// AI agent orchestrator
Wazear orchestrates AI agents in a structured pipeline. One builds. Another tears it apart. You get what survives.
$99 once. Free updates on same version. Pay again only when we earn it and you find it's worth it.
// 02 — Features
01
What ships with Wazear
Wazear comes with a default pipeline of eight agents: Planner, Planner Reviewer, Architect, Architect Reviewer, Programmer, Programmer Reviewer, Fixer, and QA. Each has a defined role and a defined place in the sequence. You can modify any of them, remove them, or build entirely new ones. This is just the starting point.
02
A process, not a conversation
Most AI coding tools are one conversation going wherever it goes. Wazear runs a pipeline where each agent reads only what it needs — the plan, the architecture, the code — and produces exactly its part. No agent inherits another's confusion. The structure is the product.
03
Every output can have an opponent
The default pipeline pairs every producing agent with a Reviewer. The Reviewer reads the spec and the output — nothing else. Not the producer's reasoning, not its intent. Just what was asked and what was delivered. That gap is where most mistakes surface. Fewer reach you.
04
Two quality gates, not one
The Producer Reviewer catches issues before the output leaves that stage — think of it as the programmer's team lead. QA is different. It runs after everything is built, takes your screenshots and your observations, adds its own analysis, and hands a full report to the Fixer. Internal review and user acceptance testing. Both, by default.
05
Guardrails you own
Wazear ships with a default list of banned commands. You can add your own. Agents whose job is to think — planners, reviewers, architects — can't run commands at all. Not restricted. Blocked entirely.
06
Your pipeline, your rules
The default pipeline is a starting point. Use it as-is, modify it, strip it down, or build something entirely your own. Agents are prompts with a small config — if you can describe a role, you can create one. No programming required. We are working on a UI-only solution that streamlines this further.
07
You choose your AI, including free options
Wazear runs on Qwen Code. Qwen currently offers a free usage tier when you sign up — check their site for current limits. Need more or something different? Plug in your own API key — OpenAI, Gemini, or OpenRouter, which opens access to virtually every AI provider available.
// 03 — How It Works
Simplicity is at the heart of Wazear. Your pipeline is nothing more than agent definitions running in sequence — each doing one thing and one thing only.
Create a project
Choose where it lives on your disk and write a brief describing what you want to build.
Choose your tech stack
Pick what you want to build with. Don't know? Leave it — the Architect will decide based on your brief.
Choose your agents
Select which agents run in your pipeline. Wazear ships with a roster that simulates a full Software Development Life Cycle — ready to use out of the box.
Start the pipeline
One click. The first agent picks up your brief, does its job, and hands off to the next. You don't manage the handoffs. Wazear does.
The Planner
Takes your brief and turns it into a list of actionable goals. The Planner Reviewer reads it cold — spotted an issue? It goes back to the Planner with feedback until it's right.
The Architect
Lays out the technical architecture and tech stack. Its own Reviewer checks the work. Same loop — issues go back, get fixed, move forward.
The Programmer
Builds your product. The Programmer Reviewer checks the code against the architecture as the criteria. Anything off goes back to the Programmer to fix.
The QA
Runs a final check on everything. You test the app yourself, take screenshots, add notes. All of it gets compiled into a report.
The Fixer
Works through every issue in the QA report. Once done, it loops back to QA automatically. That cycle continues until you're satisfied and mark the pipeline complete.
At any point, you can step in, review any output, and modify it before the next stage continues. Or don't. Your call.
// 04 — Pricing
One time. No subscription. No credit meters. No surprises from us.
$99 once. Free updates on same version. Pay again only when we earn it and you find it's worth it. Pay today, not every day.
$99 is an introductory price that may increase with v2. Buyers of v1 get v2 free — two versions for the price of one.
Mac and Windows versions are coming. Leave your email and we'll notify you when they're ready.
// 05 — FAQ
Not to use Wazear. The default pipeline ships ready to run — you can start a project without touching a single agent file. Creating or customizing agents means editing a markdown file — think of it like writing a structured Notion page — and choosing a few options in the UI. If you can describe a role clearly, you can create an agent for it. We're working on a UI-only solution that streamlines this further.
It will. That's not a flaw in Wazear — it's a reality of working with AI. What Wazear does is put structure around it. Reviewer agents challenge every output before it moves forward. QA gives you a stage to manually test your app, take screenshots, and flag what's wrong — which gets compiled into a report the Fixer works from. You'll surface more mistakes earlier than you would with a single AI session. You won't catch all of them. Gotta catch 'em all is a different product.
Same-version updates are free. If you buy v1, every update within v1 — v1.2, v1.5.8, v1.9.11 — costs you nothing. If you want v2, you buy it separately. Either way, you never pay for anything you don't feel is worth it and we didn't earn.
$99 is an introductory price that may increase with v2. Buyers of v1 get v2 free — two versions for the price of one.
Wazear itself has no external dependencies — it's a local desktop app. However, it currently runs on Qwen Code as its AI engine, and Qwen Code requires Node.js. In practice: install Node.js, then install Qwen Code with one npm command. The app walks you through it on first launch, and a full setup guide is available. Future versions of Wazear may reduce or remove this requirement entirely.
Wazear is fully local. It doesn't connect to any server, doesn't phone home, and collects nothing. It coordinates between agents running on your machine — that's it. That said, those agents are Qwen instances, which means whatever you feed into the pipeline goes to Qwen under their terms. If you use your own API key — OpenAI, Gemini, or OpenRouter — your data goes to that provider under their terms. We don't touch it, intercept it, or see it.
Yes. Wazear currently runs on Qwen Code, which offers a free usage tier — check their site for current details. If you need more capacity or a different model, plug in your own API key. OpenAI, Gemini, and OpenRouter are supported. OpenRouter alone opens access to virtually every AI provider available, including additional free tiers.
Those are single-agent tools. You prompt them, they produce something, you review it, you prompt again. You are the orchestrator — managing context, catching mistakes, deciding what goes where. Wazear is the orchestrator. It runs a structured pipeline where each agent has a defined role, reads only what it needs, and hands off to the next stage automatically. Reviewer agents challenge outputs before they move forward. You step in when you want — not because the process requires you to babysit it. It's also not an IDE. It doesn't replace your editor. It runs alongside whatever tools you already use.
Wazear currently runs on Linux. Mac and Windows support is in development. If you're on Mac or Windows, get in touch via the Contact page or leave your email in the waitlist above.