2026 Global Open-source AI Challenge (GOAI)「Boundless Agents」Track Open for Registration

Bringing AI Agents into Real-World User Scenarios

On July 21, 2026, the first Global Open-source AI Challenge (GOAI) officially kicked off at the Hangzhou Cloud Valley Center. As one of the core competition tracks, the Boundless Agents Track targets real industrial scenarios and encourages participating teams to develop operable, demonstrable, and replicable agent applications.

At the launch ceremony, guests from academia, industry and open-source communities reached a consensus: laboratory technological breakthroughs are merely a starting point. Only when AI moves beyond demo environments and is deployed in real-world scenarios can it truly create value across industries. The launch of the Boundless Agents Track directly addresses this core industry demand.

Track Background: Why the Boundless Agents Track Matters

Currently, AI agents are rapidly evolving from general-purpose conversational tools into industry-specific applications, accompanied by prominent challenges in industrial implementation. A financial risk control Agent must not only comprehend complex business rules, but also connect multiple internal systems, comply with compliance workflows, and output auditable decision-making basis. An on-site industrial Agent needs stable operation under limited computing power and to deliver reliable judgments amid noisy equipment data. An intelligent cockpit Agent requires multi-modal interaction integrating voice, vision and touch to deliver low-latency, highly secure user experience in driving scenarios.

All these challenges lead to a clear conclusion: for AI Agents to evolve from "capable of chatting" to "capable of getting work done", only stronger general capabilities are insufficient. What matters is in-depth understanding of industrial scenarios and precise adaptation to real business workflows. This constitutes the core logic of the Boundless Agents Track: to drive AI Agents out of chat windows and into real industrial sites across all sectors, demonstrating and creating tangible value through solving concrete problems.

Track Positioning: Innovative AI Agent Applications for Real Industrial Scenarios

The Boundless Agents Track focuses on the implementation of AI Agents in real industrial scenarios. Teams may select use cases from high-frequency fields including personal terminals, automotive mobility, financial services, education and industrial manufacturing, or propose more segmented application scenarios based on authentic user demands, and build agent applications equipped with task comprehension, workflow orchestration, tool invocation, knowledge enhancement, multi-turn interaction and result delivery capabilities. Submissions with multi-modal features are particularly welcome.

This track discourages generic chatbots, single-point Q&A tools or simple content generators. Instead, it prioritizes entries built around real users, authentic business workflows and closed task loops, to verify how AI Agents can be deployed in industrial sites, boost efficiency, optimize user experience, and form viable application prototypes with lasting practical value.

In short, the Boundless Agents Track does not seek showcases of "what large models can theoretically achieve", but solid solutions to practical industry problems solvable by AI Agents.

Core Requirements: Prioritize Closed Task Loops Over Functional Quantity

This track centers on closed task loops rather than the sheer number of functions. Entries shall focus on high-frequency scenarios across five verticals: AI + Smart Glasses, AI + Automotive, AI + Finance, AI + Education, AI + Industrial Manufacturing. Teams are encouraged to integrate large models, knowledge bases, tool calls, multi-modal interaction and workflow orchestration into practical user-facing products. Competitive, usable entries must clearly answer core questions: Who is the end user? What pain points do users face? How does the Agent interpret demands, invoke capabilities, execute workflows and deliver outcomes? Judges expect a complete end-to-end pipeline spanning user input, task comprehension, plan execution, tool/knowledge invocation, result delivery, exception handling and outcome validation.

Teams that enable AI to reliably "complete real work" in authentic scenarios --- accurately interpreting demands, fully executing workflows, delivering dependable results and properly handling exceptions --- will build industry-ready Agents that align with GOAI's core pursuit.

Eligible Participants: Developers "Capable of Deploying Agents to Real-World Sites"

The Boundless Agents Track is open to global developers, startup teams, university students, research institutions, corporate technical teams, product managers and open-source community contributors. Eligible participants include AI engineers, industry-savvy operation managers, full-stack developers, UE designers, academic researchers with published papers, and startup founders with demos. Regardless of background, any participant focused on deploying AI Agents to deliver tangible industrial outcomes is welcome.

The track highly expects teams with "deep domain expertise". You do not need state-of-the-art models, but you must identify critical user pain points; you do not need overly complex architectures, but you must deliver fully functional workflows and valid results. Teams proficient in smart glasses, automotive, finance, education or industrial manufacturing, capable of building complete closed loops from interaction design to outcome validation, are exactly who this track targets. Participants may register individually or as teams, and cross-disciplinary collaboration is strongly encouraged. Detailed team formation rules shall follow the official competition website and participant handbook.

