
Acme Co
Acme Co
Search…
Search…
⌘
⌘
K
K
Policies
Policies
Data sources
Data sources
Personalization
Personalization
Evaluation
Evaluation
Advanced
Advanced
Scenarios
Scenarios
New account onboarding
New account onboarding
Loan application support
Loan application support
Fraud alert handling
Fraud alert handling
Credit card replacement
Credit card replacement
Supporting docs
Supporting docs
Compliance guidelines.pdf
Compliance guidelines.pdf
ID verification.csv
ID verification.csv
Escalation flowchart.png
Escalation flowchart.png
Brand
Brand
Rules
Rules
Remaining memory
Remaining memory
High
High
Last locked by
Last locked by
Internal system
Internal system
Dev mode
Dev mode

Banking AI support agent
Banking AI support agent
Multi-modal
Multi-modal
draft
draft
updated a min ago
updated a min ago
Deploy agent
Deploy agent
Scenario: New account onboarding
Scenario: New account onboarding
Scenario 1: “Account Creation” Inquiry
Scenario 1: “Account Creation” Inquiry
Say
Say
“Welcome to Acme Bank! We’re excited to get you started with your new account. May I confirm your full name and date of birth to begin?”
“Welcome to Acme Bank! We’re excited to get you started with your new account. May I confirm your full name and date of birth to begin?”
Customer provides info
Customer provides info
@info_confirmed
@info_confirmed
proceed to
proceed to
Identity Verification
Identity Verification
Customer refuses
Customer refuses
@info_not_provided
@info_not_provided
proceed to
proceed to
Escalation
Escalation
Scenario 2: Identity Verification
Scenario 2: Identity Verification
Say
Say
“Thank you. Please upload a photo of your government-issued ID, such as a passport or driver’s license.”
“Thank you. Please upload a photo of your government-issued ID, such as a passport or driver’s license.”
Upload is successful:
Upload is successful:
verify_identity_check
verify_identity_check
proceed to
proceed to
Compliance Checks
Compliance Checks
Upload fails: retry up to 2 times
Upload fails: retry up to 2 times
@info_not_provided
@info_not_provided
if still fails, proceed to
if still fails, proceed to
Escalation
Escalation
Scenario 3: Compliance Checks
Scenario 3: Compliance Checks
Say
Say
“We’ll now verify your details against our compliance system. This will only take a moment.”
“We’ll now verify your details against our compliance system. This will only take a moment.”
Pass
Pass
accepted
accepted
Account Confirmation
Account Confirmation
Fail (watchlist or mismatch):
Fail (watchlist or mismatch):
trigger
trigger
compliance_flag
compliance_flag
proceed to
proceed to
Escalation
Escalation
Scenario 4: Account Confirmation
Scenario 4: Account Confirmation
Say
Say
“Great news! Your account has been successfully created. We’ve sent your login details to your registered email address. Please check your inbox and follow the instructions to sign in.”
“Great news! Your account has been successfully created. We’ve sent your login details to your registered email address. Please check your inbox and follow the instructions to sign in.”
Customer confirms receipt
Customer confirms receipt
mark
mark
account_onboarded
account_onboarded
proceed to
proceed to
Closing
Closing
No confirmation
No confirmation
resend email
resend email
Still no confirmation
Still no confirmation
idle
idle
proceed to
proceed to
Escalation
Escalation
Step 5: Escalation
Step 5: Escalation
Say
Say
“I’ll need to connect you with a specialist to complete your onboarding. Please hold while I transfer you to our Tier 2 team.”
“I’ll need to connect you with a specialist to complete your onboarding. Please hold while I transfer you to our Tier 2 team.”
If
If
Action
Action
trigger
trigger
escalation_ticket
escalation_ticket
Step 6: Closing
Step 6: Closing
Say
Say
“Thank you for choosing Acme Bank. Your account is now ready to use, and our support team is here if you need anything else. Have a wonderful day!”
“Thank you for choosing Acme Bank. Your account is now ready to use, and our support team is here if you need anything else. Have a wonderful day!”
Acme Co
Search…
⌘
K
Policies
Data sources
Personalization
Evaluation
Advanced
Scenarios
New account onboarding
Loan application support
Fraud alert handling
Credit card replacement
Supporting docs
Compliance guidelines.pdf
ID verification.csv
Escalation flowchart.png
Brand
Rules
Remaining memory
High
Last locked by
Internal system
Dev mode

