Maum AI

Korea's publicly listed multimodal AI platform for contact centres, robotics, and on-premise enterprise LLMs

Business & Finance Enterprise Has API Open Source
Researched · Published · Reviewed
RECATOOLS Score
6.2 / 10
Capability
7
Value for money
5.5
Ease of use
5
ASEAN readiness
6
API quality
6
Founded
2014
HQ
Seongnam-si (Pangyo), South Korea
Users
~115–124 employees; enterprise customer base (undisclosed count)
Launched
KOSDAQ IPO: 23 Nov 2021 (ticker 377480)
Developer
None (independent KOSDAQ-listed company)

Overview

Maum AI (formerly MINDs Lab) is a South Korean AI platform company founded in 2014 and publicly listed on KOSDAQ since November 2021. Its core offering is the maum.ai platform — a suite of deep-learning AI engines covering conversational AI, AI Contact Centre (AICC) automation, an on-premise large language model appliance (MAAL), an on-device voice AI (SUDA), and an AI English education service. The company serves enterprises in finance, telecommunications, public sector, and manufacturing, and has a notable presence in the Philippines alongside South Korea, the United States, and Canada.

Beyond its SaaS and AIaaS lines, Maum AI is pivoting toward Physical AI through its WoRV research division, which develops Vision-Language-Action (VLA) foundation models for robotics and autonomous vehicles. Open-source model releases on GitHub and Hugging Face — including MAAL-8B (Llama-3 fine-tune), CANVAS (commonsense robot navigation, Outstanding Paper at NeurIPS 2024), and audio models with over 1,000 GitHub stars — demonstrate genuine research credentials. Revenue for fiscal 2025 was approximately 9.8 billion KRW (~USD 7 million), and the company continues to operate at a loss as it invests in expanding its foundation-model portfolio and Physical AI data factory initiative.

Advertisement

Pricing

Pricing shown for reference only. These figures reflect RECATOOLS research as of 11 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.

Enterprise
Custom
Full solution deployments quoted per project
  • SUDA, MAAL and WoRV platform lines
  • AI contact centre and industry solutions
  • Contract pricing via sales

Use cases

Deploying AI-powered voice and chat agents in enterprise contact centres to handle inbound customer queries without human agents Running a secure on-premise LLM (MAAL) for internal document Q&A, summarisation, and workflow automation in regulated industries such as finance, healthcare, and defence Providing AI-native English speaking practice and conversation coaching to Korean students and corporate language-training programmes Equipping industrial robots and autonomous construction equipment with Vision-Language-Action models (WoRV/CANVAS) for real-world task execution Building multimodal AI pipelines via the maum.ai platform that combine speech recognition, NLU, and chatbot modules for kiosk, smart-factory, or smart-city deployments

What you can produce with Maum AI

  • Reduction in contact-centre human-agent handling time through AICC voice and chat automation
  • Secure on-premise LLM deployment with MAAL scoring 9.06/10 on enterprise task benchmarks vs. GPT-4o's 9.59
  • Open-source CANVAS navigation framework awarded Outstanding Paper at the NeurIPS 2024 Workshop on Open-World Agents for robot commonsense navigation
  • Publicly available MAAL-8B and MAAL-11B-Vision Hugging Face models for Korean-language and multimodal fine-tuning
  • One of four companies selected by the Korean government in 2023 (alongside Naver Cloud, KT and VAIV Company) to provide hyperscale AI services to government agencies and SMEs
  • Physical AI Data Factory targeting 300 TB of annual industrial robot training data in partnership with Dongguk University
  • MOU with Doosan Bobcat (June 2026) for VLA-model-driven autonomous compact construction equipment
Advertisement

ASEAN Perspective

Maum AI in Southeast Asia

Maum AI has an established office presence in the Philippines — a natural beachhead given that country's large BPO and contact-centre industry — and lists Southeast Asia as a target market for AICC and AI English education. However, its customer references and publicly documented deployments remain concentrated in South Korea (banking, telecoms, public sector). Korean-language optimisation is a core competitive strength, which reduces out-of-the-box utility for non-Korean ASEAN deployments; organisations in the region would need to evaluate multilingual or English-language pipeline capabilities separately. The company's Physical AI Data Factory and WoRV robotics work are aligned with ASEAN industrial automation growth, but commercial traction in the region has not yet been publicly confirmed as of mid-2026.

