Miner Technology releases FY2025 results: Fully entering the AI-native operations era, with accelerated implementation of intelligent agent services

Miner Technology (2718.HK) officially announced its full-year results for the fiscal year ended December 31, 2025. During the reporting period, the company achieved operating revenue of 1.426 billion yuan, representing a year-on-year increase of 3.2%; gross profit reached 790 million yuan, growing by 10.8% year-on-year; adjusted net profit (Non-HKFRS) reached 42 million yuan, turning losses into profits.

While financial performance and operational quality continued to improve, Miner Technology has fully entered the AI-native operations era. AI is accelerating its transformation from an auxiliary tool to a key productive force embedded in real business flows, driving the company from 'providing data intelligence' towards 'delivering quantifiable results.' Among these developments, the Agentic Services business model has completed a critical breakthrough from capability validation to commercial implementation, becoming a significant marker of Miner Technology's advancement in AI commercialization.

Wu Minghui, founder, CEO, and CTO of Miner Technology, stated: 'In 2025, Miner Technology completed a pivotal transformation—from helping clients 'understand data' to helping them 'achieve results.' The significance of Agentic Services lies in advancing the value of AI from delivering tools to delivering outcomes. This not only changes the logic behind how clients procure professional services but is also reshaping service methods and business models, creating greater growth opportunities for Miner’s future.'

Continuous optimization of operational quality, with AI deeply reshaping business flows

In FY2025, Miner Technology continued to build on its foundation of Data Intelligence, using Agentic AI as its core driver to promote continuous optimization of its business structure. The company has currently empowered approximately 2,100 well-known brand clients and over 240,000 enterprise users, continuously helping clients optimize decision-making and improve operational efficiency. During the reporting period, the renewal rate of major clients reached 96%, with the core client base remaining stable, showcasing strong resilience and stickiness across overall operations.

Leveraging the proprietary enterprise-grade Trusted Business Intelligence Agent Platform DeepMiner, the company has systematically integrated AI into its core business lines, accelerating the transformation of its delivery logic towards an 'Agent-driven' approach. Significant progress has been achieved across multiple key scenarios, leading to operational efficiency improvements in the Data Intelligence business unit. Delivery efficiency for marketing intelligence operations has increased up to fourfold, while operational intelligence has reduced single work order resolution time by over 30%.

During the reporting period, the Miaozhen Systems team fully advanced the Agentic AI transformation of workflows, achieving end-to-end integration within 60 days and systematically addressing the integration challenges of social media data processing with AI applications. With necessary human-machine collaboration retained, the AI system achieved a 90% automation rate across the entire workflow. The production of in-depth review reports saw a 20-fold increase in labor efficiency, with notable improvements in insight stability, framework completeness, deep analytical capabilities, and micro-trend identification novelty. AI is gradually evolving from an auxiliary tool to a key driver of productivity within business processes, enabling the company to transition from traditional labor-intensive delivery models toward a more efficient and reusable intelligent operating system.

The Agentic Services model is accelerating its implementation and has already generated scaled revenue.

Building on extensive accumulated marketing data, client scenario understanding, and industry know-how, MiningLamp Technology has proactively positioned itself in the Agentic Services model and is driving its accelerated deployment. During the reporting period, the company’s Agentic Services segment generated revenue exceeding 100 million yuan. As a crucial direction for the company's AI commercialization efforts, Agentic Services goes beyond tool output or single-point capability delivery, instead focusing on quantifiable business objectives for clients and providing end-to-end services closely aligned with results.

During the reporting period, over 30% of the company’s new major clients came from the Agentic Services business segment, indicating that this model has demonstrated strong commercial adaptability and practical appeal across diverse industry demands. Focused on marketing scenarios, MiningLamp Technology has streamlined an end-to-end workflow spanning data insights, content production and distribution, and ad placements, helping clients achieve an average 20% improvement in marketing effectiveness with nearly three times the operational efficiency.

Continuous enhancement of the AI-native infrastructure is building long-term barriers for trusted intelligence.

Whether it is the improvement of core business efficiency or the accelerated implementation of the Agentic Services model, both rely on the ongoing support of the underlying technological architecture. As MiningLamp Technology’s self-developed enterprise-grade Trusted AI Agent platform, DeepMiner has become the unified infrastructure connecting all aspects of the company’s business capabilities, continuously supporting the transition from point-based efficiency gains to systemic-level collaboration.

DeepMiner employs a multi-agent hierarchical collaboration architecture of 'scheduling-decision-execution,' enabling full-chain collaboration in complex business scenarios from task comprehension, path planning, to execution. As of October 2025, the company’s proprietary GUI model Mano ranks second overall on the OS-World leaderboard and first on the Mind2Web leaderboard; the reasoning model Cito leads in the small-size model category in BFCL. Through continuous innovation and industry-leading technical strength, MiningLamp Technology further solidifies its long-term competitive edge in the trusted AI space.

Notably, MiningLamp Technology has deeply embedded AI into the daily operations of its organization. During the reporting period, 100% of employees were connected to the DeepMiner platform. Employees can independently create and deploy customized business Agents based on specific operational scenarios, promoting internal standard operating procedures to evolve from unidirectional human execution to human-machine collaborative orchestration. Through this ongoing internal practice, DeepMiner has gradually evolved into organizational-level infrastructure supporting the company’s day-to-day operations.

Looking ahead, we aim to drive the scaled development of Agentic Services.

Minglue Technology will firmly focus on the core direction of 'trustworthiness' and continue to advance the in-depth development of its business. On one hand, the company will promote the integration of software and hardware deployment to further solidify the Agentic AI infrastructure; on the other hand, it will accelerate the replication of the Agentic Services model across more scenarios, continuously expanding its application boundaries in content marketing, entertainment ecosystems, and more digital white-collar work areas, while penetrating labor and expertise-intensive industries.

Addressing practical pain points such as system fragmentation, process breakpoints, and data silos, the company will drive Agent-based transformations across more industries through the scaled output of trusted AI labor.

At the same time, Minglue Technology will continue to externally deliver end-to-end AI scheduling capabilities, systematically expanding its global business footprint and actively embracing the era of the Internet of Agents (IoA), exploring new paradigms for human-machine collaborative productivity.

From providing tools to delivering results, from localized efficiency improvements to AI-native operations, Minglue Technology is continuously validating the industrial value of trusted Agentic AI through real business closed loops. It aims to gradually establish the standards of 'verifiable outcomes, auditable processes, and transferable knowledge' as universal principles for native AI delivery in the industry. Along the path of 'making AI truly integrate into business, achieve outcomes, and scale,' Minglue Technology will continue to move forward with determination.

14K Views