Banking AI support agent
Multi-modal
draft
updated a min ago
Deploy agent
Scenario: New account onboarding
Scenario 1: “Account Creation” Inquiry
Say
“Welcome to Acme Bank! We’re excited to get you started with your new account. May I confirm your full name and date of birth to begin?”
Customer provides info
@info_confirmed
proceed to
Identity Verification
Customer refuses
@info_not_provided
proceed to
Escalation
Scenario 2: Identity Verification
Say
“Thank you. Please upload a photo of your government-issued ID, such as a passport or driver’s license.”
Upload is successful:
verify_identity_check
proceed to
Compliance Checks
Upload fails: retry up to 2 times
@info_not_provided
if still fails, proceed to
Escalation
Scenario 3: Compliance Checks
Say
“We’ll now verify your details against our compliance system. This will only take a moment.”
Pass
accepted
Account Confirmation
Fail (watchlist or mismatch):
trigger
compliance_flag
proceed to
Escalation
Scenario 4: Account Confirmation
Say
“Great news! Your account has been successfully created. We’ve sent your login details to your registered email address. Please check your inbox and follow the instructions to sign in.”
Customer confirms receipt
mark
account_onboarded
proceed to
Closing
No confirmation
resend email
Still no confirmation
idle
proceed to
Escalation
Step 5: Escalation
Say
“I’ll need to connect you with a specialist to complete your onboarding. Please hold while I transfer you to our Tier 2 team.”
If
Action
trigger
escalation_ticket
Step 6: Closing
Say
“Thank you for choosing Acme Bank. Your account is now ready to use, and our support team is here if you need anything else. Have a wonderful day!”
Solve your biggest support problems without breaking a sweat
Solve your biggest support problems without breaking a sweat
Solve your
biggest support problems without breaking a sweat
Put the power of 50,000 AI agents to work for your business. Agent Canvas makes it easy to set up, customize, and iterate on your ideal agent so you can solve the biggest support problems your customers face.
Put the power of 50,000 AI agents to work for your business. Agent Canvas makes it easy to set up, customize, and iterate on your ideal agent so you can solve the biggest support problems your customers face.
Built for complexity
Designed to handle your biggest support problems
Built for speed
Implement complex changes instantly
Built for scale
Navigate millions of customer scenarios with high precision
Get a new agent up and running in no time
Get a new agent up and running in no time
AGENT BUILDER
Effortless setup
Design new voice, chat, or multi-modal agents with as much or as little specificity as you need. Get started with transcripts, live recordings, or SOPs.
Create new agent

Chat
Voice
Multi-modal
Add training documents
Attach files to give your agent business context
Drag files here or click to browse
Cancel
Create agent
AGENT BUILDER
Effortless setup
Design new voice, chat, or multi-modal agents with as much or as little specificity as you need. Get started with transcripts, live recordings, or SOPs.
Create new agent

Chat
Voice
Multi-modal
Add training documents
Attach files to give your agent business context
Drag files here or click to browse
Cancel
Create agent
AGENT BUILDER
Effortless setup
Design new voice, chat, or multi-modal agents with as much or as little specificity as you need. Get started with transcripts, live recordings, or SOPs.
Create new agent

Chat
Voice
Multi-modal
Add training documents
Attach files to give your agent business context
Drag files here or click to browse
Cancel
Create agent
AGENT BUILDER
Effortless setup
Design new voice, chat, or multi-modal agents with as much or as little specificity as you need. Get started with transcripts, live recordings, or SOPs.
Create new agent

