AI Product Manager

Pantas; Kuala Lumpur; On-site; MYR 180,000 - 240,000; Full time

Job summary

Pantas is an AI company. We’re hiring an AI Product Manager to own AI product work across the ESG platform and Axsona—not a single backlog. You’ll translate customer needs into specs and work with engineering until the feature ships and is used.

This is a flat, AI‑first team where the CEO codes and runs monthly AI workshops. You should have 3+ years as a PM in software or SaaS and real examples of AI you built—an agent, automation, or eval harness. Visa is not provided.

Qualifications

• 3+ years as a PM in software or SaaS with end-to-end ownership and shipped AI work.
• Proven ability to define specs and guide AI features from concept to production.
• Strong written communication and judgement in ambiguous data scenarios.

Responsibilities

• Own AI product area end to end across ESG and Axsona.
• Decide what to build and what not to.
• Write specs engineers can implement from.
• Define success metrics and ensure quality before launch.
• Prioritize across ESG and Axsona; defend trade-offs.

Skills

AI product management

Shipping AI features

Backlog ownership

Spec writing

User research

Tools

Claude Code

Codex

Gemini

Databricks

OpenRouter

HuggingFace

Vertex AI

Job description

Most PM roles hand you a roadmap. This one hands you a portfolio.

Pantas is an AI company. We’re best known for the platform that helps financial institutions and corporates across Southeast Asia measure, manage, and report carbon and ESG risk — but that is one product, not the company.

We also built Axsona and its flagship AI Twins: demographically accurate AI agents deployed at scale as synthetic populations, so decision-makers can stress-test messaging, products, and strategy before they spend a cent finding out. More AI products are coming behind it.

We’re hiring an AI Product Manager to own AI product work across that portfolio — not to babysit one backlog.

We are unable to provide visa. Please do not apply if you require a visa for this role. Your application will not be reviewed.

The role

You will own AI product work across the Pantas portfolio — AI-native features in the ESG platform and simulation products in Axsona — from figuring out what customers actually need, to writing the spec, to working with engineering until it ships and people use it. You do not need to be an AI researcher. You need to be a strong software product manager who already builds with AI, has taste about what is genuinely useful versus merely impressive, and is willing to make the call when the data is incomplete.

What you’ll actually do:

• Own a product area end to end across both the ESG platform and Axsona.
• Decide what gets built and what doesn’t.
• Write specs engineering can build from without guessing.
• Define what “working” means for an AI feature — accuracy, usefulness, when a human stays in the loop — and check the output yourself before customers see it.
• Understand the customer problem, decide what to build, and prioritise what does not get built — across both ESG platform and Axsona surfaces.
• Hold a prioritised backlog rather than a wish list, and defend the trade-offs when someone senior disagrees.

What “AI-first” actually means here:

• A flat team doing advanced agentic work. No layers — just people building at the frontier.
• Your CEO codes, and runs monthly AI deep-dives on prompting, agents, MCP, context and harness engineering.
• Best-in-class tooling — Claude Code, Codex, Gemini — used daily by every role, technical or not.
• You prototype your own ideas. PMs here build, they don’t just write tickets.

Who we’re looking for:

3+ years as a product manager in software or SaaS, owning an area end to end — with evidence you’ve shipped. And the one thing we care about most: show us something real you’ve built with AI. An agent, an automation, a prototype, an eval harness, an LLM‑powered feature. Evidence beats credentials.

The one thing that won’t fly:

Chatting with an AI, finishing courses, or listing tools. Show us something you built, shipped, or automated.

How we work: AI-first, for real

Your CEO codes — and runs monthly AI workshops. Our founders build alongside the team, leading deep-dives on prompting, agents, MCP, context engineering, and harness engineering. That is rare for a product manager.

AI is the default, not a side team. Everyone — including non-technical roles — uses Claude Code, OpenAI Codex, Gemini, and others daily to prototype, build agents, script, and automate. We share playbooks so adoption compounds.

