ShanghAI - QwenWork
Hey Lobsters, welcome to this new edition of ShanghAI.
Every founder who saw QwenWork's pricing page did the same thing: pulled up Claude Cowork, saw $20 next to $20, and closed the tab. Wrong move. That's the most expensive tab-close of the quarter.
Here's the trick nobody caught: Claude Cowork is a coworker who clocks out the second you shut the laptop. Artifacts, Computer Use, Claude Desktop — all session-bound.
Close the chat, kill the client, log out: the agent stops mid-thought.
Want it running at 3am scraping competitor pricing? You're now the DevOps team, wiring the Claude API into a Lambda function you own, patch, and pay for on top of the $20.
QwenWork shipped that Lambda function as a $0 line item inside the same $20.
Cloud-Native Scheduled Tasks run on Alibaba's own backend and keep firing after your laptop's asleep — with an isolated audit log per run, the same audit trail Zapier gates behind its higher tiers.
Add native full-stack hosting — custom domain, built-in database, password gate, zero Vercel bill — and the math flips: you're not paying $20/mo for an agent.
You're paying $20/mo for the agent, the orchestrator, and the hosting layer Cowork users are still sourcing at $50–$170/mo extra.
OUR PARTNER
THE RUNDOWN AI
Learn AI in 5 minutes a day
You don't have to scroll every AI thread, track every new tool, or watch every demo.
The Rundown AI breaks it all down for you — the latest AI news, tools, and tutorials in one free 5-minute email every morning.
Trusted by 2M+ professionals at Apple, Google, and NASA.
HOW IT WORKS
THE ALL-IN ONE AUTONOMOUS WORKSPACE
QwenWork is Alibaba’s unified agent environment built directly on Alibaba Cloud’s serverless infrastructure.
Western platforms sell a conversational model and leave you to figure out hosting, triggers, and state storage. QwenWork collapses all three into a single runtime.
The platform operates across three native capabilities:
First, the Persistent Execution Engine runs tasks as asynchronous cloud processes rather than session-bound threads.
When instructed to monitor competitor catalog updates every morning, Alibaba Cloud manages the cron schedule, headless execution, and error handling entirely server-side without an active browser tab.
Second, the Built-in Database and Edge Deployment layer allows any script, dashboard, or data table to publish instantly to a live URL.
It provisions its own relational storage, compiles an access-controlled frontend, and routes traffic through custom domains without connecting to external services like Vercel or Supabase.
Third, the Awareness Engine indexes working habits, brand guidelines, and schema rules directly from past execution runs.
Instead of repeatedly uploading static prompt context files, teams can convert proven workflows into shared organization-wide templates with a single click.
By eliminating external webhook configurations and third-party hosting, QwenWork transforms a fragmented three-vendor automation stack into a self-contained execution node.
SEE ALSO
THE NEW TOOL OF HUBSPOT
Stop losing deals you should be winning.
Bad leads. Slow deals. Objections you didn't see coming.
That's not a sales problem. You're selling to the wrong people.
Most founders never stop to define exactly who their product is actually for. So they pitch everyone, convert few, and wonder why the funnel's broken.
HubSpot for Startups built a free tool that fixes the root cause in 2 minutes. Paste your URL, answer a few quick questions, and get a clear profile of your best-fit customer.
WHAT TO USE
TOP 3 AGENTS PLATFORMS FOR BACKGROUND AUTOMATION - SEPTEMBER 2026
BEST QUALITY/PRICE: QwenWork Plus — $20/month for 2,000 credits, replacing your stack with native cloud scheduling and bundled full-stack hosting.
BEST LOCAL: Dify + Qwen2.5-Coder-32B (Q4) — Runs offline on a single 24GB GPU to deliver unlimited automation loops with zero data leakage and zero API fees.
ShanghAI CHOICE: QwenWork Enterprise Standard — At $30/seat/month, its pooled team credits and centralized admin controls make deploying shared background agents frictionless.
THIS WEEK OFFER (Affiliate): Notion Business - 3 MONTHS FREE
APPLY TO YOUR BUSINESS
KILLING THREE BILLS WITH ONE CREDIT POOL
This workflow audits your current Zapier/Make + Vercel/Supabase + Cowork spend, maps each trigger-action chain to a QwenWork Scheduled Task, and cuts over hosting with zero downtime.
Step 1: Audit & Map the Orchestrator
Export your active scenario list from Zapier or Make. Group workflows into: (1) simple inbound webhooks and (2) interval-based recurring crons (e.g., daily scraping, weekly analytics digests).
Open QwenWork and initialize a dedicated Scheduled Task for each recurring scenario.
Replace multi-step Zapier parsing blocks with a single prompt-instructed Python execution step running natively on Alibaba’s backend. Unlike Zapier, where each sub-step burns task quota, QwenWork executes the entire transformation script as one task run with an isolated execution audit log.
Step 2: Migrate Hosted Front-Ends & Persistent State
For internal portals, client reports, or dashboards currently served on Vercel or Retool: prompt QwenWork to rebuild the frontend directly from your existing JSON data payloads.
Bind the output directly to QwenWork's built-in relational table rather than routing to an external Supabase instance.
Deploy directly to production using QwenWork’s native URL publisher. Add your company subdomain via CNAME and toggle the built-in password/SSO gate. (Plus tier covers up to 10 live sites; Pro covers 50).
Step 3: Configure the Awareness Engine for Persistent Memory
Turn on the Awareness Engine in the desktop client settings. Unlike Claude Projects—which requires re-uploading static markdown context files every session—the Awareness Engine passively indexes your team's code formatting rules, output schemas, and brand tone across runs.
Once a prompt pipeline works reliably in testing, click "Promote to Skill". This locks the chain into an organization-wide reusable module that non-technical team members can invoke without prompting from scratch.
Step 4: Run the Two-Week Burn Audit & Deprecate Legacy Seats
Allocate QwenWork Plus’s 2,000 monthly credits across your migrated workloads. Monitor burn rate for 14 days: scheduled lightweight background tasks consume fractional credits compared to interactive UI generation.
Once credit consumption stabilizes below the 2,000 monthly cap, cancel Zapier/Make paid tiers and downgrade auxiliary cloud databases to free-tier read-only backups.
The Bottom Line: Setup takes 45–60 minutes for teams with under 10 active automations. The ROI isn't the $20 agent seat—it’s the immediate $50–$150/month reduction in SaaS bloat that hits your balance sheet starting next billing cycle.
If your engineering or ops team is ready to move beyond manual prompting, audit your recurring SaaS bloat, or deploy production AI agents at scale:
Until next week,
The ShanghAI Guy