Chat
Voice
Multi-modal
Add training documents
Attach files to give your agent business context
Drag files here or click to browse
Cancel
Create agent
BUILT-IN AI SUPPORT
Easy updates
If you can think it, Atlas can do it. Share simple chat instructions with your built-in AI developer as you design your ideal agent.
Atlas
Enabled
History
Hello! I’m Atlas, your AI developer. How can I help you today?
Atlas
• 1 min ago
Can you add these APIs:
https://docs.google.com/document
Just now
Great. I’ll integrate those APIs and guide you through the setup.
Atlas
• Just now
Type your message...
BUILT-IN AI SUPPORT
Easy updates
If you can think it, Atlas can do it. Share simple chat instructions with your built-in AI developer as you design your ideal agent.
Atlas
Enabled
History
Hello! I’m Atlas, your AI developer. How can I help you today?
Atlas
• 1 min ago
Can you add these APIs:
https://docs.google.com/document
Just now
Great. I’ll integrate those APIs and guide you through the setup.
Atlas
• Just now
Type your message...
BUILT-IN AI SUPPORT
Easy updates
If you can think it, Atlas can do it. Share simple chat instructions with your built-in AI developer as you design your ideal agent.
Atlas
Enabled
History
Hello! I’m Atlas, your AI developer. How can I help you today?
Atlas
• 1 min ago
Can you add these APIs:
https://docs.google.com/document
Just now
Great. I’ll integrate those APIs and guide you through the setup.
Atlas
• Just now
Type your message...
BUILT-IN AI SUPPORT
Easy updates
If you can think it, Atlas can do it. Share simple chat instructions with your built-in AI developer as you design your ideal agent.
Atlas
Enabled
History
Hello! I’m Atlas, your AI developer. How can I help you today?
Atlas
• 1 min ago
Can you add these APIs:
https://docs.google.com/document
Just now
Great. I’ll integrate those APIs and guide you through the setup.
Atlas
• Just now
Type your message...
Control exactly how your agents show up
Control exactly how your agents show up
POLICY WRITING
Precise customization
Define support rules and workflows with specificity so your agent feels like an extension of your business.

Banking AI support agent
Multi-modal
Draft
Updated a min ago
Deploy agent
Fetch User Profile
Loads a user’s profile from your API and stores name + tier for later steps.
Code
Input Schema
Callable functions:
APIs
Global Variables
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
# Loads a user's profile and updates global store
store_updates = {}
assistant_message = ""
try:
# API result
result = await get_user_profile(user_id)
if not result.get("success"):
assistant_message = "Could not fetch profile. Check the user ID."
return {"store_updates": store_updates, "assistant_message": assistant_message}
store_updates["currentUserName"] = result["name"]
store_updates["currentUserTier"] = result["tier"]
assistant_message = f"Profile loaded for {result['name']}."
except Exception as e:
assistant_message = f"Unexpected error. {e}"
return {"store_updates": store_updates, "assistant_message": assistant_message}
POLICY WRITING
Precise customization
Define support rules and workflows with specificity so your agent feels like an extension of your business.

Banking AI support agent
Multi-modal
Draft
Updated a min ago
Deploy agent
Fetch User Profile
Loads a user’s profile from your API and stores name + tier for later steps.
Code
Input Schema
Callable functions:
APIs
Global Variables
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
# Loads a user's profile and updates global store
store_updates = {}
assistant_message = ""
try:
# API result
result = await get_user_profile(user_id)
if not result.get("success"):
assistant_message = "Could not fetch profile. Check the user ID."
return {"store_updates": store_updates, "assistant_message": assistant_message}
store_updates["currentUserName"] = result["name"]
store_updates["currentUserTier"] = result["tier"]
assistant_message = f"Profile loaded for {result['name']}."
except Exception as e:
assistant_message = f"Unexpected error. {e}"
return {"store_updates": store_updates, "assistant_message": assistant_message}
POLICY WRITING
Precise customization
Define support rules and workflows with specificity so your agent feels like an extension of your business.

Banking AI support agent
Multi-modal
Draft
Updated a min ago
Deploy agent
Fetch User Profile
Loads a user’s profile from your API and stores name + tier for later steps.
Code
Input Schema
Callable functions:
APIs
Global Variables
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
# Loads a user's profile and updates global store
store_updates = {}
assistant_message = ""
try:
# API result
result = await get_user_profile(user_id)
if not result.get("success"):
assistant_message = "Could not fetch profile. Check the user ID."
return {"store_updates": store_updates, "assistant_message": assistant_message}
store_updates["currentUserName"] = result["name"]
store_updates["currentUserTier"] = result["tier"]
assistant_message = f"Profile loaded for {result['name']}."
except Exception as e:
assistant_message = f"Unexpected error. {e}"
return {"store_updates": store_updates, "assistant_message": assistant_message}
POLICY WRITING
Precise customization
Define support rules and workflows with specificity so your agent feels like an extension of your business.

