Skip to main content

Enei AI

Artur Inozemtsev avatar
Written by Artur Inozemtsev
Updated yesterday

Enei.ai is a tool for automatic communication analysis (calls, chats) that helps companies improve customer service quality and manager efficiency.

Instead of manually listening to hundreds of conversations, Enei automatically evaluates dialogues based on specified parameters, generates analytics, identifies strengths and weaknesses, and provides insights for making management decisions.

Main Features

  • Dashboard — a general dashboard with key metrics for a quick overview of the situation.

  • Review Journal — detailed analysis results for each conversation with filtering and appeal options.

  • Analytics — summary reports by departments, teams, or individual managers, performance dynamics, identification of weak points.

  • AI Insights — AI answers to any questions about already analyzed data (for example, "What are the most common customer objections over the past month?").

  • Change Requests (appeals) — the ability for managers or supervisors to challenge analysis results.

  • File Upload — uploading audio or chats for analysis, both individual and bulk.

  • Prompts (checklists) — settings that determine what exactly needs to be evaluated in conversations and by what method.

  • Integrations — settings for connecting Enei.ai to external services: telephony, support systems, or other business tools.

Dashboard

The Information Panel is a general dashboard that provides a quick overview of all analyses conducted over a selected period. It helps understand overall dynamics, workload volumes, and key metrics without needing to dive into details.

1) Main Metrics

The top section of the panel displays key metrics:

  • Total Analyses — the number of communications analyzed within the selected date range.

  • Total Call Duration — the total time of all conversations that were analyzed.

  • Representatives — the number of unique managers included in the analysis.

2) Daily Analysis Trend

Shows a breakdown of analyses by day.

Allows you to quickly track peak days, activity drops, or processing consistency.

3) Analyses by Prompt

Reflects how many conversations were analyzed according to each created checklist.

4) Top Representatives

A list of managers with the highest number of analyses conducted.

For each one shown: number of analyses and percentage of the total.

Review Journal

Review Journal displays all AI-analyzed communications - calls/chats and allows you to quickly access details, submit appeals for score changes, filter, and build your own table views.


Column and View Configuration

1) You can customize the display of exactly the data that interests you. To do this, click the "Customize Columns" button.

2) In the "Add Columns from Fields" block, select the desired Prompt Template so that the fields of this specific template appear.

3) Open the "Prompt Structure Fields" tab and select the parameters you want to see in your view.

4) You can select the type of data that will be displayed in the field, the width and length of the column.

5) Apply the changes and save the view.

6) Now you can select the created view from the list by clicking "Select View"

Viewing a Conversation Card

To open a card with conversation analysis, click the green eye icon button in the "Actions" column.

The card contains three sections:
1) Analysis Details — AI scores and comments for each parameter in the prompt

2) Transcription — text transcription of the conversation with timestamps

3) Audio Player — call audio that can be listened to

Change Request

If the AI score is incorrect (for example, the AI didn't recognize something or gave a score different from what the manager should have received), the employee or their supervisor has the ability to submit a change request:


1) In the analysis card, click the pencil icon next to the required parameter.

2) Set the parameter that you believe is correct, or make changes to the text and describe in the comment the reason why the parameter needs to be changed.

3) Submit the change request

Submitted requests are displayed in the "Change Requests" section (in the side menu) for further processing

A request can be submitted by an employee who has access to the account, as well as by their supervisor or account administrator.

Analytics

Analytics allows you to automatically collect and analyze data about calls and chats without manual work. In a few clicks, you get summary metrics for employees or departments for any period. This tool helps quickly identify productive team members, detect problem stages in the funnel, and explore customer inquiry topics.

How to Build Analytics:

1) Select a prompt template

2) Set the date range for which you want to take communications for analysis
3) Configure the blocks:

  1. Group By Fields - You can choose exactly how to group the results; grouping can be done by any data, but it is most appropriate to group by managers or by departments. This provides different levels of detail and answers different questions.

  2. Data for evaluation:

  • Score Parameters (numerical) — (average scores for conversation stages, final score, etc.)

  • Text Parameters — data that shows what exactly managers and clients said: topics, needs, objections, key phrases

4) Click "Generate Analytics"

5) After generating the report, you will see:
1. Total number of analyzed communications and the period for which they were taken.

2. Overall average scores for the department/departments or selected managers

3. Average scores by the type of data you selected for grouping (representatives, departments, date, etc.)

4. Data can be grouped by various indicators, including by days, which will allow you to evaluate the average dynamics of your employees.

5. Text parameters

Important:

  1. Text parameters will be available only if they are present in your prompt (that is, you evaluate the conversation not only on a specific scale, but also receive a description of certain stages)

  2. You can select the fields you need that you want to include in the generated analytics, depending on which metric/metrics you need to track.

  3. Text parameters will be especially useful if we correctly set the task for the AI to collect data in our chat.

Text parameters can be a regular text array (1) or a list/enumeration (2) of the data we specified for tracking.

The "Percentage" column reflects the percentage share of each type of objection from the total number.

For example: "Cost" occurred 2 times, which is 22.2% of all 9 recorded objections.

The rest ("Information", "Employment Guarantees", "Training Schedule", etc.) 1 time each — that's 11.1% each.

That is, this column displays the percentage of a specific parameter among all available in the analysis category.

Data Filters

Allow you to limit the sample before building analytics to analyze exactly the data set that is important now (for example, taking only one department, newcomers to the team, specific call types, a separate product line, etc.)