PMs build, not just brief. You will be expected to prototype your own concepts, build evals for AI features, and put a working artefact in front of engineering rather than a wall of text. You do not need to be a software engineer. You do need to be comfortable creating with AI.

A flat team across the portfolio. No layers to navigate. You will work across frontier and open-source models (GPT, Claude, Gemini, DeepSeek, Qwen, Kimi, GLM) and modern infra (Databricks, OpenRouter, HuggingFace, Vertex AI).

Ship AI-native products

• Take AI features from concept to production: LLM-based extraction, document understanding, decision support, agentic workflows, and audience simulation.
• Prototype with AI yourself — then refine into stable features with engineering and data science.
• Turn AI‑generated PRDs into work engineering can execute, and keep the loop tight between prototype and production.

Define what “working” means

• Set the quality bar for AI features: accuracy, usefulness, failure modes, and when a human should stay in the loop.
• Build and run evals. Check the output yourself before it ships to customers — for simulation work, “looks plausible” is not a standard.
• Track whether shipped work actually moved the number it was meant to move, and say so honestly when it did not.

Work with the business

• Work day to day with engineering, forward‑deployed analysts, and account managers — grooming, sprint planning, unblocking decisions, reviewing what ships.
• Talk to customers and business development regularly; translate what you hear into a roadmap.
• Connect AI capability to commercial value — what a client will pay for, and why.

What you’ll need

• 3+ years as a product manager in software or SaaS, owning a product area end to end.
• A track record of shipping — features that reached real users, not just specs and decks.
• Evidence that you already build with AI, not just use it. This matters more to us than your years of experience.
• Clear written communication. You can put a problem, the options, and a recommendation on one page.
• Technical literacy: comfortable with knowledge of AI, data, and how the product is actually built. You do not need to code, but you will build with AI‑assisted tools.
• Judgement and taste — the ability to tell an impressive‑looking output from an accurate, fit‑for‑purpose one.
• Comfort with ambiguity, fast‑changing ground, and making the call when the data is incomplete.

Nice to have

• Shipped AI or ML features in a live product, especially with LLMs or agents.
• Experience with enterprise customers and their procurement, security, and compliance reviews.
• Exposure to research, consumer insights, market research, or simulation — relevant to AI Twins.
• Sustainability, ESG, carbon accounting, or financial services domain knowledge — useful, but learnable.

Show us your work

This is the part that matters most. We’ll ask you to show something real you’ve built with AI — an agent, an automation, a prototype, an eval harness, an internal tool, an LLM‑powered feature. It does not need to be production software or a launched product. Side projects, hackathons, internal tools, and self‑directed work all count.

• the problem or objective,
• what you personally built or decided,
• the AI tools and models you used,
• the result or value it produced,
• and a link, repo, demo, or screenshot (a concise write-up is fine where the work is confidential).

Listing tools you’ve tried, courses you’ve finished, or chats you’ve had with an AI is not enough. We want evidence you’ve built something.

What this role is — and isn’t

• It is a portfolio role — not a single‑product ESG PM job. If only the sustainability part interests you, this is the wrong role.
• It requires building with AI — but not production infrastructure or foundation‑model training.
• It is not a delivery‑manager or ticket‑writing seat. You own outcomes, not throughput.
• It is not an intro AI training ground for people who are interested in AI but haven’t yet built anything with it. We will sharpen real ability fast — we’ll teach applied AI from zero.

What we offer

Real ownership across a portfolio of AI products with paying enterprise customers — not a greenfield experiment, and not someone else’s playbook to inherit. A flat, fast team where you build alongside a technical CEO who still codes. Direct access to customers and senior decision-makers. Hybrid, flexible work in Malaysia. Competitive salary, ESOP, monthly AI workshops, and top‑tier AI tooling.

Confidentiality

Our product methods, internal workflows, and client engagements may be confidential. Specific client and project details may be disclosed at offer stage under NDA. You must be comfortable exercising discretion and maintaining strict information security throughout.

We’re proud to be an equal opportunity employer and celebrate diversity across race, religion, colour, national origin, gender, age, marital status, sexual orientation, and disability status.