The system should not rot.
Interface
The product should keep working when life gets complicated, not collapse because the user stopped maintaining it for three days.
Vision
The AI is the operator. The human is the executive.
An AI-operated planning system for people who are done maintaining their own planning system.
AI is the interface. Conversation becomes the fastest control surface for planning, replanning, and negotiating trade-offs.
That only works if the assistant can operate a real system with explicit actions, durable state, and a native UI designed for humans.
Lorvex is that system. Your assistant carries more of the planning burden. You keep judgment, taste, and final responsibility.
The right mental model is not a smarter todo list. It is a chief of staff for your commitments.
Planning is the product.
Problem
The hard part is not storing tasks. It is balancing deadlines, meetings, energy, dependencies, and backlog pressure.
The app still matters.
Surface
AI can operate the system, but humans still need a native, legible place to scan, edit, and stay oriented.
What changes when the human stops being the operator
Most productivity apps assume the human is also the scheduler, triage engine, and maintenance worker. That is why so many systems rot as soon as life gets busy.
In Lorvex, your assistant can capture, reorganize, schedule, and revise. You stay in the executive role: set direction, negotiate constraints, and decide when trade-offs are worth it.
Free chat becomes the write interface. The assistant translates that conversation into explicit operations so the system stays coherent as the day moves.
The result is a simpler human surface. You open the app to see the plan, not to assemble it from scratch.
You: "Plan today. I have meetings 10-11 and 15-16. Deadline Friday."
Assistant:
Schedule
09:00-09:45 Draft keynote opening (45m)
11:15-11:25 Reply to venue (10m)
13:30-15:00 Protected focus block (90m)
16:15-16:45 Move design review to tomorrow
You: "I'm low-energy. Keep only two items."
Assistant:
Updated: kept the deadline-critical work, moved the rest.
Planning is the hard part
Planning is not picking items from a list. It is balancing a field of constraints: deadlines across projects, calendar collisions, energy, dependencies, and the work you keep quietly avoiding.
That is where AI changes the experience. Instead of asking you to constantly maintain the system, Lorvex lets the system absorb more of the planning work.
With global context, an assistant can explain why it moved work earlier, why a day is overloaded, or why a task keeps being deferred.
Examples
1) "Wednesday is overloaded. I moved two tasks to Monday while your afternoon is still open."
2) "You only have about 90 usable minutes tomorrow. Do you want the keynote work protected there?"
3) "You've deferred this six times. Archive it, or decide the blocker."
Open workflow, real operations
When AI is the interface, tool capability becomes product capability. Lorvex exposes explicit actions so assistants can do real work instead of stopping at good-looking prose.
That matters both inside Lorvex and across broader AI workflows. Your assistant should be able to use Lorvex alongside the rest of your tools in one session.
Lorvex uses MCP for that interoperability. The point is not the protocol itself. The point is that the system stays open and composable.
Why a native app still matters
AI-native does not mean UI-light. Humans still need a place that is fast to read, calm to live in, and direct to edit.
Chat is the quickest way to express intent. The app is the quickest way to absorb the answer. Both surfaces matter, and both should feel deliberate.
Lorvex is built as a native app because this is daily-use software, not a disposable wrapper around a prompt box.
Durable context
Most assistants reset between sessions. Lorvex gives them durable state to work from, so the system remembers your projects, patterns, history, and real constraints.
Over time, planning state becomes more than a backlog. It becomes a record of what you did, what you avoided, what repeatedly slipped, and what actually fit into a real day.
That continuity is what makes AI planning compound instead of restarting from zero every time.
Principles
- Executive, not operator
- The human should not carry the full maintenance burden of the system. The assistant carries more of the operational load; the human keeps judgment and taste.
- Planning, not storage
- The product is not a prettier place to store tasks. Its job is to turn real constraints into a workable plan.
- Simplicity is earned
- The interface gets simpler only because AI absorbed more of the cognitive work. Minimalism is the consequence, not the goal.
- Reasoning must stay legible
- AI output should come with reasons a human can judge, and corrections should stay cheap when the AI gets intent wrong.
- Open workflow, owned continuity
- The system should remain open to broader AI workflows while preserving durable, user-owned context across sessions.
- Native software still matters
- Even when AI is the interface, the human-facing app should stay fast, calm, and worth living in every day.
Project
Lorvex is open source. Issues are contributions.
This project is built in an AI-coding loop. All implementation is written with Codex + Claude Code, guided by human design direction and feedback.
If you have an idea, file a GitHub issue. High-signal, well-scoped issues are treated as implementation requests and pushed through a fast path: issue -> PR -> code, with unusually short distance between feedback and shipped work.