Stellah AIApply

Remote role for Philippines-based operators

Conversational AI Analyst

Catch AI conversation failures before they become customer or client operations problems.

This is not script-based support. You will review AI-handled conversations, handle first-line customer issues, diagnose what went wrong, and write clear escalation notes for the internal team.

Only apply if you can:
  • Read detailed playbooks and apply rules in the right order.
  • Understand the root issue behind a customer report or client request.
  • Explain messy cases clearly without needing constant supervision.
First-line supportAI conversation reviewRoot-cause diagnosisInternal escalation
PHP 40,000-65,000/monthCustomer support, QA, SaaS, IT, BPO, or ops experienceAssessment emailed if shortlisted
Apply nowSee process

About Stellah AI

We build AI systems for customer conversations.

Stellah AI helps service businesses handle customer conversations, qualify leads, support booking flows, and keep customer operations running smoothly.

This role sits between customers, client playbooks, and the AI system. You help spot the details that could create wrong expectations, confused customers, or avoidable work for the internal team.

What you will own

You will be the human judgment layer for customer conversations.

You are not only answering tickets or checking AI replies. You help Stellah understand what is really happening across the customer, client, playbook, AI system, and internal operations.

Understand reported issues and manage expectations

Act as first-line support when customers surface issues: clarify context, set expectations, and decide what needs investigation.

Review AI conversations against client playbooks

Check whether the AI followed the correct rule, asked the right question, and avoided risky claims.

Diagnose root causes across connected variables

Understand how customer context, client policies, service constraints, AI behavior, and workflow outcomes affect each other.

Coordinate fixes with the internal team

Write short notes that explain the issue, likely root cause, why it matters, and what should happen next.

Fit filter

Strong candidates usually have this pattern.

We care more about judgment, root-cause thinking, and consistency than polished corporate language. Specific examples beat generic claims.

Likely a fit

  • Direct AI operations experience is not required; we can train people who demonstrate strong judgment, detail, clear writing, and a willingness to learn
  • Able to commit full-time to this role without another job, client, or side commitment
  • Strong written English with clear customer-facing replies and short, useful internal notes
  • Careful and methodical when reviewing conversations, playbooks, and edge cases
  • You like messy cases where the first customer message does not explain the real problem
  • Comfort reading detailed playbooks and applying rules in the correct order
  • Able to connect client requests, customer context, AI behavior, and operational constraints
  • Can identify the root issue behind a feature request, bug report, or unclear customer conversation
  • Able to spot edge cases, risky replies, missing context, and unclear instructions
  • Self-starting and reliable with minimal supervision
  • If you have relevant experience, can describe each employer/account, role title, work period, duration, support channels, and KPIs specifically

Probably not a fit

  • Part-time availability, 20-30 hours/week, or keeping another job/client
  • Claims experience only as '7 years' or 'BPO/support' without company-by-company periods, roles, channels, or KPIs
  • Needs a fixed script for every edge case
  • Follows instructions literally without understanding the underlying problem
  • Struggles to connect policies, customer context, data, and workflow outcomes
  • Repeated grammar or clarity issues in written answers
  • Rushes repetitive work or skips details
  • Wants basic ticket answering without QA or process ownership
  • Uncomfortable with AI-assisted work being reviewed and scored
  • Needs constant direction before taking the next obvious step

Hiring process

The process is intentionally practical.

We are filtering for careful operators, not people who can write a perfect cover letter.

1

Application

5-8 min

Answer the written questions carefully. This is the first step only, so you can apply from mobile and complete deeper tasks later.

2

Video intro

2-3 min

Shortlisted applicants send a short video with examples, availability, and why this judgment-heavy role fits them.

3

Practical assessment

75-90 min

If you pass the earlier steps, we email a detailed playbook assessment to complete later with focused time at your computer.

4

Zoom interview

30-45 min

We discuss your reasoning, judgment, and how you handle unclear customer or client cases.

5

Paid trial

Finalist stage

Finalists complete a realistic task before any long-term decision.

Judgment preview

The work often starts with an unclear customer issue.

We do not expect you to solve every case alone. We do expect you to slow down, check the right variables, and explain the issue well before escalating.

Example case

A customer says the AI promised they were eligible for a booking.

The client playbook says medication history may require human review, but the conversation notes are incomplete and the customer is now expecting confirmation.

A strong operator would check

  • What the AI actually said versus what the playbook allows.
  • Which missing details affect the customer's eligibility or expectations.
  • Whether this is a one-off conversation issue, unclear playbook rule, or system behavior that may repeat.
  • What to tell the customer now, and what internal note the team needs to resolve the root issue.

Application

Start with the short application.

The practical assessment is emailed only if you are shortlisted. Read each question carefully. Include the word playbook in your final answer to confirm that you followed the instructions.

Having trouble? Use the backup Notion form

FAQ

Before you apply

What experience is most useful?

Customer-facing tech support, SaaS support, IT support, BPO, QA, CX operations, CRM operations, and similar roles are useful if you can show strong judgment, root-cause thinking, and attention to detail.

Is this a normal customer support job?

No. You will still support customers, but the role is judgment-based. The core work is monitoring AI-assisted customer conversations, checking them against client playbooks, understanding what issue a client or customer is really surfacing, and writing useful internal notes.

How much time does the hiring process take?

The first application usually takes 5-8 minutes. If shortlisted, the video intro is 2-3 minutes and the practical assessment is designed for 75-90 minutes of focused time.

Can I use AI tools during the process?

You may use AI tools where the instructions allow it, but you must disclose how you used them and how you checked the output. Blind AI answers are a weak signal.

Why is there a practical assessment?

The role depends on real judgment: finding hidden edge cases, applying rules in order, understanding relationships between variables, and knowing what to try before escalating. The assessment tests that directly.

What is the salary range?

The budgeted range is PHP 40,000-65,000 per month, depending on experience, fit, and scope. Final compensation is confirmed only at offer stage.