TellWang
Agent-native PaaS
TellWang turns prompts into apps, APIs, automations, and headless agents, then runs them on a complete, isolated stack.
How it works
You already know the rules — what a booking needs before it's confirmed, when an invoice counts as overdue, who signs off on a refund. What kept that knowledge from becoming software was never the code. It's everything around the code: environments, schema changes, verification, rollback, monitoring. A Wok carries all of that, so plain sentences are enough to start.
“Quotes over $2,000 need a second approval.” Wang turns sentences like that into schema, functions, and interface — on a Wok born as a running app with the whole backend already wired.
Wok provisioned · stack readyWang works on Preview and pushes through git. Browser checks must find real content, function calls must return real answers, and schema rides a migration log. Only the exact verified build promotes to Published.
git push · verified · promotedTelemetry, function logs, client errors, and session replay show how each release behaves for real visitors. A bad one rolls back to the last good release.
Rollback target · last good releaseOne prompt, more than a screen
Start with an outcome. Wang decides which parts belong in an interface, which should run quietly in the background, and how they connect.
Public experiences, private tools, sign-in, records, uploads, payments, and a real domain.
Give an agent tools, knowledge, triggers, and approval rules. Call it by API, webhook, schedule, inbox, or chat.
Webhook received · run #128Schedule reliable jobs, retries, alerts, and delivery checks without stitching together another workflow host.
Next run · tomorrow 08:00Ship authenticated endpoints and MCP tools with authorization enforced at the resource boundary.
POST /mcp · 200 OKBuilt with TellWang · Wheeldo
Wheeldo turns time into a living wheel. TellWang helped turn that unusual interface into a secure, responsive, agent-assisted SaaS product with accounts, private schedules, public booking, invitations, and a safer path from Preview to production.
Found 45 minutes tonight for the license renewal.
The platform underneath
AI builders often stop when the interface looks finished. TellWang builds on a live Wok, where the software and everything it depends on are created, tested, and operated together.
Explore the Wok →Build guides
Each guide gives your coding agent the prompt, platform shape, safety boundaries, and checks needed to finish on a live Preview.
Working from a terminal or a coding agent? npm i -g @lollipopai/cli puts the whole platform on $PATH, and lollipop skill install teaches your agent how to run a Wok — see the setup.
Pricing
The subscription pays for the operating shape. Usage beyond included allowances uses prepaid credits while postpaid billing is disabled.
One live app, its linked Preview, 1M model tokens, DeepSeek Flash in Wang, and one hosted run at a time.
10M included model tokens, Kimi access, custom domains, and prepaid usage beyond the allowance.
Five live apps, four concurrent Wang runs, and prepaid usage beyond the allowance.
Scale plus database standby, failover, and redundant REST.
What should exist next?
Start with an app, an automation, an API, or an agent. The Wok is ready underneath it.