During the early days of smartphone proliferation, Nokia CEO Jorma Ollila once lamented: "We didn't do anything wrong, but somehow we lost." Times have changed.

Today, the mapping industry stands at a similar turning point. With OpenAI releasing GPT-5 and DeepSeek training cutting-edge models using domestic chips, AI continues to surge forward. From traditional GPS navigation to intelligent route planning, map applications have natural connections with the AI era. As AI technology continues to advance, spatial intelligence is viewed as an important paradigm, and when applied practically, AI maps are positioned at the forefront of this trend.

AutoNavi is leading the charge. On August 4th, AutoNavi, a subsidiary of Alibaba Group, announced the completion of comprehensive AI transformation and launched what it defines as the "world's first AI-native map application" - AutoNavi Map 2025 version.

Previously, AutoNavi had used AI for point-by-point optimization in navigation, positioning, and search functions. Now, with comprehensive AI transformation of its product format, AutoNavi demonstrates its full commitment to embracing AI.

AutoNavi told industry observers that the comprehensive AI transformation aims to meet users' increasingly complex search demands. More importantly, this AI reconstruction and introduction of intelligent agents could bring AutoNavi greater commercial imagination.

Over the years, AutoNavi has evolved from a simple mapping tool to a comprehensive life service platform with complete "search, discovery, transaction, and fulfillment" capabilities, becoming a super app with over 800 million monthly active users. However, it still faces commercialization pressures.

Despite attempting various monetization approaches including advertising, ride-hailing, and local life services, AutoNavi has struggled to change its "use-and-go" tool-like nature. Now, as "DeepSeek and others" once again elevate AI possibilities, whether spatial intelligence will become AutoNavi's new breakthrough remains to be validated by time.

**From Tool to Intelligent Agent: AutoNavi Map Reconstructs Itself**

In July this year, at the 2025 YC Global Entrepreneur Summit, AI pioneer Professor Fei-Fei Li stated directly: "Without spatial intelligence, AGI (Artificial General Intelligence) is incomplete."

The essence of AGI lies in large models or intelligent agents being able to perceive, understand, and act, while spatial intelligence achieves these capabilities within three-dimensional space. Compared to language models that give AI expressive abilities, spatial intelligence focuses more on action, providing AutoNavi with prerequisites for embracing AI.

As AutoNavi Map CEO Guo Ning noted, over these twenty years, AutoNavi has accumulated dynamic cognitive understanding of people, vehicles, roads, and stores in spatiotemporal contexts.

This comprehensive transformation to AI-native application is not a simple upgrade, but a reconstruction of itself. In terms of specific functionality, the intelligent agent "Teacher Xiao Gao" is undoubtedly the biggest highlight of this update. It's not just a voice assistant, but an intelligent companion with natural language interaction capabilities that can understand user needs and provide personalized solutions.

During testing, when asking Teacher Xiao Gao to plan a route for "eating, drinking, and beautiful citywalking" in Beijing's Dongsi area, within less than a minute, Teacher Xiao Gao planned a two-day route with themes divided into "Hutong Culture and Record Store Exploration" and "Anime Merchandise and Scholarly Journey," including various dining venues like cafes and restaurants.

Additionally, Teacher Xiao Gao provides thoughtful reminders such as "comfortable shoes recommended for hutong walking" and "braised pork restaurants require queuing in the evening, consider off-peak timing."

If Teacher Xiao Gao meets users' fixed travel needs, then functions like "AI Instant" and "AI Search" satisfy users' immediate travel requirements, with the latter testing applications' proactive thinking abilities.

The "AI Instant" function allows AutoNavi to automatically recommend services or destinations users are most likely to need next, based on current time, location, and habits. For example, office workers might be recommended coffee shops, while self-driving tourists might be recommended gas stations.

AI Instant is based on a "time progression + spatial evolution" dual-axis ranking model, integrating current spatiotemporal location to accurately predict users' immediate travel needs and proactively plan next steps.

