AI Phone System (Sam)¶

Overview¶

The AI Phone System is Scott Recycling's automated call handling solution powered by the AI assistant "Sam." Built on the scott_ai_csr module, Sam answers incoming phone calls via Asterisk/Sangoma integration, identifies callers by matching phone numbers against contacts, detects what the caller needs, extracts structured data from the conversation, creates pickup requests, and can transfer calls to human agents when needed.

All phone system activity is managed under CSR Phone in the main menu.

Last Updated: February 2026 Target Audience: CSR Managers, Admins

The CSR Phone menu contains four sections:

Menu Item	What It Shows
Active Calls	Calls currently in progress (ringing or connected)
Phone Calls	Complete history of all calls handled by the system
Pickup Requests	Pickup requests created from phone conversations
Pickup Schedule	Calendar view of scheduled pickups from phone requests

Phone Calls¶

List View¶

The Phone Calls list shows all calls with the following columns:

Column	Description
Call ID	Unique identifier for the call
Phone Number	Caller's phone number
Caller	Name provided by the caller or identified from caller ID
Contact	Matched Odoo contact record (if found)
Intent	What the caller was trying to do (detected by AI)
Status	Current call status (Ringing, Active, Completed, Transferred, Abandoned)
Duration	Total call length
Outcome	Final result of the call

Click any row to open the full call detail form.

Phone Calls list view Phone Calls list — all calls handled by the AI phone system with status, intent, and duration

Call Detail Form¶

The call detail form shows everything about a single call:

Status Workflow¶

Calls progress through these statuses:

Ringing --> Active --> Completed / Transferred / Abandoned

Status	Meaning
Ringing	Incoming call detected, Sam is connecting
Active	Call is in progress, Sam is speaking with the caller
Completed	Call ended normally after Sam resolved the caller's request
Transferred	Call was handed off to a human agent
Abandoned	Caller hung up before the call was resolved

Call Information Fields¶

Field	Description
Phone Number	The caller's number
Contact	Linked Odoo contact (auto-matched by phone number)
Detected Intent	The AI-classified intent of the call (see Intent Detection below)
Conversation Transcript	Full text transcript of the conversation between Sam and the caller
Extracted Slots	Structured JSON data extracted from the conversation (e.g., address, material types, preferred date)
Recording	Audio recording of the call (if available)
Duration	Call length in minutes and seconds
Outcome	Final call outcome (see Call Outcomes below)

Shadow Monitoring Fields¶

Field	Description
Shadow User	Human agent currently monitoring this call
Shadow Active	Whether shadow monitoring is currently enabled
Shadow Notes	Notes added by the monitoring agent

Call Outcomes¶

Every completed call receives one of the following outcomes:

Outcome	Meaning
Resolved	Sam fully handled the caller's request without human intervention
Pickup Scheduled	A pickup request was created from the conversation
Transferred to Human	The call was handed off to a human agent
Callback Requested	The caller asked to be called back at a later time
Abandoned	The caller disconnected before the request was resolved
Failed	A system error prevented the call from being handled

Intent Detection¶

Sam automatically classifies each call into one of the following intents based on the conversation:

Intent	Description
schedule_pickup	Caller wants to schedule an e-waste pickup
check_materials	Caller is asking what materials Scott Recycling accepts
get_directions	Caller wants directions to a Scott Recycling facility
request_certificate	Caller needs a recycling certificate or documentation
check_status	Caller is checking the status of an existing pickup or service
speak_to_human	Caller explicitly requests a human agent
general_question	Catch-all for other inquiries

Intent detection happens in real time during the conversation. Sam uses the detected intent to guide the conversation flow, ask relevant follow-up questions, and extract the right data fields.

Pickup Requests¶

When Sam identifies that a caller wants to schedule a pickup, it creates a Pickup Request record with the information extracted from the conversation.

List and Calendar Views¶

Pickup requests can be viewed in list or calendar format. The calendar view shows requests by their preferred pickup date, making it easy to see upcoming scheduling demand.

Pickup Requests from phone calls Pickup Requests — pickup requests created automatically from AI phone conversations

Pickup Request Fields¶

Field	Description
Contact	The matched or newly created contact
Phone Number	Caller's phone number
Address	Pickup location extracted from conversation
Material Types	Types of e-waste the caller described
Preferred Date	When the caller wants the pickup
Preferred Time	Time of day preference (morning, afternoon, etc.)
Data Destruction	Whether the caller requires data destruction services
Certificate Required	Whether a certificate of recycling is needed
Notes	Additional details from the conversation
Status	Current status of the request
Assigned To	CSR or dispatcher responsible for scheduling

Pickup Request Status Workflow¶

New --> Confirmed --> Scheduled --> In Progress --> Completed

Status	Meaning
New	Request just created by Sam, needs human review
Confirmed	CSR has verified the details with the customer
Scheduled	Pickup has been added to the truck schedule
In Progress	Pickup is being executed
Completed	Pickup finished successfully

Transfer Wizard¶

When Sam determines that a call needs human intervention -- or the caller asks to speak with a person -- Sam can transfer the call. Managers can also manually initiate transfers from the call detail form.

The Transfer Wizard includes:

Field	Description
Transfer To	Select a specific user, call queue, or external phone number
Reason	Why the call is being transferred (e.g., complex request, caller preference, escalation)
Notes	Context for the receiving agent about the conversation so far

The receiving agent sees the conversation transcript and extracted data so the caller does not have to repeat information.

Shadow Mode¶

Shadow mode lets a human agent silently monitor a live AI call without the caller hearing them. This is used for quality assurance and training.

Go to CSR Phone > Active Calls
Find the call you want to monitor
Click Shadow to join the call in listen-only mode
Add notes in the Shadow Notes field as the call progresses
Optionally, take over the call by clicking Take Over if intervention is needed

Shadow sessions are logged for review, including which agent monitored and any notes they added.

AI Conversations — full transcripts of Sam's phone interactions with callers

Common Tasks¶

Viewing Active Calls¶

Navigate to CSR Phone > Active Calls
The list shows all calls currently ringing or in progress
Each row shows caller number, matched contact, current status, and duration
Click a row to see the live transcript and extracted data

Reviewing Pickup Requests¶

Navigate to CSR Phone > Pickup Requests
Filter by status to find New requests that need review
Open a request to see the extracted details and conversation transcript
Verify the information, update any fields, and change status to Confirmed
Assign the request to a dispatcher for scheduling

Scheduling from a Pickup Request¶

Open a Confirmed pickup request
Review the preferred date, material types, and location
Use the information to create a pickup through the standard Scheduling a Pickup workflow
Update the pickup request status to Scheduled

Transferring a Call¶

Open an active call from CSR Phone > Active Calls
Click Transfer
Select the transfer target (user, queue, or external number)
Enter the reason and any notes for the receiving agent
Click Confirm Transfer -- the call is handed off immediately

Tips¶

Review new pickup requests every morning

Sam creates pickup requests around the clock. Check CSR Phone > Pickup Requests filtered to New status first thing each morning to process overnight requests promptly.

Use shadow mode for quality assurance

Periodically shadow live calls to evaluate how Sam handles different scenarios. This helps identify conversation patterns that could be improved and ensures customers are getting accurate information.

Check the transcript before calling back

When following up on a pickup request, read the full conversation transcript first. The caller already provided details to Sam -- referencing them shows the customer their time was valued and avoids asking the same questions twice.

Monitor call outcomes weekly

Review the Phone Calls list grouped by Outcome to spot trends. A rising number of transfers or abandoned calls may indicate conversation flows that need tuning.

Questions? Contact your supervisor or refer to the AI & Automation overview.