RECATOOLS Verdict

Maum AI is one of the few AI companies in Asia that's actually audited: it's been on KOSDAQ since November 2021, so its losses, R&D spend, and mixed revenue trend are all a matter of public record rather than founder claims. That transparency is unusual for an AI vendor and worth crediting on its own.

The technology case is real too. CANVAS, its robot-navigation framework, won an Outstanding Paper award at a NeurIPS 2024 workshop, and the open Llama-3-based MAAL-8B model sits on Hugging Face for anyone to test. The June 2026 Doosan Bobcat MOU on autonomous construction equipment shows the WoRV robotics pivot has real industrial interest behind it.

The catch is financial: operating losses persist even as 2025 revenue growth resumed, and ASEAN presence beyond a Philippines office is thin. Treat this as a promising but still-proving Korean specialist, not a plug-and-play SaaS pick.

Independent AI-assisted assessment by RECATOOLS.

What people say

Start with the number that makes Maum AI unusual among AI vendors: it's a public company. Ticker 377480 on KOSDAQ since November 2021, formerly MINDs Lab, which means anyone can pull the actual revenue and loss figures instead of trusting a pitch deck. FY2024 revenue fell to roughly 7.8 billion KRW, down nearly 24% from the prior year, before growth resumed through 2025 — Q1 2025 revenue was up about 9% year over year, though the company was still posting an operating loss of roughly 1.9 billion KRW that quarter.

The product line is genuinely broad for a company this size: AICC contact-centre automation, the MAAL on-premise LLM appliance, the SUDA on-device voice engine, and — the newer bet — WoRV, a Vision-Language-Action research arm chasing Physical AI. That last one isn't vaporware. CANVAS, Maum AI's commonsense robot-navigation framework, picked up an Outstanding Paper award at a NeurIPS 2024 workshop, and the company signed an MOU with Doosan Bobcat in June 2026 to explore autonomous construction equipment. MAAL-8B, a Llama-3 fine-tune, is published openly on Hugging Face.

What doesn't hold up as well is any claim of broad international traction: outside Korea, the company's footprint is essentially a Philippines office plus early-stage North American presence, not a Southeast Asian sales operation. There's also no independent G2 or Gartner review trail to check vendor claims against — evaluation runs through the company's own investor filings and press releases, not third-party user feedback. For enterprises comparing AI vendors, that mix of financial transparency and thin outside review coverage cuts both ways.

Summary of public user & expert reviews, compiled by RECATOOLS.

Notable facts

  • Maum AI's VoiceFilter repo — an unofficial PyTorch implementation of Google's audio separation research — has accumulated over 1,200 GitHub stars, making it one of South Korea's most-starred AI open-source projects.
  • CEO Yoo Taejoon holds a CPA license from Samil PwC, making him one of the few AI company founders whose career began in accounting rather than engineering.
  • Maum AI's CANVAS robot-navigation system won an Outstanding Paper Award at the NeurIPS 2024 Workshop on Open-World Agents, placing it among a small group of Korean AI labs with top-tier international workshop recognition.
  • The company's WoRV division (World Models for Robotics and Vehicle Control) operates a fleet of hundreds of robots in real industrial sites — agricultural, logistics, defense, and construction — as training data generators for its foundation models.

Frequently asked questions

What is the difference between MAAL and SUDA?
MAAL is Maum AI's on-premise large language model appliance — a 70B-parameter model deployed on a company's own servers for organisations with strict data-security or network-separation requirements. SUDA is a separate on-device voice conversational AI designed to run locally on edge hardware, targeting voice-based interfaces in environments without reliable cloud connectivity.
Does Maum AI offer a public API or self-serve developer plan?
Maum AI provides a Minds API SDK covering Korean and English speech recognition, natural language analysis, and machine reading comprehension. Access requires contacting the team directly ([email protected]) to receive credentials; there is no confirmed public self-serve or free-tier sign-up as of mid-2026.
Is Maum AI open source?
Partially. Maum AI publishes model weights and research code on GitHub and Hugging Face under open licences (primarily BSD-3-Clause). Notable releases include Llama-3-MAAL-8B-Instruct, the CANVAS robotics navigation framework, and several audio/speech models. Its commercial platform products (AICC, MAAL appliance, SUDA) are proprietary.

About this listing

Researched on
Published on
Last reviewed

This entry was compiled from publicly available data including Maum AI's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Maum AI unless explicitly stated.

Data accuracy

Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.

For the latest details, please refer to Maum AI directly →

Spotted something out of date? Suggest an update →

Advertisement