Meanwhile, the "AI Search" function, upgraded from the original "Nearby" feature, focuses on recommending "beyond expectations" potential destinations within broader spatiotemporal dimensions. For instance, when opening AI Search, AutoNavi displays a "Beverages to Start an Efficient Morning" section based on morning coffee habits, recommending nearby coffee shops.

Essentially, whether it's Teacher Xiao Gao, AI Instant, or AI Search, all make AutoNavi more vivid and enriched as a navigation tool, capturing more potential user needs and increasing daily usage rates.

The core of comprehensive AI transformation lies in converting AutoNavi Map from a static data warehouse into a "living" intelligent agent with soul. In this AI transformation wave, AutoNavi chose not to make minor adjustments but to initiate thorough reconstruction of itself.

Deconstructing the complex real world is crucial on the path toward spatial intelligence.

**Comprehensive AI Transformation: AutoNavi Proactively Matches User Needs**

For a long time, AutoNavi was an obedient map navigation tool, with Lin Chi-ling's voice package gently saying "Although the road is congested, you are on the optimal route." It would dutifully accompany users to their destinations, but only to destinations.

AutoNavi wouldn't consider whether destinations were open; it would simply direct users wherever they searched, more like a point-and-go tool.

Now, on the path toward AGI, spatial intelligence differs fundamentally from language intelligence. While the former emphasizes dialogue and thinking, the latter stresses perception and decision-making in three-dimensional spacetime, placing higher demands on AI map functionality - from passive response to proactive thinking.

Regarding the difference between proactive and passive approaches, in April this year, AutoNavi's technical leader made a vivid comparison: "Traditional navigation is like acting from a script, mechanically executing preset routes; while our intelligent agent is more like an experienced 'veteran driver' who can perceive road conditions in real-time, predict risks, and proactively adjust strategies."

In May 2024, when road surface collapse occurred on Guangdong's Meida Expressway, AutoNavi detected multiple vehicles simultaneously dropping to zero speed in a short time, identified the risk, and issued "vehicles ahead are braking suddenly" warnings to following vehicles.

This transformation from passive response to proactive matching is strongly driven by large model technology. From a technical architecture perspective, AutoNavi adopts a "master-slave" intelligent agent coordination mechanism in its product architecture, deeply integrating Qianwen large models with spatial scenarios.

In this architecture, Teacher Xiao Gao serves as the "master intelligent agent" like an intelligent butler, receiving user needs. Based on processed perceptual information, the master agent can proactively conduct multi-round cycles of reasoning, planning, execution, and reflection.

More importantly, after initial analysis and sub-task breakdown, the master agent distributes sub-tasks to corresponding "slave agents." Slave agents function like domain experts - some responsible for route planning, others for finding attractions, etc.

Finally, the master agent integrates expertise from all sources, summarizing a feasible travel solution to meet users' complex needs.

This seemingly simple integration of search and recommendation distribution systems requires precise tool chain invocation and effective user service, which depends on data nourishment.

For AutoNavi, this unique data consists of spatiotemporal data closely connected to the real world. Over the years, AutoNavi has accumulated user data covering travel, life services, spatial information, and multiple other domains. This data serves as a vast treasure trove, providing rich nutrients for AutoNavi Map's AI transformation and representing AutoNavi's greatest confidence in spatial intelligence layout.

Compared to others, AutoNavi has been on the spatial intelligence path for many years. Over the past twenty years, AutoNavi was not only among the first map companies to obtain Grade A surveying and mapping qualifications but also established cooperation with domestic highway management departments. Most importantly, data cultivated by 1 billion users provided AutoNavi with nutrients for risk capture and risk assessment.

Of course, spatial intelligence is a relatively broad track, and AutoNavi's ambitions extend beyond this. With the advancement of the "AMAP-AI Inside" strategy, AutoNavi aims to enable spatial intelligence to support comprehensive leaps for AutoNavi APP, automotive navigation, smart glasses, embodied intelligence, and more.

