Access control
Users and role permissions
| User | Role | Coach mapping | Status | Pages | Actions |
|---|
0% conversion rate
Eligible for AI analysis
Cannot be analyzed
AI processing complete
Ready to analyze
Lead queue
Sales intelligence table
Coach queue
All coach cases
Audit summary
Call conversion profile
High-signal fields generated from the transcript audit.
Audit engine
--
Structured form
Metric scorecard
Follow-up management
Task, disposition, and history
Track every follow-up interaction and next callback commitment.
--
--
--
--
Role inference
Speaker segmentation
Blockers
Decision friction
Evidence
Field support
Transcript
Full call transcript
Complete active transcript uploaded for this lead.
Version history
Transcript uploads
Step 1
OpenAI connection
OpenAI endpoint is handled automatically.
The app uses the official OpenAI API endpoint by default, so only the API key is required here.
Step 2
Audit model strategy
Step 3
Audit execution mode
Batch API is selected by default.
Normal audits will be submitted to OpenAI Batch API using the selected model. Choose Direct API only for urgent, small-volume audits.
Training AI flow
Golden data, calibration, silver training, and fine-tune activation
Checking status
Use this flow in order.
Create the latest dataset, calibrate scoring, generate the silver train/eval set, start fine-tuning, then activate the completed model for future audits only. Your LLM Settings audit model is separate from the fine-tune base model because not every audit model supports fine-tuning.
Current audit model
--
1
Dataset
Create / update golden dataset
Pulls new audited Mongo rows into the versioned dataset without rewriting existing lead scores.
2
Calibration
Calibrate AI scoring
Updates probability, band thresholds, and call-score calibration for future audits.
3
Training set
Generate silver train/eval set
Builds de-identified examples from the latest dataset and tags them with the selected audit and fine-tune models.
4
Fine-tune
Start fine-tune job
Uploads train/eval files and starts the OpenAI fine-tuning job using the supported fine-tune base selected above.
5
Activation
Refresh and activate
Refresh job status while training runs, then activate the fine-tuned model for future audits.
No golden dataset has been generated yet.
Silver training has not been prepared yet.
Account security
Change password
Update your password anytime. Temporary passwords must be changed before using the app.
Step 1
SMTP connection
Gmail-only configuration is enabled here.
Use your Gmail address and an app password. This same setup will later be used for automated dashboard emails with embedded snapshots, PDF attachments, and Excel reports.
Step 2
Sender identity
Test before we automate anything.
Send a simple email first so we know the credentials, sender identity, and SMTP security mode are working correctly.
Step 3
Send HeatMap dashboard email
Embedded dashboard email.
The email body is generated from the latest HeatMap data at send time, with AI insights on the right. It also attaches the latest complete lead report Excel and a HeatMap dashboard PDF.
Step 1
Create backup export
Use this before moving the app to a server.
The backup exports MongoDB app collections: leads, audits, follow-ups, users, golden datasets, calibration settings, training history, and dashboard cache settings.
Step 2
Import and restore
Replace mode is powerful.
Use replace mode for a fresh server migration. It clears app collections before restoring the uploaded backup. Merge mode is safer for adding missing records.
History
Backup log
| Date & time | File | Type | Size |
|---|---|---|---|
| No backup history yet. | |||