Day 1
Get started
Scout ingests your past policies and auto-drafts a baseline. You review key intents and edge cases.
Agent Canvas
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.
AAcme 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 memoryHigh
Last locked byInternal system
Dev mode
Banking AI support agentMulti-modaldraftupdated a min agoDeploy 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_confirmedproceed toIdentity Verification
Customer refuses@info_not_providedproceed toEscalation
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_checkproceed toCompliance Checks
Upload fails: retry up to 2 times@info_not_providedif still fails, proceed toEscalation
Scenario 3: Compliance Checks
Say
“We’ll now verify your details against our compliance system. This will only take a moment.”
PassacceptedAccount Confirmation
Fail (watchlist or mismatch):triggercompliance_flagproceed toEscalation
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 receiptmarkaccount_onboardedproceed toClosing
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
ChatVoiceMulti-modal
Add training documentsAttach files to give your agent business context
Drag files here or click to browse
CancelCreate agent
Built-in AI Support
Easy updates
If you can think it, Scout can do it.
Share simple chat instructions with
your built-in AI developer as you
design your ideal agent.
ScoutEnabled
HistoryHello! I’m Scout, your AI developer. How can I help you today?
Scout• 1 min ago
Can you add these APIs:
https://docs.google.com/document
Great. I’ll integrate those APIs and guide you through the setup.
Scout• Just now
Type your message...
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 agentMulti-modal
draftupdated a min agoDeploy agent
Fetch User ProfileLoads a user’s profile from your API and stores name + tier for later steps.
CodeInput SchemaCallable functions:APIsGlobal Variables
1# Loads a user's profile and updates global store
2store_updates = {}
3assistant_message = ""
4
5try:
6 # pretend API result
7 result = {"success": True, "name": "Alex", "tier": "Pro"}
8
9 if not result.get("success"):
10 assistant_message = "Could not fetch profile. Check the user ID."
11 return {"store_updates": store_updates, "assistant_message": assistant_message}
12
13 store_updates["currentUserName"] = result["name"]
14 store_updates["currentUserTier"] = result["tier"]
15 assistant_message = f"Profile loaded for {result['name']}."
16
17except Exception as e:
18 assistant_message = f"Unexpected error. {e}"
19
20return {"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 Rate99%Based on 1,250 simulations
Simulations passed1,240All conditions met
Simulations failed10Failure conditions met
Test caseStatusPass rate
Response Accuracy TestPassed100%
User Interaction SimulationPassed100%
AI Agent Error Handling Test CaseDeviated80%
HOW IT WORKS
Lightning-quick deployment
Day 6
Build & Validate
Modular code + APIs deliver a review-ready version. Run manual QA; Scout refines revisions into a release-ready policy.
Day 12
Full launch
Test-launch to 1% of call traffic, iterate with logs + feedback, then scale to 100% based on KPI insights.