Scenarios

Four ways to build
every test scenario

Visual workflow builder, call script upload, dataset import, or paste your SOP and let AI generate the graph. Template variables turn each row into a unique test case - with live validation so nothing breaks at runtime.

Start for Free Book a Demo
Scenario: Insurance Claim Flow Graph 7 nodes · 247 rows
Nodes
Conversation
End Call
Transfer Call
Variables
{{persona}} {{situation}} {{outcome}}
Start Greeting
Conversation Verify Identity
Branch Claim Type?
Conversation New Claim
Conversation Check Status
Transfer To Agent
End Call Resolve
Node: Verify Identity
Type
Conversation
Prompt
Ask the caller for their name and policy number. Persona: {{persona}}
Transitions
→ Branch: Claim Type?
→ End: Invalid ID
Conditions
If verified → Branch
If failed 3x → End Call
Method: SOP → Auto-generated Rows: 247 test cases Variables: 3 valid
Agent: Insurance Claims v2
Core Features

Build scenarios any way
your team works

Workflow Builder
drag & drop
SCENARIO GRAPH - INSURANCE CLAIM FLOW NODE PALETTE Converse Branch End Call Transfer Auto-gen drag → Start Greeting Verify ID Claim Type new New Claim existing Check Status End Call Transfer LEGEND Conversation Branch End Call Transfer 7 nodes · 6 connections · drag-and-drop canvas
SOP Import
parsing
SOP → SCENARIO GRAPH CONVERSION insurance_claim_sop.pdf TXT DOCX PDF 1. Greet customer warmly 2. Verify identity (name, DOB) 3. Determine claim type 4. If new: collect details 5. If existing: pull records 6. Provide resolution steps 7. Confirm satisfaction 8. Close & summarize call Parsing... 12 turns detected parse Generated Graph Auto-detected structure Start Greet Verify ID Claim Type? New Claim Check Resolve End SOP auto-parsed → 8 steps → 7 nodes · 2 branches
Template Editor
3 valid · 1 missing
PROMPT TEMPLATE - LIVE VALIDATION 1 You are a customer service agent. 2 The caller is {{persona}} who needs help. 3 4 Their situation: {{situation}} 5 Expected outcome: {{outcome}} 6 7 Risk level: {{risk_level}} 8 Handle with care and empathy. 9 // End of prompt template ✓ 3 valid · ! 1 missing risk_level not found in dataset DATASET PREVIEW - MATCHED COLUMNS persona situation outcome Elderly, confused Auto claim, rear-end File new claim Angry, impatient Delayed reimbursement Escalate Tech-savvy, direct Policy renewal question Provide info 3 valid · 1 missing · live variable resolution
Dataset Generator
generating
DATASET - IMPORT & SYNTHETIC GENERATION Upload CSV Sample Data AI Generate PROMPT Generate 50 insurance customer profiles with varied ✦ demographics and claim types Generating... 32/50 rows 64% GENERATED DATA # persona situation outcome 01 Maria, 67, retired Water damage, home flood File new claim 02 James, 34, engineer Fender bender, highway Check status 03 Aisha, 28, student Laptop theft, campus File new claim 04 Robert, 55, manager Roof damage, storm Escalate 05 Generating... DISTRIBUTION Age Range 20-70 Claim Types 4 types Outcomes new chk esc 50 profiles · 4 claim types · varied demographics · AI synthetic

Drag conversation, end-call, and transfer nodes onto a canvas and connect them into branching conversation flows. Configure each node with prompts, messages, and conditional logic - or enable auto-generate and let AI build the graph from a description. No competitor offers a purpose-built visual graph editor for test scenarios.

Build your first scenario

Upload a call script (TXT, DOCX, or PDF) and the system parses Customer/Agent dialogue into graph nodes and scenario rows automatically. Or paste a numbered Standard Operating Procedure - AI converts each step into a conversation graph with personas, situations, and outcomes. No other tool converts documents to test scenarios.

Upload a script

Reference scenario columns in your simulator prompt with {{persona}}, {{situation}}, {{outcome}}, or any custom variable. The editor highlights variables green when the column exists and red when it's missing - so you catch data mismatches before you run. Each row in the scenario table becomes a unique test case.

See template syntax

Upload a CSV or Excel file, use a sample dataset, or generate synthetic data from a description. Synthetic generation creates realistic customer profiles - demographics, objection types, risk profiles - with balanced distributions, not random noise. Add 1–100 rows per request, or let AI generate rows from a prompt.

Generate test data
Use Cases

From SOP to test suite
in minutes, not weeks

1. Greet customer 2. Verify identity 3. Route to dept 1 2 3 SOP steps → executable test graph

Convert SOPs to automated tests

Paste your call center Standard Operating Procedure as numbered steps. AI converts it into a conversation graph and generates scenario rows - with personas, situations, and outcomes. Prove compliance without writing a single test case.

