Current State
Two years of fixing what breaks and adapting to model advances.
Project
VibeOS
A development system built from the hard parts of AI-assisted software work: lost work, broken repos, drift, review loops, and receipts
The development system behind the rest.
HarnessQuality controlDevelopment systemShowcase
The Problem
What problem was worth solving?
I build things to solve a specific pain, not to decorate a portfolio. This is the pressure that made this project necessary.
AI coding tools can generate code, but the hard part is still managing the work: deciding what should happen next, keeping scope tight, checking quality, and making sure claims are backed by evidence.
That overhead is where the painful lessons live: lost work, broken repositories, over-trusted automation, vague handoffs, and sessions that feel productive until nobody can prove what changed.
VibeOS is the development system behind that idea: clear planning, tight scope, specialist help, quality checks, review loops, and handoff evidence.
The Build
What I built around the problem
VibeOS is the private development system behind Latif's AI-assisted software work, with VibeOS Bootstrap as the public companion. The durable architecture is loop-based: start from real source material, keep each work lane small, use specialist help, run checks, review the output, and carry the learning forward with evidence.
Execution model
Plan · build · test · audit
The useful pattern is work in loops: clarify the task, build the smallest useful piece, test it, review it, then carry the learning forward.
Structured decisions
Rules before claims
Architecture choices, quality gates, compliance checks, and planning decisions are structured before public claims are asserted.
Quality automation
Checks and receipts
Quality checks, validation, architecture enforcement, and evidence capture are treated as part of the work, not optional cleanup.
Built-in knowledge
Reference docs
The harness carries guidance, product requirements, and agent configuration so AI-assisted work stays aligned to the intended outcome.
Reality Check
The harness came from things breaking
The point is not to pretend the work was clean. This is what the project taught, where it changed shape, or where the boundary still matters.
01
What has been hard
VibeOS is not a neat framework invented from theory. It came from lost work, broken repos, bad assumptions, and the repeated experience of watching AI-assisted development drift without enough structure.
02
What had to change
The work needed loops: clarify the task, narrow the lane, build the smallest useful piece, test it, review it, and leave enough evidence that someone else can continue.
03
Current boundary
The public page should talk about the operating method and the lessons it carries, not pretend every private hook, gate, or workflow is a finished product.
Value
What changed because it existed
This is the practical value of the work. Where a project is unfinished, shelved, or still changing, the value is tied to what it actually taught.
How it works
Conversation-first harness
The interface starts from intent, but the system value is the work loop behind it: plan, execute, test, audit, and hand off.
Specialist agents
Specialist lanes
Research, build, test, audit, and documentation work are separated into specialist lanes so responsibilities stay clear.
Product model
Private system + open-source companion
VibeOS has a private working system and a free, open-source Bootstrap framework that shares the basic working model.
Alignment system
Built-in guardrails and session audits
The harness carries product requirements, quality standards, and prompt guidance so AI-assisted work stays aligned to the intended outcome.
Technical Layer
How the system is built
This is the implementation behind the work: the architecture choices, integrations, controls, and workflow decisions that made the project real enough to learn from.
Private development harness
VibeOS Bootstrap public framework
Work-order specs
Specialist work lanes
Quality checks
Audit notes
Decision records
Reference docs
jq + git + python3
Build Story
How the thinking unfolded
This is the reasoning path behind the output, not only the finished artifact.
01
The harness came from real failures
The point of VibeOS is not that AI development becomes effortless. The point is that lost work, broken repos, and vague handoffs make it obvious that AI-assisted development needs a visible loop around the work.
02
The user should not have to manage every detail
The core insight is not a specific inventory count. It is the interaction model: people should be able to describe what they want while the harness imposes planning, checks, and evidence discipline.
03
Autonomy requires structure
The natural-language workflow needs explicit scope, specialist roles, quality checks, decision rules, and reference material. That structure is what turns AI coding assistance into work someone can inspect and continue.
04
Private system, public framework
VibeOS has a private working system and an open-source Bootstrap companion that lets teams adopt the working model without needing the private system.
05
Staying on track is built in
Product requirements, quality standards, and session audits are embedded into the harness so AI-assisted work can be checked before anyone treats it as finished.
Evidence
What this project shows
This evidence is strongest when it is tied to specific work rather than broad claims.
Development-Harness Design
Designed the system around controlled work loops: intent capture, scoped work, specialist lanes, quality checks, and handoff evidence.
Voice-Driven Technical UX
Kept the interface oriented around natural language while grounding the work in explicit checks and evidence.
Developer Tooling
Designed the split between a private working system and a public open-source framework so the working model can travel with any project.
Quality Automation
Encoded quality checks, safety validation, and alignment monitoring so AI-assisted work can be reviewed before it is represented as done.
Continue
Want to go deeper?
Ask the Latif AI Guide about the architecture, the commercial logic, or the hard lessons behind this project.