Flion – Agentic super app with plugins, skills and automations
Chat with @ plugins/skills, role agents, browser dock, automations and release gates
Agentic super app for operating AI work in one cockpit: chat with @ plugins/skills, role agents, browser dock, recurring automations, operational memory and release gates.
Context and objective
Agentic super app for operating AI work in one cockpit: chat with @ plugins/skills, role agents, browser dock, recurring automations, operational memory and release gates.
I repositioned Flion as an agentic super app that brings chat, plugins, skills, specialized agents, automations, browser/computer dock, memory and release operations into one workspace. The public demo is sanitized: it shows the product shape, including @ and / command-palette workflows, a capability registry, role agents and a runnable Data Cloud vertical slice that proves MCP-style tools, contracts, an async worker, QA and release evidence without exposing the private runtime, local vault, sessions or credentials.
The challenge
The core challenge here was not just building a polished interface. It was designing a flow that fit the real business context, reduced operational noise and turned a fragmented process into something clearer, faster and more reliable.
That usually means aligning customer service, decision-making, record keeping, automation and follow-up under one coherent logic. In other words: making the product support the operation, not the other way around.
The solution
I repositioned Flion as an agentic super app that brings chat, plugins, skills, specialized agents, automations, browser/computer dock, memory and release operations into one workspace. The public demo is sanitized: it shows the product shape, including @ and / command-palette workflows, a capability registry, role agents and a runnable Data Cloud vertical slice that proves MCP-style tools, contracts, an async worker, QA and release evidence without exposing the private runtime, local vault, sessions or credentials.
Operational chat with @ plugins/skills, / commands, context chips and model control
Public registry representing 21 native plugins and 50+ skills organized by capability status
Role-agent canvas for scope, design, architecture, implementation, QA and release
Browser/computer dock and permission broker modeled for explicit-approval execution
Recurring automations with model selection, run status and auditable handoff
Data Cloud vertical slice with MCP-style tools, an async worker, tests and release gates
Real impact
The super app brings chat, plugins, skills, agents and automations into one cockpit
Permission broker and release gates reduce risk before execution and deployment
Sanitized repo proves agentic architecture without exposing the private runtime
Technology
My responsibilities
- Designed the agentic super-app architecture, including chat, command palette, plugin registry and role agents
- Implemented the sanitized public demo, generated console data and runnable Data Cloud vertical slice
- Built the operator console for fast review by recruiters and technical teams
- Documented public/private boundaries, technical positioning and release gates
Gallery
Links
Final CTA
If you want to build something like this or reorganize an operation that still depends too heavily on manual work, overloaded customer service or a confusing interface, this is exactly the kind of project where I usually create the most value.