SOP Auto-generate Compliance
PDF script.pdf PARSED DIALOGUE Customer: "I need to cancel..." Agent: "Let me help you with..." Customer: "What about refund?" PDF DOCX TXT CSV Upload scripts → auto-parse dialogue turns

Upload existing call scripts

Upload TXT, DOCX, or PDF call scripts with Customer/Agent dialogue. The system parses turns into graph nodes and creates test rows automatically - no reformatting needed. Use [EXPECTED: ...] tags to define pass/fail criteria inline.

TXT DOCX PDF
PERSONA SCENARIO INTENT Angry Refund request cancel Elderly Password reset support Confused Billing inquiry billing Impatient Order tracking track ... ... ... balanced 42/50 rows

Generate synthetic test data

Describe the customer profiles you need - demographics, objection types, risk levels - and generate 1–100 realistic rows per request. Synthetic data follows balanced distributions, not random noise, so your test coverage matches real-world traffic.

Synthetic CSV Excel
Start ? Happy path resolve issue End Error path API failure Retry Escalation transfer agent Hold Branch on condition → test every path

Design branching conversation flows

Use the visual graph builder to drag conversation, end-call, and transfer nodes onto the canvas. Connect them with conditional edges to model happy paths, error paths, and escalation flows - then generate scenario rows from the graph.

Graph Visual Builder
PROMPT TEMPLATE You are a {{persona}} agent. Handle this {{situation}} scenario. Use tool: {{missing}} ← undefined 2 resolved ! 1 missing variable

Parameterize with template variables

Reference any column in your scenario table using {{persona}}, {{situation}}, {{outcome}}, or custom variables. The prompt editor validates variables live - green for valid, red for missing - so every test case gets the right data.

Variables Data-driven
AGENT DEF │ Customer support bot, refund policy tools: lookup_order, process_refund, transfer_agent Scn 1 Refund for damaged item 3 turns Scn 2 Order status for lost package 4 turns Scn 3 Escalation: angry about wait time 5 turns Agent definition → auto-generated test scenarios

Scale coverage with AI auto-generation

Enable auto-generate on any scenario type and the system creates conversation paths, personas, and outcomes from your agent definition and description. Go from zero test cases to hundreds without writing a single row manually.

Auto-generate AI
How It Works

Choose a method, configure,
and run

01

Choose your creation method

Build scenarios with the visual workflow builder, upload a call script (TXT/DOCX/PDF), import from a dataset (CSV/Excel), or paste your SOP and let AI generate the graph. Each method produces the same output: a scenario with graph, table, and simulator prompt.

Choose Creation Method
SELECT A METHOD TO CREATE YOUR SCENARIO Visual Builder Drag & drop nodes to build flows TXT DOCX PDF Upload Script Import from text documents A B C Import Dataset CSV or JSON data import ✨ AI-Powered SOP → AI Paste your SOP, AI builds scenarios RECOMMENDED
02

Configure graph, variables, and rows

Edit nodes and edges in the graph, set template variables in the simulator prompt with {{column}} syntax, and add test case rows - from existing datasets, AI generation, or manual entry. The prompt editor validates every variable live.

Configure Graph, Variables & Rows
Scenario Graph Start Greet Verify Reject Approve End 5 nodes · 6 edges Prompt Template You are a support agent. The caller is a {{persona}} dealing with: {{situation}} Expected result: {{outcome}} Respond helpfully and follow the SOP steps. Variables detected: persona situation outcome 3 valid Scenario Data persona situation outcome Angry Refund Resolved Elderly Billing Escalate Confused Setup Guided Techie API err Fixed VIP Upgrade Upsold 247 rows ✨ AI Generate or drag CSV to upload
03

Run simulations and review

Select your scenario in a test run and execute calls or chats at scale. Every row becomes a unique test case. Review per-call results - transcripts, audio, CSAT, compliance - and iterate on the scenario until coverage is complete.

Run Simulations & Review
Complete
SIMULATION PROGRESS 247/247 complete RESULTS SUMMARY Pass Rate 94% 232 / 247 Avg CSAT 4.2 / 5.0 Failures 15 scenarios SCENARIO RESULTS # Scenario Persona CSAT Status 1 Refund request - standard Angry Customer 4.5 PASS 2 Billing dispute - complex Elderly Caller 2.1 FAIL 3 Account setup - guided Confused User 4.8 PASS Drill into failures →

Powering teams from
prototype to production

From ambitious startups to global enterprises, teams trust Future AGI to ship AI agents confidently.