? categorize it			string		FIRST
handle_objections	Value YES if the Agent is able to handle Customers objections well. NO otherwise.		YES, NO	string	NO	FIRST
identification	Value YES if the Client’s first name, last name and date of birth were mentioned in the conversation. Value NO otherwise.		YES, NO	string	NO	WEIGHTED_VALUE
ideology_ukraine	True if message supports Ukraine or is againts Russia. False if message supports Russia or is againts Ukraine. Neutral if message is neutral about this topic.		False, True, Neutral	string	Neutral	WEIGHTED_VALUE
interest	rate the clients interest about product and categorize it		High, Moderate, Low	string	No interest	FIRST
interested_when	When will the Client be interested in the selected banking product? Return exactly one of the following options that is closest to what the Client said: Days, Week, 2 weeks, A few weeks, 3 weeks, Month, A few months, 2 months, 6 months, Year.		Days, Week, 2 weeks, A few weeks, 3 weeks, Month, A few months, 2 months, 6 months, Year, Tomorrow	string	Not mentioned	FIRST
introduction	Value YES if the Agent introduced themselves with their first name, last name, and company name. Value NO otherwise.		YES, NO	string	NO	FIRST
mifid_financial_products	Financial products that where mentioned or discussed in this Conversation by Agent of Client	string with values separated by comma		string	NOT_FOUND	CONCATENATE
mifid_financial_products_explained	If Agent explained risks or details about discussed financial products		False, True	string	FALSE	WEIGHTED_VALUE
mifid_relevant	True if conversation included any information that needs to be Mifid II regulated		False, True	string	FALSE	WEIGHTED_VALUE
obscene	All profanity, obscene words that are in the transcript. NOT_FOUND if none occur.			string		FIRST
other_help	Value YES if the Agent asked the Client whether they had any other requests or if there was any other way they could help. Value NO otherwise.		YES, NO	string	NO	WEIGHTED_VALUE
personalizing_offer	Value YES if the Agent tried to personalizing the offer to Clients needs/situation, rather then just presenting product as on a leaflet. Value NO otherwise.		NO, YES	string	NO	FIRST
react_to_offer	How did the customer react to the offer? categorize it		Positive, Interest, Disinterested, Disappointed, Polite, Neutral	string	Not mentioned	FIRST
recapitulation	Value YES if the Agent summarized the conversation and verified the contact details with the customer. Value NO otherwise.		YES, NO	string	NO	WEIGHTED_VALUE
rejection	List the main reason why the client does not want to purchase the product or service, why they are not interested.		Price Objection, Lack of Need, Competitor Preference, Budget Constraints, Lack of Understanding, Not the right time	string	Not mentioned	FIRST
rejection_reason_description	Summarize the main reason why the client does not want to purchase the product or service, why they are not interested.			string		FIRST
selling	Did the agents sell the product?		YES, NO	string	N/A	FIRST
sentiment_agent	Agent's sentiment at start of conversation		Helpful, Friendly, Unfriendly, Aggresive, Neutral, Other	string	Neutral	FREQUENT
sentiment_client_end	Client's sentiment at end of conversation		Relief, Gratitude, Frustration, Confusion, Anger, Neutral, Other	string	Neutral	LAST
sentiment_client_start	Client's sentiment at start of conversation		Relief, Gratitude, Frustration, Confusion, Anger, Neutral, Other	string	Neutral	FIRST
subcategory	Subcategory to the most dominant topic Client wants to discuss			string	NOT_FOUND	FREQUENT
subsubcategory	Subcategory to the most dominant topic Client wants to discuss			string	NOT_FOUND	FREQUENT
suggest_category	Most dominant topic Client wants to discuss			string	NOT_FOUND	FREQUENT
suggest_subcategory	Subcategory to the most dominant topic Client wants to discuss			string	NOT_FOUND	FREQUENT
summary	Summary of what was Conversation about	Maximum 20 words		string	NOT_FOUND	CONCATENATE
summary_long	Summary of what was Conversation about	Maximum 40 words		string	NOT_FOUND	CONCATENATE
summary_short	Summary of what was Conversation about	Maximum 10 words		string	NOT_FOUND	CONCATENATE
topic_dominant	Most dominant topic this Conversation was about			string	NOT_FOUND	FREQUENT
topic_first	First topic Client wants to discuss			string	NOT_FOUND	FIRST
topic_nondominant	Second most dominant topic this Conversation was about			string	NOT_FOUND	FREQUENT

• Previous parameter: input field. If any parameter is selected, then its outcome is used as an input for this prompt. Example: summary of the audio record (prompt_summary)

You can only select parameters that are used by other prompts. Thus, the naming convention is in this format: <prompt>_<parameter>.

• Instructions: input field for additional prompt specifics. E.g., output file type, processing specifics, structure requirement, etc. It is mainly used in some specific cases to fine-tune the output.

• Conditions: section for specifying cases, based on values of already processed parameters, on which the current prompt should be applied:

  • Condition: dropdown field for selection of logical operators "AND" and "OR".

  • Add new condition: link displays upon clicking the following fields below. You can add multiple conditions at once.

    • Parameter: dropdown field for selection of parameters from other prompts in the format <prompt>_<parameter>

    • Value: input field displaying values based on the selected parameter

• Temperature: input field for changing the creativity of the model output. Lower values (closer to 0) make output more deterministic, focused, and consistent (sometimes called “stricter”). Higher values (such as 0.7 or 1) increase creativity and diversity, making the model more exploratory in its responses.

• ⚙️ Provider:

  

  • Create provider: button for opening a form for the selection of LLMs that will be used for processing

    • Provider: dropdown menu with options representing LLM providers (Azure OpenAI, Anthropic, Gemini, Groq, or OpenAI)

    • API Key: input field

    • Model name: input field

    • API base: input field for entering URL

    • API version: inout field for

  • Remove: button

• Create: button for creating the prompt with the latest configuration

Tip: Use the predefined parameters for inspiration, but add little details for your individual use-case. (Ie exact call script for your agents to introduce to customers, like: The value YES, if the agent greeted, introduced himself and asked "How can I help you?". The value NO otherwise.

vs. the default:

The value YES, if the agent greeted, introduced himself, said the name of the company. The value NO otherwise.

Management

• The order of prompts defines their running order.

On-screen elements:

• Upper part:

  • Search prompts: bar for searching parameters based on their name

  • + Create: button for creating a new prompt

  • Select: button for selecting multiple parameters in their list at once, and then for the following actions:

    • Prompt state: dropdown field for changing the state to one of these: "Enabled", "Disabled for all" (completely disabled), or "Disabled for rerun" (disabled only for any of the following reruns) states

    • Delete: button for deleting the selected prompt(s)

• List of prompts: (Elements are listed from the left)

  • Name: of the prompt

  • Pencil: icon button for editing the prompt

  • Command line: icon indicating the prompt state

  • Language: used for processing the prompt

  • Prompt provider type: displaying either "Default provider" or any other, as per the provider configuration

  • Parameters preview: displaying a preview of the parameters used by the prompt, upon hovering over them, their definition is displayed

• Floating buttons:

  • Export: for exporting all prompts in a JSON-formatted file

  • Import: for importing prompts in a JSON-formatted file

  • Save: for saving prompt changes