Platform onboarding · 1 of 8
What best describes you?
We'll shape your experience around how you work — terminology, examples, and metrics will all adapt.
Business leader
I own outcomes — ROI, efficiency, and strategic direction
CX / ops manager
I run customer or employee experience day-to-day
Developer / builder
I build, integrate, and ship AI-powered systems
IT / enterprise architect
I govern infrastructure, security, and compliance
Platform onboarding · 2 of 8
What are you here to solve?
Pick everything that applies. We'll build your live demo around these — so what you see is relevant to your actual problems.
Customer support at scale
Employee self-service
IT helpdesk automation
Sales & lead qualification
Search & knowledge access
Voice / IVR replacement
After-hours coverage
Multilingual support
No configuration happens here — your selections just personalise the demo on the next screen.
Platform onboarding · 3 of 8
Meet an agent already working
No setup. Pre-built for customer support — try it as a customer first, then see exactly what happened behind the scenes.
Lumio Electronics — customer view
Hi! I'm Lumio's support agent. How can I help you today?
Where's my order #LM-4821?
Your order shipped Apr 4 via FedEx — expected Apr 9. Tracking link sent to your email.
Can I change the delivery address?
Let me check if it's been picked up yet — one moment.
Querying order management system...
Agent reasoning
Intent detected
order_status_lookup · 0.97
NLU
Knowledge retrieved
Order #LM-4821 · shipped Apr 4, FedEx
RAG
Policy applied
Share carrier + ETA + tracking link
Policy
Intent detected
address_change · 0.91
NLU
Action triggered
Querying OMS for pickup status
Action
Escalation check
Confidence above threshold — continuing
Guard
The reasoning trace runs in real time alongside every response — showing exactly what was retrieved, which policy applied, and why the agent acted or escalated.
Agent builder · 4 of 8
Name your agent
Think of this as writing a job description — not building a bot. Give it a name and describe what it handles in plain language.
Agent name
What does this agent handle?
Or start from a template
Customer support
Returns, orders, billing, escalation
HR self-service
Leave, payroll, onboarding, policy
IT helpdesk
Tickets, access, troubleshooting
Agent builder · 5 of 8
Teach your agent
Upload anything that explains how your support works. Even one document visibly changes the agent's behaviour — you can re-test immediately after.
Drop files or paste a URL
PDF · DOCX · TXT · or paste a knowledge base URL
Suggested sources
Return policy PDF
FAQ page URL
Product manual
Past chat transcripts
SOP documents
After upload, we'll surface the top intents we found in your content — so you can see your knowledge become agent behaviour immediately.
Agent builder · 6 of 8
Rules & voice
Write your agent's operating principles in plain language. No intent trees or rule engines — just describe what it should and shouldn't do.
Always...
Apologize before denying
Respond in customer's language
Summarise context on escalation
Offer an alternative if declining
+ Add rule
Never...
Approve refunds above $200
Share internal ticket IDs
Discuss competitor pricing
+ Add rule
When stuck, the agent should...
Hand off to a human agent
Summarise the conversation and pass with a draft response.
Flag for async review
Send a holding message and notify your team to follow up.
Define a custom flow
Build a decision tree for specific edge cases.
Brand voice
Friendly
Warm, uses contractions
Formal
Professional, complete sentences
Empathetic
Feelings first, then resolves
Direct
Short, factual, no filler
Live previewauto-updates on tone change
"Hey! Great news — your return has been processed and the refund is on its way. You should see it within 3–5 business days. Anything else I can help with?"
Paste sample responses from your brand guidelines and we'll match the style automatically.
Agent builder · 7 of 8
Coverage & simulation
Based on what you've uploaded, here's what your agent can confidently handle — and where the gaps are before a single real customer sees it.
82%
Simulated resolution rate
↑ vs 61% industry avg
11%
Escalation rate
↓ target <15%
0
Guardrail violations
All policies held
4
Coverage gaps found
Action recommended
Intent coverage by category
Order status
92%
Returns & refunds
85%
Product support
79%
Billing & payments
64%
Account management
51%
Cancellations
18%
Warranty claims
9%
3 failures linked to missing cancellation and warranty policies. Upload those docs to close the gap before going live.
Agent builder · 8 of 8
Deploy your agent
Go live on one channel first — add more at any time. Before you flip the switch, choose how you want to start.
Choose your channels
Chat widget
Embed on any page
Email
Forwarding address
Voice
Phone number provisioned
WhatsApp
Business number
MS Teams
Internal & external
SMS
Two-way text
How do you want to start?
Recommended for enterprise
Shadow mode
Agent runs silently alongside your team. Answers are logged and reviewed — not sent. After 1–2 weeks, you'll get an accuracy report and a go-live recommendation.
Live mode
Agent handles real customers now. You can switch to shadow mode at any time from settings. Best when simulation results show 85%+ resolution rate.
Based on your simulation — 82% resolution rate, 0 guardrail violations — your agent is ready. Estimated 3–4x handle time reduction once live.