AutoNavi introduced that under this open strategy, multiple partners in smart glasses, automotive, robotics, and low-altitude flight domains have already integrated AutoNavi's spatial intelligence.

Predictably, when maps no longer simply connect point A to point B but truly understand physical world operating principles, AutoNavi seems to have found a path toward AGI.

**AutoNavi's Local Life Service Ambitions: How Much Can AI Help?**

Entering this year, DeepSeek sparked a large model application implementation climax, and AutoNavi's AI embrace accelerated. Starting in January, AutoNavi began internal redesign preparations and project initiation; in April, AutoNavi launched the world's first map-based AI navigation intelligent agent (NaviAgent).

In February, during Alibaba's Q3 2025 earnings call, Alibaba executives revealed that AutoNavi achieved profitability for the first time. Almost simultaneously, according to QuestMobile reports, as of March 2025, AutoNavi Map held 873 million monthly active users, ranking only behind WeChat, Taobao, and Alipay.

Profitability significance for AutoNavi cannot be overstated, but it hasn't changed AutoNavi's position within Alibaba. In June, with explosive growth in Alibaba's instant retail business "Taobao Flash Purchase," Alibaba integrated internal resources, formally incorporating Ele.me and Fliggy into Alibaba's China e-commerce business group. Both businesses maintain independent company operations while jointly promoting Alibaba's upgrade toward a "major consumption platform."

This means that originally, local life group members "Fliggy and AutoNavi" are moving toward higher-frequency consumption scenarios, leaving only AutoNavi outside.

Actually, over these ten-plus years, AutoNavi has done much consumption-related work. Besides aggregated ride-hailing, local life services represent AutoNavi's most attempted domain.

In March 2023, AutoNavi and Koubei formally merged, with all of Alibaba's local life to-store businesses unified under AutoNavi Map's entrance. The same month, AutoNavi launched Starbucks "curbside pickup" functionality, allowing AutoNavi self-driving users to order Starbucks through AutoNavi and pick up coffee curbside without getting out of their cars when passing stores.

Unfortunately, such seemingly consumer-connected functionality showed limited effectiveness under AutoNavi's tool-like nature. This function had insufficient integration with AutoNavi Map's core navigation business, failing to fully integrate into users' daily travel scenarios, resulting in low user participation and lukewarm response.

Holding rich user traffic for navigation needs, but how to tightly combine abundant life services with core navigation functions, allowing users to naturally generate consumption behavior while using navigation, represents an urgent problem for AutoNavi to solve.

This perhaps represents the underlying logic of AutoNavi's full AI embrace - AutoNavi seeks clearer commercialization pathways for itself.

Upgraded AutoNavi, with "Teacher Xiao Gao" as core intelligent agent, organically connects navigation, routes, ticketing, restaurants, and other services, forming a complete travel-life service closed loop. Within this closed loop, every user need can be satisfied, while AutoNavi has opportunities to obtain merchant commissions from various transaction links, maximizing commercial value.

Of course, AutoNavi is not alone in local life services but faces intense competition from strong rivals like Meituan, TikTok, and JD.com.

Meituan, as traditional local life service giant, has accumulated deep user bases and rich merchant resources in dining, food delivery, hotel booking, and other domains through years of cultivation. Cross-industry player TikTok leverages its powerful short video content ecosystem and massive user traffic to rapidly enter local life service markets through content seeding and live commerce, attracting numerous young users and small-medium merchants. JD.com, relying on its powerful e-commerce logistics system and brand influence, is gradually emerging in local life services.

Under such competitive landscape, for AutoNavi to stand out in local life services, it must fully leverage its advantages, promoting AI transformation from "dialogue tool" to "action partner," using intelligent agents to help users find optimal spatial intelligence solutions.

From "carrying clothing, food, housing, and transportation on one map" to today's "living map understanding user needs," AutoNavi has continuously sought pathways for itself.

The AI era has arrived, and AutoNavi's opportunities have come as well.