Candidates with open-source contribution experience, polished product presentation skills or technical documentation writing capabilities will receive extra evaluation credit. GOAI believes that the developers who successfully deploy agents to real sites often are not in the spotlight but work behind the scenes within core industries --- we are looking for you.

Submission Requirements: Feasibility & Practical Deployability

Covering five verticals --- AI + Smart Glasses, AI + Automotive, AI + Finance, AI + Education, AI + Industrial Manufacturing --- this track connects real daily and workplace scenarios. All entries must deliver at least one demonstrable, verifiable closed task loop, with clear explanations of target users, scenario pain points, interaction pipelines, technical roadmaps, data sources, compliance boundaries and follow-up iteration plans.

Specifically, entries shall demonstrate the following core capabilities:

  • I. Operability: Functional demos, videos or other verifiable materials;

  • II. Explainability: Clear traceability of results and rationale behind recommendations for end users;

  • III. Controllability: Prominent alerts and human confirmation mechanisms for missing data, failed tool calls and high-risk operations.

From "building standalone AI functions" to "delivering complete end-to-end services" --- the Boundless Agents Track targets AI Agents that move beyond demo showcases and fully integrate into daily life, work and industrial production.

Evaluation Criteria: Measured by "Real Problems, Authentic Scenarios & Complete Closed Loops"

To align with the evaluation philosophy of "clear scenarios, smooth workflows, complete evidence and full closed loops", six weighted evaluation dimensions are set as follows:

  • Industrial Scenario Value: 25%

  • Agent Capabilities & Closed Task Loop: 25%

  • Product Experience & Demo Completion: 20%

  • Depth of Technical Implementation: 15%

  • Security, Compliance & Traceability: 10%

  • Open-source & Reusability Contribution: 5%

The weighting structure highlights that evaluation prioritizes genuine problems and well-grounded scenarios as well as the Agent's ability to really resolve problems and fully complete end-to-end workflows --- these two dimensions serve as the core threshold for entries of the Boundless Agents Track. Valid scenarios go before technical sophistication; credible problem framing precedes implementation complexity. A small closed-loop solution with clear scenarios, smooth workflows and complete supporting evidence outperforms bloated, barely functional all-in-one platforms.

Schedule & Awards: Progressive Path from Preliminary to Final Rounds

The competition registration opened on July 15. Participating teams may register via the official website, and the deadline for preliminary round submissions is August 16; the list of semi-final shortlisted teams will be released on August 24, with the Top 30 teams advancing to the semi-finals; the deadline for semi-final submissions is September 3; the final shortlist will be announced on September 10, and the Top 15 teams will qualify for the finals; offline final defenses and demonstrations will be held on September 22; the award ceremony, project showcase and ecosystem matchmaking will take place on GOAI DAY, September 23. All schedules are subject to final official notices issued by the organizing committee.

Evaluation of preliminary round focuses on solution design and feasibility. Teams must elaborate their target users, core scenario pain points, Agent operation logic, coordination between data and tools, compliance boundaries and long-term deployment roadmaps. Upon advancing to semi-finals and finals, entries undergo rigorous engineering verification. Functional demos, code repositories or equivalent verifiable engineering materials, clear deployment guides, credible data sources and authentic runtime evidence become core metrics for judges to assess real project value.

In terms of awards, the cash prizes for the track's champion, first runner-up and second runner-up are RMB 500,000.00, RMB 300,000.00 and RMB 100,000.00 respectively; outstanding projects will also be eligible to compete for the RMB 1,000,000.00 overall GOAI Grand Prize.

Closing Remarks: Shift AI Agents from "Demo-Ready" to "Delivery-Ready"

Leveraging the GOAI platform, this track aims to collaborate with global developers to address a core question: What tangible value can AI Agents deliver in industrial sites? How can their value be validated? What constitutes a complete, deliverable solution? When developers universally adopt these standards to measure Agent performance, the era where AI moves beyond "peak performance in demos" to "delivery as the starting benchmark" will truly arrive.

This is the transformation that the GOAI Boundless Agents Track strives to drive: enabling AI Agents to transcend flashy demo-only performance, embed seamlessly into daily user scenarios, and evolve into reliable, trustworthy, deliverable collaborative power.

Registration is now open. Visit the official website goaihz.com to sign up, download the track participant handbook and view full challenge details.

Long press to scan the QR code and sign up instantly

Total prize pool of 5 million RMB awaits you （Competition Registration QR Code） Scan the QR Code to add GOAI Assistant

Your exclusive competition guide is online （GOAI Assistant QR Code）