See It In Action

Real Demos.
Real Data.

These aren't mockups. Every scenario below runs against real databases with production data. Watch the agent reason, act, and deliver results.

Recommended Demo Order by Audience

AudienceDemosTime
ExecutivesOne-Call Workflow → Execution Traces5 min
DevelopersOne-Call Workflow → Multi-Step Workflow → Traces → Agent Builder12 min
Full WalkthroughAll 6 demos in order20 min
⚡ FLAGSHIP2 min

One-Call Workflow Creation

The breakthrough: a user says "Build me an onboarding workflow" and the AI creates it in one API call. ~150 tokens, 2 seconds, $0.001.

The Prompt:

"Create a client onboarding workflow with intake, document collection, verification, and activation"

1
calls
~150
tokens
2s
time
$0.001
cost
demo-one-call-workflow
# What happens under the hood:
1.Agent picks activity IDs from its catalog
2.Calls workflow_builder.create_from_steps with compact JSON
3.Server validates ordering: Start first, Completion last
4.Creates template, levels, and states in a single transaction
5.Returns: "Created workflow #1045 — 5 levels, 5 states"
✅ Total: 1 tool call(s) · ~150 tokens · 2s · $0.001
Detailed API3 min

Complex Multi-Step Workflow

Watch the agent build a complex workflow through 12+ individual API calls. Shows the full range of the Workflow Builder actions.

The Prompt:

"Build an invoice processing workflow with three stages: Receipt, Review, and Approval. Each stage needs Start, Processing, and Completion states."

12+
calls
600-1200
tokens
18s
time
$0.007
cost
demo-multi-step-workflow
# What happens under the hood:
1.workflow_builder.create_template → templateId: 2048
2.workflow_builder.add_level × 3 (Receipt, Review, Approval)
3.workflow_builder.add_state × 9 (Start, Processing, Completion per level)
4.workflow_builder.add_transition × 6+ (connecting states across levels)
5.Agent summarizes: "Created workflow #2048 with 3 levels, 9 states"
✅ Total: 12+ tool call(s) · 600-1200 tokens · 18s · $0.007
Data Analysis3 min

Data Pipeline Analysis

The agent investigates real pipeline data across multiple sources — analyzing errors, verifying records, and creating tasks for resolution.

The Prompt:

"Give me an overview of our processing pipeline — how many records are in each status?"

Follow-up prompts:
"Drill into the error records. What's causing failures?"
"For validation errors, create tasks for manual review."
5+
calls
~400
tokens
10s
time
$0.005
cost
demo-pipeline-analysis
# What happens under the hood:
1.pipeline.stats → Pipeline overview with status counts
2.pipeline.search(status=error) → List of error records
3.record.details(recordId) → Specific error codes per record
4.record.analyze → Root cause identification
5.task.create → Follow-up tasks for manual review
✅ Total: 5+ tool call(s) · ~400 tokens · 10s · $0.005
Full Workflow5 min

End-to-End Issue Resolution

Complete automated workflow: find errors → analyze root cause → verify source data → auto-resolve what’s possible → escalate the rest.

The Prompt:

"Find today's batch with the most errors, analyze the top 3, verify the source data, auto-resolve what's possible, and escalate the rest."

7+
calls
~800
tokens
15s
time
$0.010
cost
demo-e2e-resolution
# What happens under the hood:
1.batch.list(hasErrors=true) → Identifies batch #567
2.batch.errors(567) → 12 error records found
3.record.analyze × 3 → Root cause analysis on top errors
4.source.verify → Cross-references against source data
5.record.resolve → Auto-resolves records with known fixes
6.task.create × 2 → Escalates remaining issues for manual review
✅ Total: 7+ tool call(s) · ~800 tokens · 15s · $0.010
Observability1 min

Execution Trace Explorer

Every agent action is traced with full detail — tool name, arguments, results, latency, token count, and cost. Enterprise-grade transparency.

The Prompt:

Open the traces dashboard to browse execution history.

N/A
calls
N/A
tokens
Instant
time
Free
cost
demo-execution-traces
# What happens under the hood:
1.Browse execution history with agent name, duration, tokens, cost
2.Click into a trace to see individual tool calls
3.View input arguments and output results for each call
4.Analyze cost by model, by agent, over time
5.Full compliance audit trail — who, what, when, result
Configuration2 min

Agent Builder UI

See how agents are configured — system prompts, allowed actions, model settings, guardrails, and full version history.

The Prompt:

Open the Agent Builder to view and configure agents.

N/A
calls
N/A
tokens
Instant
time
Free
cost
demo-agent-builder
# What happens under the hood:
1.View all agents with status, model, and action count
2.Inspect system prompts with embedded activity catalogs
3.Configure allowed/denied actions with wildcard support
4.Set model parameters: maxToolCalls, maxTokens, temperature
5.View version timeline — every config change tracked

Want to See This Live?

We'll run these demos against real data in a 30-minute call. Bring your own workflow — we'll show you what's possible.

Schedule a Live DemoExplore All Capabilities