AI agentSupport for WhatsApp, web chat & email
Onfre learns your docs, policies and past tickets, then resolves customer questions on its own: it looks up orders, issues refunds, books appointments, and hands the tricky ones to your team with the whole story.
Priya
Indigo Loom · via WhatsApp
What Onfre did
Resolved by AI
Handed to humans
Resolution rate
Support teams running on Onfre
<0 s
first reply, any hour
0+
languages, incl. Hinglish
0/7
on every channel
0
channels, one agent
The problem
Customers message when it suits them, mostly on WhatsApp and often at night. More people help with the hours, but not with answering the same 40 questions over and over.
Most support questions are repeats: order status, returns, delivery areas, plan changes. They pile up after hours, when nobody is there to answer.
Your team spends the first half of every week replying to people who have already waited two days. Some of them have already left a review.
Slow replies show up as cancelled COD orders, chargebacks and customers who quietly switch to a competitor.
How it works
No flows to draw, no intents to label, no decision trees to maintain. You give Onfre the knowledge your team already uses, and decide where the lines are.
Link your WhatsApp Business number, add the web chat widget and forward your support inbox. Your number stays yours.
~15 min
Point Onfre at your help center, policies, product catalog and past tickets. It learns your answers and your tone.
~30 min
Choose what it does on its own, what needs approval, and when it must always hand over to a person.
~20 min
Onfre drafts replies to your last 500 tickets so you can review them. Switch on one channel, then widen.
same afternoon
What Onfre does
Every reply is grounded in your own docs, policies and data, with the source linked. No source, no answer: Onfre hands over instead of guessing.
Can I return something I bought in the sale?
Sale items can be exchanged for a different size within 7 days, but they can't be returned for a refund.
Inside your tools, within the limits you set.
On request, on frustration, or when your rules say so.
Handed to Meera · Payments
Summary: Refund failed 3×, UPI verified, customer frustrated. ₹6,400 needs approval.
Replies in whatever the customer writes in.
bhai mera parcel kab aayega?
Aaj shaam 7 baje tak pahunch jayega! 📦
Learns your tone from your best agents' replies.
A weekly report shows what people asked, what Onfre resolved, and which questions had no good answer in your docs, so you know exactly which page to write next.
Under the hood
Here's what happens in the few seconds between a customer's message and Onfre's reply.
bhai mera order 6 din se nahi aaya 😟
Aaj shaam 7 baje tak pahunch jayega! 📦
Step 1
Works out what the customer wants, in which language, and how they're feeling.
Step 2
Pulls only what this question needs from your docs and systems.
Step 3
Applies your guardrails: confidence, blocked topics, approval limits.
Step 4
Does whatever the answer needs, inside the limits you set.
Step 5
Answers in the customer's language and your brand's tone.
Step 6
Logs the sources and outcome, and feeds your weekly report.
Try it yourself
Pick a business and a question. Watch Onfre answer, show its sources, take action, or decide it's one for a human.
1. Pick a business
D2C apparel · 1,900 orders/week
2. Ask a customer question
Scripted examples for a fictional business. In your account, Onfre answers from your own docs and data.
Pick a question to see Onfre work
You'll see the reply, the sources it used and anything it did.
Solutions
Each industry has its own repeat questions, its own actions and its own lines that must never be crossed.
Order status, returns, sizing and COD questions, answered in seconds.
Questions it handles
Actions it takes
Example conversation
hi, ordered a kurta in M but it's too tight. can i get L?
Of course! Size exchanges are free within 15 days. I've set up an exchange for order #IL-50917: a courier will collect the M and drop off the L on Thursday.
perfect, do I need to pack it?
Why Onfre
Rule-based bots are cheap but frustrating. More hires are great but slow and expensive to scale. Onfre sits where you actually want to be.
| Capability | Rule-based chatbot | Hiring more agents | Onfre |
|---|---|---|---|
| Answers at 2 a.m. and on Sundays | Yes | Only with night shifts | Always |
| Understands messy, Hinglish questions | Needs exact keywords | Yes | Yes, 30+ languages |
| Answers from your real policies | Only scripted answers | After training | Yes, with sources |
| Takes actions (refunds, bookings) | Rarely | Yes | Yes, within your limits |
| Knows when to bring in a person | Loops until users give up | They are the person | Yes, with full context |
| Setup and upkeep | Flows to build and maintain | Hiring and training | An afternoon, then it learns |
| Cost as volume grows | Low | Grows with every hire | Per resolution, never per seat |
Integrations
Onfre reads from your knowledge base, acts in your store and billing tools, and syncs every conversation to your helpdesk. No migration, no new inbox for your team to learn.
Customer stories
We used to start every morning with 300 unread WhatsApp chats. Now Onfre clears the overnight queue before anyone logs in, and my team only sees the conversations that actually need a person.
Ritika Sharma
Head of CX, Indigo Loom
71%
of chats resolved without an agent
“The handover is what sold us. When Onfre passes a chat over, the agent gets the order, the history and what the customer already tried. Nobody has to say 'can you repeat that?' anymore.”
Daniel Fernandes
Support Lead, Northbay Fibre
9 s
median first reply, down from 3 h 40 m
“Patients book, reschedule and ask about prep instructions in Hindi, Marathi and English. Onfre answers from our own clinic protocols and never guesses at anything medical. It hands those questions to our staff.”
Dr. Anjali Rao
Operations, Tandem Clinics
1,140
appointments handled in the first month
Estimate your savings
Move the sliders to match your support desk. The estimate uses our published pricing and your own numbers.
Your estimate, per month
1,800
Conversations resolved by Onfre
180
Agent hours freed up
$1,080
Agent time saved
$223
Onfre Starter plan
Estimated monthly saving
$857
plus answers in seconds for 1,800 customers, at any hour
A rough estimate from your inputs, not a promise. Resolution rates depend on your docs, your rules and the kind of questions you get. Onfre drafts answers to your real tickets during the trial so you can see your own rate before paying anything.
Pricing
Every plan includes a 14-day trial. Handovers to your team are never billed.
$79/month
600 AI resolutions / month
One WhatsApp number and a web chat widget, answered day and night.
$249/month
2,500 AI resolutions / month
Every channel, plus actions in your store, billing and helpdesk.
$749/month
8,000 AI resolutions / month
For high-volume teams that need custom actions and controls.
Security
Encrypted end to end
In transit and at rest
Never trains shared models
Your data improves only your agent
Sensitive data masked
Cards, IDs and OTPs redacted
Guardrails you control
Blocked topics and approvals
FAQ
Onfre only answers from the sources you connect: your help docs, policies, product data and past tickets. Every answer is linked to the source it came from. If it can't find a confident answer, it says so and hands the chat to your team instead of guessing.
Connect WhatsApp, point Onfre at your help docs, and see it draft answers to your real tickets before any customer does.