Skip to content
ONE3.ai

Customer · App 04

Resolve customer requests, not just reply to them

One inbox for WhatsApp, email and web chat, with SLAs, cited AI answers and governed Actions that resend invoices, create payment links and open work items from inside the ticket.

Planned · Year 1 (W1)7 modulesYear 1: Partly in
Service Desk is where a business's customers get help on the channel they already use, and where supervised AI Digital Employees and human agents resolve requests by acting on the same customer, invoice, order and work records. It commits to three things: resolve, not just reply; one customer record shared with CRM, Billing and Marketing; and evidence, not claims, with AI and channel cost per ticket and verified outcomes flowing into the Investment & ROI Ledger.

What Service Desk is

Service Desk is One3's customer support app. It is designed for WhatsApp-heavy Indian SMBs, digital agencies supporting many clients, and mid-market B2B teams with contractual SLAs.

Every message from email, WhatsApp, web chat, a web form, the API or the portal is matched to a person or company on the shared customer record. It becomes a ticket with an owner, a queue and SLA timers. Calls and community posts join the same record in later waves. Consent, contact points and history are never copied into a separate support database.

Resolve, not just reply

Most helpdesks can answer a question, but fixing the problem still means opening another system. In Service Desk, a curated set of resolution Actions sits in the ticket side panel: resend an invoice or GST e-invoice copy, create and send a UPI payment link, show payment status, or create a client work item. These run through One3's governed Action Bus with risk tiers and approvals. Refunds and credit notes are only ever proposals, routed to the Approvals Inbox with an evidence pack and a second factor.

Supervised Digital Employees

The Support Triage & Reply Assistant starts in shadow mode, sorting tickets by category, priority, sentiment and language while people check its accuracy. As it earns autonomy, it can send knowledge-grounded first replies on WhatsApp, web chat and email, with AI disclosure and a structured handoff note when a person is needed. Every reply cites Knowledge sources. Hard AI budget caps per tenant, agent and channel stop AI work when reached and route conversations to the human queue.

A strict resolution definition

A ticket counts as a verified resolution only when it was resolved through the proper Action, an AI check confirms the request was addressed, it was not reopened inside the reopen window, no negative CSAT arrived, and no Action was reversed. Only verified resolutions feed the ROI Ledger, so statements reflect real outcomes.

How it fits with One3

Helpdesk reads the Customer 360 view from CRM and Billing. Knowledge grounds every answer, and unanswered clusters become article drafts. The client portal lets agency clients see reports, approve creatives from Creative Studio, pay invoices and read ROI statements. Later waves add a public Help Center, a Contact Center with Indic voice agents, Voice of Customer themes that feed Customer Success, a Status Page that bulk-updates incident tickets, and moderated Community spaces.

India first

WhatsApp is planned for every edition. India is the default data region for India tenants. Hindi and Tamil UI and AI are planned for W2, each enabled only after its quality eval passes. DPDP basics, including consent at intake, purpose tags, retention and data-principal requests, are designed in from W1 (verify with counsel).

Waves

Helpdesk MVP is planned for W1 and GA for W2, alongside Portals lite. Help Center & Docs, Contact Center, Voice of Customer and Status Page are planned for W3, and Community for W4.

Problems solved

What Service Desk fixes

Support is scattered across a WhatsApp phone, a shared inbox and spreadsheets

Messages get lost and nobody owns them. Service Desk puts WhatsApp, email and web chat in one inbox, and every thread becomes a trackable ticket with an owner and an SLA.

AI answers, but a person still has to open billing to fix anything

Action-taking AI usually needs integration projects. Service Desk ships curated resolution Actions on native Billing, CRM and Work objects, gated by risk tier and approvals.

AI helpdesk bills are unpredictable

Add-on and stacked AI charges are hard to forecast. Service Desk applies hard budget caps, shows cost per ticket on the ticket and feeds a strict, published resolution definition into the ROI Ledger.

Needed features are locked in the top tier

The planned core edition includes WhatsApp, SLAs, macros, copilot and knowledge-grounded replies. Add-ons are whole modules, not single features.

Agencies support many clients from many logins

Clients cannot see progress. A partner tier gives client sub-accounts, per-client SLAs, a white-label client portal, and per-client reports and ROI statements.

Data must stay in India and be ready for DPDP

India is the default region for India tenants, with consent at intake, purpose tags, retention and erasure workflows designed in. Verify obligations with counsel.