Banking AI support agent
Multi-modal

Draft
Updated a min ago
Deploy agent
Fetch User Profile
Loads a user’s profile from your API and stores name + tier for later steps.
Code
Input Schema
Callable functions:
APIs
Global Variables
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
# Loads a user's profile and updates global store
store_updates = {}
assistant_message = ""
try:
# API result
result = await get_user_profile(user_id)
if not result.get("success"):
assistant_message = "Could not fetch profile. Check the user ID."
return {"store_updates": store_updates, "assistant_message": assistant_message}
store_updates["currentUserName"] = result["name"]
store_updates["currentUserTier"] = result["tier"]
assistant_message = f"Profile loaded for {result['name']}."
except Exception as e:
assistant_message = f"Unexpected error. {e}"
return {"store_updates": store_updates, "assistant_message": assistant_message}
SIMULATIONS
Refine as you go
It’s easy to run test cases and iterate agent behavior for support that keeps getting better.

Banking AI support agent
an hour ago
Test results
Pass Rate
99%
Based on 1,250 simulations
Simulations passed
1,240
All conditions met
Simulations failed
10
Failure conditions met
Test case
Status
Pass rate
SIMULATIONS
Refine as you go
It’s easy to run test cases and iterate agent behavior for support that keeps getting better.

Banking AI support agent
an hour ago
Test results
Pass Rate
99%
Based on 1,250 simulations
Simulations passed
1,240
All conditions met
Simulations failed
10
Failure conditions met
Test case
Status
Pass rate
SIMULATIONS
Refine as you go
It’s easy to run test cases and iterate agent behavior for support that keeps getting better.

Banking AI support agent
an hour ago
Test results
Pass Rate
99%
Based on 1,250 simulations
Simulations passed
1,240
All conditions met
Simulations failed
10
Failure conditions met
Test case
Status
Pass rate
SIMULATIONS
Refine as you go
It’s easy to run test cases and iterate agent behavior for support that keeps getting better.

Banking AI support agent
an hour ago
Test results
Pass Rate
99%
Based on 1,250 simulations
Simulations passed
1,240
All conditions met
Simulations failed
10
Failure conditions met
Test case
Status
Pass rate
HOW IT WORKS
Lightning-quick deployment
Lightning-quick deployment
Day 1
Get started
Atlas ingests your past policies and auto-drafts a baseline. You review key intents and edge cases.
Day 6
Build & Validate
Modular code + APIs deliver a review-ready version. Run manual QA; Atlas refines revisions into a release-ready policy.
Day 12
Full launch
Test-launch to a 1% of call traffic, iterate with logs + feedback, then scale to 100% based on KPI insights.
Day 1
Get started
Atlas ingests your past policies and auto-drafts a baseline. You review key intents and edge cases.
Day 6
Build & Validate
Modular code + APIs deliver a review-ready version. Run manual QA; Atlas refines revisions into a release-ready policy.
Day 12
Full launch
Test-launch to a 1% of call traffic, iterate with logs + feedback, then scale to 100% based on KPI insights.
Day 1
Get started
Atlas ingests your past policies and auto-drafts a baseline. You review key intents and edge cases.
Day 6
Build & Validate
Modular code + APIs deliver a review-ready version. Run manual QA; Atlas refines revisions into a release-ready policy.
Day 12
Full launch
Test-launch to a 1% of call traffic, iterate with logs + feedback, then scale to 100% based on KPI insights.
Day 1
Get started
Atlas ingests your past policies and auto-drafts a baseline. You review key intents and edge cases.
Day 6
Build & Validate
Modular code + APIs deliver a review-ready version. Run manual QA; Atlas refines revisions into a release-ready policy.
Day 12
Full launch
Test-launch to a 1% of call traffic, iterate with logs + feedback, then scale to 100% based on KPI insights.
GET A PERSONALIZED DEMO
Ready to see the Giga AI agent in action?
Ready to see the Giga AI agent in action?
Start scaling your customer support with powerful AI.
Giga
Giga
Giga


