WheelWright — Stop reinventing the wheel with every AI session

The cycle you're stuck in

You've been here before

Without WAI

Every session is Groundhog Day

You burn tokens re-explaining context that the AI already knew five minutes ago — in a different tab.

Re-describe your architecture, stack, and decisions every session
Switch LLMs and start from a blank slate
Move from IDE to chat to CLI — none share a brain
Shift focus from backend to frontend and lose all prior context
Rework code the AI already got right last time

With WAI

Context rolls forward

Drop a WAI-Spoke/ folder in your project. Every AI session picks up where the last one left off.

Architecture, decisions, and state persist across sessions
Switch between Claude, ChatGPT, Copilot, Grok — all aligned
IDE, chat, and CLI all read the same memory
Shift focus freely — your context travels with you
Zero rework. Progress only moves forward.

The six variants

Six ways alignment breaks

Each time agent, session, tool, focus, decision tree, or history changes, you restart from scratch. MAP keeps them all aligned simultaneously.

⟳

Agent

Claude in the morning, ChatGPT at night — neither remembers.

◷

Session

Close a tab and your decisions evaporate with it.

⬡

Tool

Your IDE, your chat window, your CLI — all separate brains.

◎

Focus

Switch from backend to design and context resets entirely.

🌳

Decision tree

You explore branches, prune paths, and still end up re-hashing the same options because the model forgets which routes you already ruled out.

⏳

Lossy remembrance

Long sessions push early decisions out of view. Nuance, trade-offs, and "why we chose this" slowly disappear from the conversation.

The protocol

MAP — Multi-Variant Alignment Protocol

WAI treats all of it — agent, session, tool, focus, history — as one problem: context continuity. When anything changes, your project data stays put. No restarts.

MAP is file-based and LLM-agnostic. No database. No server. No vendor lock-in. Just structured files that any AI can read and any developer can own.

Two products for two kinds of work — thinking and building. Both speak the same protocol, so whichever you pick, nothing is wasted if you later want the other.

Start with Tracks — free, no install

Two products, one protocol

Which kind of work is yours?

Tracks is for thinkers — where the reasoning is the artifact. The Harness is for builders — where the system has to keep running. Same protocol underneath, so neither locks you out of the other.

For thinkers

WAI Tracks

Portable context files that teleport your working state to any AI, any session, any tool. Stop repeating yourself in under five minutes.

One file captures state, decisions, and next steps
Works with any LLM — no vendor commitment
Mid-session checkpointing by updating a single Track file
No install, no config — drop a file and go

Explore Tracks

For builders

WAI Harness

Your own hub-and-spoke wheel. The hub remembers across every project; each spoke does the work and sends back what it learned. We give you the blueprint — your agent builds it.

Hub-first: inbox, capabilities graph, taste, and registry
Your agent builds your wheel from the blueprint — no account, no server
Cut from the harness we run daily, so it cannot drift into a brochure
PEV execution: Perceive / Execute / Verify

Explore the Harness

Not a ladder — two jobs. Take Tracks when the thinking is the work, the Harness when the system is. And they fit together: Tracks is the transport layer the Harness runs on, so nothing you learn in one is wasted in the other.

Origin

Born from necessity

WheelWright wasn't designed in a lab. It started as a way for me — Mario Vaccari — to stop losing the thread every time I opened a new AI session.

I'm a systems-minded, solutions-driven product manager learning AI by building with it hands-on. As I hit the same pitfalls while wiring AI into my own work, I built WheelWright first as a session continuity fix.

Pushing it further, it became clear the same protocol could take on more and more of the delegatable work. I've always believed that if you enjoy what you do, you care how you do it — that principle has shaped my entire career, and it's baked into WAI.