Customers write and speak in many languages

Language detection and Indic-aware triage, Hindi and Tamil UI and AI planned for W2, voice-note transcription, and Indic voice agents in Contact Center from W3.

The same complaint comes back every week

Knowledge gap detection drafts articles, Voice of Customer themes create close-the-loop tasks and churn signals, and bugs link to Issues & Sprints.

Key capabilities

What you can do

One omnichannel inbox

Email-to-ticket, WhatsApp Business with service window tracking, and a web chat widget with cited knowledge answers, all on one customer record.

SLA policies and escalations

First response, next response and resolution targets by priority, business hours and holidays, pause on pending statuses, and alerts on warning and breach.

AI triage in shadow mode first

Category, priority, sentiment and language detection that starts by being compared with human decisions and earns autonomy through measured agreement.

Cited suggested replies

Reply drafts grounded in Knowledge with citations, confidence and groundedness, and tracking of whether agents sent, edited or rejected them.

Resolution Actions in the ticket

Resend invoices and GST e-invoice copies, create UPI payment links, check payment status and open work items, all risk-gated.

Hard AI budget caps

Caps per tenant, agent and channel with warnings, automatic pause of non-critical AI and handoff to the human queue when spent.

Verified resolutions and ROI

A strict published definition of resolution, cost lines per ticket and verified outcomes feeding the Investment & ROI Ledger.

White-label client portal

Customers and agency clients raise and follow tickets, see reports, approve creatives, pay invoices and read ROI statements.

Agency multi-client desk

Client sub-accounts with per-client SLAs, cross-client reporting and configuration snapshots for consistent setup.

Workflows

How the work flows

End-to-end workflows inside the app and across One3.

Intake to verified resolution

How any inbound message becomes a ticket and a verified outcome.

  1. 1.A message arrives by email, WhatsApp, web chat, form, API or portal
  2. 2.The sender is matched to a person on the shared record, or a new person is created with a consent notice
  3. 3.The message joins an open conversation and creates or updates a ticket
  4. 4.An SLA policy is matched and timers start
  5. 5.The triage agent sets category, priority, sentiment and language, and assignment rules pick the queue
  6. + 4 more steps →

Invoice copy and payment link on WhatsApp

A Digital Employee resolves a billing request inside the risk gate.

  1. 1.A customer asks on WhatsApp for their last invoice and a payment link
  2. 2.The support Digital Employee retrieves policy from Knowledge and reads the invoice from Billing
  3. 3.If its skill is at the earned level, the connector is healthy and budget is available, it sends the invoice copy and UPI link
  4. 4.Otherwise it creates an approval request with a preview for an agent or team lead
  5. 5.The ticket is resolved with a handover note and cost lines post to the ledger
  6. + 1 more steps →

WhatsApp service window rule

How replies respect WhatsApp's customer service window and templates.

  1. 1.An agent or Digital Employee composes a reply
  2. 2.If the last inbound message is inside the service window, a free-form reply is allowed
  3. 3.Outside the window, an approved template for the purpose is required, with its category and rate shown
  4. 4.Without an approved template, the send is blocked and email or a template submission is suggested
  5. 5.Delivery status is recorded on the ticket

Incident communications

How an outage is communicated without answering hundreds of tickets one by one (planned W3).

  1. 1.A monitoring webhook, ticket spike or staff member opens an incident draft
  2. 2.AI drafts the customer update from internal notes
  3. 3.A publisher approves before anything goes public
  4. 4.The status page, chat widget banner and help centre banner update and subscribers are notified
  5. 5.Matching tickets are linked and receive an approved bulk reply
  6. + 1 more steps →

Voice of Customer feedback loop

Turning repeated complaints into fixes (planned W3).

  1. 1.Tickets, CSAT comments, call transcripts, NPS and survey answers are collected as signals
  2. 2.Personal data is redacted before analysis
  3. 3.Signals are tagged into themes with sentiment
  4. 4.Themes are compared with CSAT, reopen rate, churn and revenue as associations, not causes
  5. 5.A close-the-loop task goes to the owning team
  6. + 1 more steps →

AI inside

AI skills that act through governed Actions

Every AI skill runs through the same typed Actions people use, with risk tiers, approvals and an audit trail.

Ticket triage A4

Sets category, priority, sentiment and language, including code-mixed text. Starts in shadow mode and is promoted only on measured agreement.

Grounded suggested replies A2

Drafts replies from Knowledge with citations, confidence and groundedness. Sending stays with a human unless the reply skill has earned autonomy.