To configure a filter:
1) Click "Add filters"

2) In the "Field" block, select a parameter for filtering

3) In the "Operator" block, select the condition by which you want to filter data

4) In the "Value" block, enter exactly what you want to filter (manager name, department name, keyword)

AI Insights

AI Insights eliminate manual work with reports. In a few clicks, you get a summary from dozens/hundreds of already analyzed conversations: who is performing better/worse, where the "bottlenecks" are in the funnel, which topics or objections are heard most often, how quality changes over time. You don't need to open 30–50 cards — you see a ready-made data selection based on your request.

To create a new Insight:

1) Go to the "AI Insights" section and click the "New Insights" button

2) Question — Enter your question or click one of the suggested prompts

3) Select the date range for which the analysis needs to be conducted

4) Prompt Template — Select the prompt template by which the communication analysis was conducted
5) Department / Representative (optional) — Limit the insight to the needed team or employees. If not selected — all are taken.

6) Provider and Model — for deeper analytical answers, Anthropic or OpenAI are recommended. Lighter models (from Groq) may be faster, but less detailed.

7) Field Selection (Optional) — Determines from which specific fields the AI prompt will take data

8) Specify a name that is clear to us and click "Create Insight"

Upload Files

You can upload your own audio files for analysis. There is an option to upload either one file or several simultaneously

Single Upload

1) There are two options for uploading audio:

1.1 "File Upload" — drag the desired file into the upload field. Or click on the upload area and select the desired file on your computer

1.2 "Upload from URL" — paste a link to the file:

  • from Ringostat telephony

  • from third-party telephony where authorization is not required for listening

  • from Google Drive — the file/folder must be shared with editing rights to our system account ([email protected])

2) Select which prompt will be used to analyze the audio

3) Priority (optional) — by default we leave normal, but if you need to analyze the file faster than others, change the priority to high
4) Schedule (Schedule for later) — you can analyze the file later. In this field, specify the date and time to start the analysis

5) Metadata Fields - allow analysts to group analysis data by specified parameters

6) Click "Upload Audio for Analysis"


Bulk Upload (CSV)

1) Before uploading, you need to prepare the file so that it meets all requirements, so download the template

The template must contain:

  • url (required) — full URL to the audio file

  • type (required) — name of the prompt (checklist) to be used for evaluation

  • date (optional) — call date

  • any additional fields — will be saved as metadata (for example, Department, Representative, etc.)

2) When our template is ready, go to the "Bulk Upload (CSV)" tab

3) Select which prompt will be used to analyze the audio files.

4) Drag the desired file into the upload field. Or click on the upload area and select the desired file on the computer.

5) Click "Upload (CSV)"


Prompts

Prompts — allow you to create and save checklists (analysis templates) for artificial intelligence. The accuracy of AI's ability to analyze your managers' calls depends on the quality and structure of the prompt.

In a checklist, you can:

  • set an evaluation on a scale (for example, from 0 to 5 or from 1 to 10);

  • receive from AI a description of strengths and weaknesses at each stage;

  • extract an array of text from the conversation (for example, key phrases or manager quotes);

  • collect lists of words, stop-words, or triggers that were used;

  • record specific actions (for example, whether the manager introduced themselves, clarified needs, offered a solution).

Creating a New Prompt

1) Go to the "Prompts" section and click "New Prompts"

2) Describe your requirements in the AI prompt generator and click "Generate Prompt Structure"

AI may need some time to generate the prompt structure

3) In the "Structure" section, you can see each stage of the checklist, delete or add the necessary parameter

4) In the "Basic Settings" tab, you can specify:

  1. Script Name

  2. Project - The project it belongs to

  3. Role - AI role (for example, sales or quality department manager)

  4. Provider and Model AI (Anthropic, OpenAI, Google, Groq)

  5. Response Language - Response language in which we will receive the analysis result, and the transcription task (translate or transcribe)

  6. Transcription Language

  7. Final Score — set how we want to see the calculation of the overall communication score (sum, percentage)

  8. Max Score - set the upper limit of the final score (AI automatically sums all score fields and displays the overall result)

  9. LVA — this is multi-level voice analysis, which is used for in-depth assessment of the interlocutor's state during a phone conversation. The technology analyzes not only words, but also intonations, pauses, voice timbre that a person cannot consciously control

  10. Transcription Instructions (for example, complex terms, names, brand names that AI might have difficulty understanding)

  11. CompanyDetails — this way AI will better understand your niche

  12. Basic Instruction for AI, from which it will draw information for evaluating communications (context for analysis)

What LVA can determine:

  1. level of stress and tension during a response;

  2. signs of uncertainty or evasion from a direct answer;

  3. possible reliability/risk of untruthfulness of what was said;

  4. emotional and psychological state of the client at the time of the conversation.

5) In the "Metadata" tab, you define additional parameters for analytics.
By default, available:

  • Department,

  • Representative.

There is also an option to add your own, for example: product, contact channel, customer geography, etc.

This will allow you to group analytics in the future not only by managers or departments, but also by other criteria


AI Improve

Each prompt has an AI improvement tool.

It allows you to quickly modify the checklist without manual editing:


1) Open the editing of the ready prompt

2) Click the "AI improve" button

3) Describe what needs to be changed (for example, "add detailed criteria for identifying needs") and click "Improve Current Prompt"

AI automatically updates the structure while keeping the basic logic unchanged.

Did this answer your question?