AI Agents
Discovery Express

Customer ops

Customer Support Agent

Resolves tier 1 tickets inside your CRM and hands the rest to a person with full context.

In productionReviewed September 3, 2026
WhatsApp · Support
Demo data

  • Hi, where is my order 48213? It said today.

  • Looking up order 48213 in the OMS

  • Hi Ana. Order 48213 is out for delivery, arriving between 4 and 6 pm. Do you want to change the address or leave instructions for the driver?

  • Leave it with the doorman please

  • Done. I added the instruction to the delivery and the driver has it. You will get a message when it is delivered.

Resolved without a human

Ticket #48213 · 1.9 s to first answer · CSAT survey queued

KPI it moves
$6.40 to $0.85 per ticketMeasured
Channels
WhatsAppVoiceWeb chatEmail
Integrates
  • Zendesk
  • HubSpot
  • Salesforce
  • Your CRM
  • Telephony
A support lead reading her phone on a fintech operations floor in Mexico City at dusk
Customer ops

What it does

Every ticket answered inside the CRM. The hard ones handed over with context.

Where it lives

Your cloud, your repo, your IP.

1. Receives 2. Looks up 3. Decides 4. Acts 5. Escalates 6. Learns

Channels

  • WhatsApp
  • Voice
  • Web chat
  • Email

Customer Support Agent

03Decides

Systems it integrates

  • Zendesk
  • HubSpot
  • Salesforce
  • Your CRM
  • Telephony

A person on your team

01

Receives

Takes the ticket on WhatsApp, voice, web chat or email and identifies the customer against the CRM.

02

Looks up

Reads the account, the order or the balance in the system of record and the policy that applies in the knowledge base.

03

Decides

Judges whether it can resolve with the permissions it holds or whether the case needs a person.

04

Acts

Answers, updates the ticket, issues the refund or the change within its limits and writes it all to the CRM.

05

Escalates

Hands off by sentiment, amount or topic to a person, with a summary and the actions already taken.

06

Learns

Turns every escalation into eval cases each week; prompts and policy are versioned in your repo.

KPI it moves

Measured, with the source.

Every figure has a row in the Proof Ledger: source, date, owner and client permission.

Proof Ledger
Measured

Measured

$6.40 to $0.85

Cost per ticket, USD

Mexican fintech, 90-day pilot, Proof Ledger Sep 2026

Before

$6.40

After

$0.85

Measured

35% to 68%

Tier 1 tickets resolved autonomously

Same client, 90 days after go-live, Proof Ledger Sep 2026

Before 35%

Measured

42 to 71

NPS

Same client, 90 days after go-live, Proof Ledger Sep 2026

Before +42 · After +71
Reviewed September 3, 2026

How it is governed

Every action has an owner, an eval and a limit.

Set in the Blueprint, enforced by the Govern layer of the platform.

Govern · Approval queue
Demo data

Waiting for a person

Refund $1,240 · above your $500 limit

Ticket #48977 · WhatsApp

Weekly eval

Passing

Tone and policy

97.2%

Last 8 runs
94.1% → 97.2%

Regression suite passed before the last prompt change

Limit, enforced

Daily cost ceiling, auto-pause at 100%

62% of today's ceilingAuto-pause

Human approvals

  • Refunds and credits above the limit you set
  • Changes to account data or payment method

Typical evals

  • Resolution accuracy on a weekly sample of real tickets
  • Tone and policy compliance scored per conversation

Limits

  • No action outside the tools granted through MCP
  • PII masked before any model call

Time to production

8 to 12 weeks

of pilot to production under a KPI Contract, after a Blueprint of 1 to 2 weeks

As a Library agent it needs 2 to 4 weeks of tuning inside the pilot, most of it on your policies and your systems, not on the model.

DaCodes delivery standard, Sep 2026. Integration is the long pole, not the model.

How we deploy
The four phases
Commitment gate

  1. 01Discovery Express

    2 sessions, no cost

  2. 02Blueprint

    1 to 2 weeks

  3. 03Pilot

    8 to 12 weeks

  4. 04Scale & Managed

    Continuous

  • Pilot, first 8 weeks
  • Up to week 12 of the pilot
  • This agent: 2 to 4 weeks of tuning, inside the pilot
  • Commitment gate: the KPI decides Scale

Proof and demo

What backs the label.

Related agents

Same function or industry.

All agents
A clinic receptionist with a headset at the scheduling screen, a patient waiting behind
In production

Customer ops

Appointment Agent

KPI it movesMeasured

No-shows down 18% to 34%

A courier handing a package at a doorstep in Monterrey
In production

Customer ops

Order Tracking Agent

KPI it movesMeasured

Where is my order tickets down 60% to 78%

FAQ

Direct answers.

No. It closes the tier 1 tickets that follow a policy and hands the rest to your team with context. In the fintech case the team moved to tier 2 and retention work while cost per ticket fell from $6.40 to $0.85.

WhatsApp Business, voice, web chat and email on the customer side; Zendesk, HubSpot, Salesforce or your own CRM on the system side, through native APIs or MCP. It reads and writes the same tickets your team sees.

It escalates. Thresholds by confidence, sentiment, amount and topic are set in the Blueprint, and every escalation arrives with a summary and the actions already taken. Those cases become eval cases the following week.

A pilot of 8 to 12 weeks under a KPI Contract. This agent needs 2 to 4 weeks of tuning inside that pilot, most of it on your policies and CRM fields, not on the model.

Which process do we automate first?

Two discovery sessions, no cost, and you leave with two or three quick wins whether or not you build with us.

We reply in under one business day.

Loading the form