AI first response with handoff A3

Answers on WhatsApp, web chat and email with AI disclosure and hands off with a structured note when a person is asked for, sentiment is negative or confidence is low.

Resolution planning A3

Plans invoice copies, GST e-invoice copies, UPI payment links and client work items under each Action's risk tier.

Refund and credit note proposals A1

Prepares an evidence pack and proposal for human approval with a second factor. Agents never change amounts or issue refunds.

Knowledge gap detection A2

Clusters unanswered questions and drafts help articles for a Knowledge Manager to review and publish.

Voice-note transcription A4

Transcribes WhatsApp voice notes in English, Hindi and Tamil into the ticket thread, per language after its eval passes.

SLA breach risk and duplicate detection A2

Scores open tickets for breach risk and suggests duplicate or related tickets to merge. Merges stay with people.

Conversation QA scoring A4

Scores conversations and calls against the tenant's rubric, calibrated against human QA samples (planned W3).

Digital Employees

Agents that operate Service Desk

Rent them by the month. They work under your team's supervision and their cost and returns appear in the Investment & ROI Ledger.

Integrations

Connects with the tools you already use

  • WhatsApp Business Platform
  • Gmail
  • Microsoft 365 Mail
  • Inbound and outbound email (SPF, DKIM, DMARC)
  • Web chat widget
  • In-app messaging SDK (iOS, Android, React Native)
  • SMS providers with DLT registration
  • Telegram
  • Cloud telephony partners
  • LiveKit
  • UPI payment links via Billing
  • E-commerce platforms
  • Monitoring tools via webhook
  • Google Business Profile reviews
  • Freshdesk
  • Zendesk
  • Zoho Desk
  • HubSpot Service Hub
  • Chatwoot
  • Zammad

Connector availability depends on provider access and is verified before launch. How integrations work →

Security & compliance

Built to be trusted

Tenant isolation on every table

Row-level security is forced on every Service Desk table, with relationship-based access for portals, private status pages and company-private community spaces.

Every external message is gated

Public replies are treated as externally irreversible. Digital Employees send only with earned autonomy for that skill and channel; otherwise they request approval with a preview.

AI is always disclosed

Every AI-authored message follows the tenant's persona policy and is blocked if no disclosure rule exists. Voice agents announce they are AI at the start of every call.

Money stays with people

Digital Employees never change amounts. Payment links cover only outstanding invoice amounts, and refunds and credit notes are proposals needing human approval with a second factor.

DPDP-ready by design

Consent at intake, purpose tags, retention per ticket type, data-principal requests, message redaction and legal hold, with India as the default region. These are planning pointers; verify with counsel.

Sensitive data masking and full audit

Card-number, Aadhaar-like and PAN-like patterns are masked in messages and AI context, and every Action, including AI and approvals, is written to the audit trail.

Frequently asked questions

Is Service Desk available now?

Not yet. One3 is being built in waves. Helpdesk MVP is planned for W1 and GA for W2, with Portals lite in W2. Help Center, Contact Center, Voice of Customer and Status Page are planned for W3, and Community for W4.

Is WhatsApp included?

WhatsApp is planned for every edition, alongside email and web chat. Message charges are shown per ticket, and replies follow WhatsApp's service window and template rules.

Will AI reply to my customers on its own?

Only after it earns that autonomy. The support assistant starts in shadow mode, then drafts for human approval, and only sends knowledge-grounded replies itself once measured quality thresholds are met. It works for your team under supervision, always discloses that it is AI and hands off to a person when needed.

What counts as a resolution?

A verified resolution requires the ticket to be resolved properly, an AI check that the request was addressed, no reopen or follow-up inside the reopen window, no negative CSAT and no reversed Action. Only these feed ROI statements.

Can I control AI spend?

Yes. Hard budget caps per tenant, agent and channel trigger warnings, pause non-critical AI and route conversations to people when reached. Approvals are never blocked mid-flow.

Can we move from Freshdesk, Zendesk or Zoho Desk?

Migration adapters are planned: CSV and spreadsheet import in W2, a Freshdesk ticket adapter from W2, and Zendesk and Zoho Desk adapters in W3, each with a dry run and reconciliation report. Source API terms are verified first.

Does it work for agencies?

Yes. The partner tier is planned with client sub-accounts, per-client SLAs, cross-client reporting and a white-label client portal for each client.

Plan your Service Desk rollout with us

Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.