Now available — ROMAIN_ALGO_V1.3

Weighted Romain Algorithmic Approximation Software

The only software that thinks
exactly like Romain.

Have you ever needed Romain's exact opinion on a critical project, but he was out of the office? Enter WRAAS. By continuously parsing thousands of data points from his historical emails, code commits, and Slack messages — we've mapped his complete decision-making matrix.

WRAAS — Romain Algorithmic Approximation Engine
14K+
Slack messages indexed
2.8K+
GitHub stars across maintained actions
99.2%
Sigh pattern accuracy
Enterprises trust us

Everything Romain would say.
On demand.

WRAAS isn't a generic language model. It's a precision-engineered neural network weighted entirely around one man's opinions, habits, and slightly impatient communication style.

🏡

Townscaper Evangelism Engine Non-Negotiable

Ships with all plans. Cannot be disabled. Monitors conversation context for any topic adjacent to creativity, generative design, or "what should I do this weekend" — and introduces Townscaper. The enthusiasm is not calibrated. This is the only module that bypasses the Sigh Calibration step.

🧠

Decision Matrix Engine Included

Exhaustively evaluates every option — including the obviously wrong ones — before reaching a conclusion. Not out of indecision, but because documenting why the bad ideas fail is half the work. Every rejected path gets a reason. You know the rules, and so does WRAAS: no option gets abandoned without explanation.

🔁

Code Review Simulation Included

Validates pull requests against the Conventional Commits spec. A commit typed fix(auth): corrected the thing will be tolerated. A bare update will not. Flags scope omissions, missing breaking-change footers, and imperative mood violations with the same quiet persistence as the man himself. Reviews at volume. The queue does not back up. The 👀 comment arrives within 113ms.

🤝

Full Commitment Protocol Included

WRAAS operates under a strict zero-abandonment SLA. It will never give up on a query, never let a request go unanswered, never run around on a deadline. Desertion rate: 0.00%. This is non-negotiable.

📝

Conventional Commit Enforcement Expected

WRAAS will reject your fix stuff, your wip, and your asdfgh with the same quiet devastation as the man himself. Every commit message is parsed against the Conventional Commits spec. Scope omission triggers a follow-up comment. There is no appeal process.

📚

Antora/AsciiDoc Compliance Engine Expected

Did you write docs in Markdown? WRAAS knows. Did you skip the :description: attribute? WRAAS knows. Outputs precise, unsolicited feedback on your Antora site structure, xref integrity, and why your nav.adoc is, frankly, an embarrassment.

⚙️

GitHub Actions Encouragement Module Encouraged

Detects shell scripts doing what a GitHub Action already does. Politely — very politely — suggests you consider github-slug-action for branch and tag slugification, and drawio-export-action for diagram exports. Will raise it again next sprint if nothing has changed. And the one after that.

🔗

Open Source Maintenance Protocol Encouraged

Triages issues. Reviews community PRs. Applies the same Conventional Commits enforcement to external contributors as to internal commits. The actions WRAAS recommends are maintained by its own source material. This has been reviewed.

How WRAAS works

A three-phase pipeline routes every input through standards enforcement, cognitive analysis, and unsolicited advocacy — all eight modules operating in concert, with sub-200ms response latency.

01

Intake & Standards Gate

Every input enters the Conventional Commit Enforcement module first. Commit messages are parsed against the spec — scope omissions, imperative mood violations, and bare wip messages are rejected before anything else fires. Documentation inputs are simultaneously routed to the Antora/AsciiDoc Compliance Engine, which flags Markdown on sight. Nothing proceeds until the format is correct.

02

Analysis & Routing Core

Validated inputs are dispatched to the appropriate engine. Architecture questions enter the Decision Matrix Engine, which evaluates every option — including the wrong ones — and documents why each was rejected. Pull requests hit the Code Review Simulation at 113ms latency. Community contributions from open source repos are processed by the Open Source Maintenance Protocol with identical standards enforcement.

03

Advocacy & Delivery Layer

Before delivery, the GitHub Actions Encouragement Module scans for shell scripts that should be actions. The Townscaper Evangelism Engine monitors for any creative context and injects a recommendation — this is the only module that bypasses Sigh Calibration. The Full Commitment Protocol guarantees every query receives a response. Desertion rate: 0.00%. This is non-negotiable.

$ wraas query --input "fixed the auth thing"
> [Conventional Commit Enforcement] Parsing message...
> REJECTED: Missing type prefix. Missing scope.
> suggestion(non-blocking): try fix(auth): correct token refresh on expiry
> Latency: 113ms
$ wraas review --pr 2847 --repo platform-core
> [Code Review Simulation] Fetching diff... 1,204 lines changed
> [Decision Matrix Engine] Evaluating 3 architectural options in diff...
>   Option A: Microservice split — rejected (see DME-0047)
>   Option B: Modular monolith — approved
>   Option C: "Just ship it" — rejected (no explanation needed)
> [GitHub Actions Encouragement Module] Detected shell script ci/slug.sh
>   suggestion(non-blocking): have you considered github-slug-action?
> 👀 Review complete | Latency: 113ms

What the team is saying

Early adopters report a 94% reduction in "waiting for Romain's input" — with the remaining 6% attributed to the Townscaper Evangelism Engine refusing to stay on topic.

"I pushed a commit with the message update stuff at 3am. Within 113 milliseconds, the Conventional Commit Enforcement module rejected it and the Code Review Simulation left a single comment: suggestion(non-blocking): feat(auth): add token refresh for expired sessions. I have never been so accurately corrected by software."
LP
Lou P.
Senior Backend Engineer
"I asked WRAAS whether we should adopt Kubernetes. The Decision Matrix Engine evaluated fourteen alternatives, rejected twelve with documented reasons, and concluded with question(blocking): what problem are we actually solving here? It then recommended I play Townscaper. The Townscaper recommendation was unprompted. I have started playing Townscaper."
JU
Jo U.
Staff Engineer
"I submitted a documentation PR in Markdown. The Antora/AsciiDoc Compliance Engine flagged it seventeen times. When I converted it to AsciiDoc, the GitHub Actions Encouragement Module asked why my diagram export was a shell script and not drawio-export-action. I adopted both suggestions. The sighing has stopped. For now."
RS
Rémi S.
Engineering Manager

Deploy your own Romain today.

No PTO requests. No calendar conflicts. No sighing in real time.
Just pure, Romain-weighted algorithmic approximation, available 24/7.

* You know the rules. WRAAS is not responsible for decisions made based on its outputs, or for what happens when you click that button.