AI Agents Lab · Shanghai Pilot

Co-building usable AI Agents for social good.

A 4-6 week lab for youth social impact teams, education programs, and AI engineers to build runnable agents on Cloudflare AI application infrastructure.

Explore partnership Technical fit

Why this matters

Social impact teams are ready for a safe path from AI demos to deployed agents.

Nonprofits, student teams, and education programs are already experimenting with AI. This lab turns that curiosity into project-based learning: each participant brings a real problem and leaves with an agent they can run, adapt, and keep using.

01

For youth leaders

Student organizers and youth project leads can use agents for meeting notes, task breakdowns, and activity retrospectives, freeing weekly coordination time for higher-leverage work.

02

For social impact organizations

NGOs and social enterprises can unlock capacity in volunteer matching, content production, impact reporting, and knowledge retrieval, leaving more room for judgment-heavy work.

03

For AI engineers

Technical young builders can practice AI Gateway, RAG, and agent architecture in real settings, with responsible deployment included from the first line of code.

Who We Are

We come from GSC Shanghai II Hub.

GSC Shanghai II Hub is part of the World Economic Forum's Global Shapers Community, a Shanghai-based community of young changemakers across social impact, education, technology, entrepreneurship, and corporate social responsibility. We bring urban issues and social innovation projects into real-world settings.

2050 New Life Forum stage discussion with speakers seated on stage

2050 New Life Forum

Shanghai II Hub members helped connect youth innovation, sustainability, and new-lifestyle topics at the 2050 conference, bringing urban issues into public discussion.

Source
Tomás Saraceno presenting at an Aerocene project showcase

Davos Culture Leader Interview

Global Shapers @ Shanghai II Hub supported WEF art-project interviews, placing culture, technology, and public issues into a broader media context.

Source
Blue network-themed event image for How to Lead an Interview

Shaper Training Workshop

Shanghai II Hub Global Shapers joined cross-Hub skill training and shared methods for interviews, content creation, and facilitation in small workshops.

Source

Why Cloudflare · CF Fit

Cloudflare is the application layer that lets agents actually run.

Claude, GPT, and Gemini solve model access. Cloudflare solves the application layer around models: unified calls, cost visibility, rate limits, caching, knowledge retrieval, edge deployment, and security. That layer is exactly where social impact teams need a partner.

Two-color risograph illustration on deep blue paper with warm orange stairs, grids, and circles suggesting layered infrastructure
Infrastructure as foundation
AI GatewayUnified model gateway
One endpoint for all models, with usage, cost, logs, caching, and fallback models visible from one dashboard. Useful from day one.
Workers AIHosted AI inference
Transparent hosted inference pricing gives prototype teams a practical starting point.
Agents SDKAgent development toolkit
Memory, scheduling, and tool calling turn promising prompts into product-shaped agents.
Vectorize · AI SearchVector database / AI search
RAG over NGO SOPs, training materials, and case documents is one of the fastest social-good use cases to prove.
Workers · Pages · R2 · D1Edge runtime and storage
Edge functions, frontend hosting, object storage, and SQL make launch achievable for small teams.
Zero Trust · WAFSecurity protection
Volunteer data, donor information, and youth-protection scenarios need strong defaults. Cloudflare provides that safety baseline.

What Cloudflare Gets

Cloudflare gets numbers, cases, and stories.

The program is designed around that outcome. After one cohort, Cloudflare can take away market narrative, DevRel usage signals, and Greater China impact assets.

~30builders

Around 30 builders using AI Gateway, Workers AI, and Vectorize in real projects, producing usage data DevRel can inspect.

5case studies

Social impact and education offer strong new storytelling ground for Cloudflare. These cases can move directly into blogs, events, and partnership decks.

1template

An open-source Social Good Agent starter template that other regions and partners can fork, extending the impact beyond Shanghai.

Prototype Menu

The first cohort can produce five types of social-good and education agents.

Each team brings a real problem. By graduation, they ship a runnable prototype and present it live to Cloudflare, GSC, NGOs, universities, and partner organizations at Demo Day.

  1. P-01

    Volunteer / Community Inbox Agent

    Volunteer / Community Inbox Agent

    Route volunteer emails, signup forms, and community help requests into one inbox. Credentials stay in the infrastructure layer while the agent uses a scoped toolbox to classify, deduplicate, draft replies, and suggest routing.

    Toolbox · Inbox
  2. P-02

    Impact Report Agent

    Impact Report Agent

    Turn activity notes, photos, and data into a reviewable impact-report draft, compressing weeks of coordination into one afternoon.

    Reporting
  3. P-03

    Student Leadership Agent

    Student Leadership Agent

    Act as a meeting note-taker, task breakdown assistant, and retrospective coach for student organization leads. Leadership training becomes a product capability.

    Youth
  4. P-04

    NGO Knowledge Agent

    NGO Knowledge Agent

    Convert SOPs, training materials, and case documents into a searchable knowledge base. RAG is one of the Cloudflare stack's strongest fits.

    RAG · Vectorize
  5. P-05

    AI Safety Checklist Agent

    AI Safety Checklist Agent

    Run every agent through privacy, data, ethics, and youth-protection checks before launch. Responsible AI becomes a default workflow.

    Responsible AI

Program Design

A curated small cohort, 4-6 weeks, from real problem to demo-ready prototype.

The first cohort is curated around people, problem quality, and delivery potential. Scale can follow after the first cycle.

  1. Recruitment and selection

    Select 20-30 participants from GSC, partner universities, NGOs, developer communities, and corporate CSR teams.

  2. Workshop sprint

    Four hands-on workshops: agent architecture, RAG, AI Gateway operations, and responsible deployment.

  3. Prototype development

    Teams iterate with mentors and run weekly reviews across engineering, product, and impact.

  4. Demo Day

    Invite Cloudflare, GSC, NGOs, universities, and corporate CSR / ESG partners. Publish a bilingual recap after the event.

Partnership Ask

We are asking Cloudflare for a few concrete things.

We want to run Cloudflare AI application infrastructure through a real social-impact pilot. Turning the pilot into a product-shaped experience takes a few practical forms of support.

Technology Partner Supporting Partner Community Co-host

  1. Technical mentors

    Invite 1-2 Cloudflare engineers or solutions architects to join a workshop and Demo Day. The time commitment is light, with travel shaped around availability.

  2. Developer credits

    Trial credits for AI Gateway, Workers AI, R2, D1, and Vectorize, enough for 20-30 participants to run through the full prototype cycle.

  3. Demo template

    Co-create an open-source Social Good Agent starter template that other regions and partners can fork directly.

  4. Public storytelling

    Co-author a bilingual recap and impact story after the program, ready for the Cloudflare blog, GSC channels, and partner media.

Next Step

Can we schedule a 30-min conversation?

We would like to speak with Cloudflare's community partnership, developer relations, marketing, or AI platform team and find the best collaboration path.

Send partnership brief
Lead
RUOFAN WANG
Role
GSC Shanghai II Hub · Project Contributor
Focus
AI engineering · Agent implementation · Education and social impact
Email
w15365132161@gmail.com
WeChat
langteh