Every idea earns its lane.
Wolfgate is AI portfolio intake, biased toward no. You describe a problem and state its numbers; the gate shapes it into a brief a scoring committee can act on, then routes it to the cheapest honest answer — enablement, a workflow redesign, a governed build, or a process fix that needs no AI at all. This guide covers the whole path: getting in, shaping an initiative, reading the verdict, exporting the record, and the one rule the gate will never bend.
What it is — and what it isn't.
Wolfgate is the front door for "should we build something for this?" It takes a rough problem from anyone in the organization and turns it into a decision that's written down, reasoned, and comparable to every other request — without a meeting.
It is a triage instrument. It shapes a messy problem into a committee-ready brief, checks it against everything already in the register, and assigns it one of four lanes based on what the problem actually needs. It is deliberately biased toward no — the cheapest honest answer usually isn't a custom build.
It isn't an approval, a budget, or a build. A GOVERNED BUILD verdict means "this earned a place in the scoring queue," not "start coding." Nothing here commits money or people. It's the step that makes the later decisions faster and fairer.
One code. Twelve hours.
The gate is public to read, but the parts that call an AI model are behind a passcode — so the link is safe to share without leaving the model spend wide open.
The first time you shape or triage an initiative, a passcode panel appears. Enter the code you were given and press Unlock. That's it — one entry covers you for 12 hours, on that browser, and the panel won't interrupt you again until it lapses.
Open the gate
Go to gate.wolflow.ai. You can browse and read freely — no code needed just to look.
Enter the passcode when asked
It appears the moment you run your first initiative. Type it, press Unlock. A wrong code just says so, calmly — no lockouts, try again.
Work uninterrupted for 12 hours
The check happens on the server; nothing is stored in your browser that a page can read. When the 12 hours lapse, you'll simply be asked once more.
The passcode protects spend, not secrets. The register's example data is built from public signals only, so reading the site costs nothing and reveals nothing sensitive — the code simply keeps the AI calls from being run by just anyone with the URL.
Four questions. The second one is the one that matters.
Pick the department the problem lives in, then answer four short questions. Three are required; the fourth is optional but valuable. Write like you'd explain it to a colleague — the gate does the shaping.
Wolfgate will not send anything to a governed build without a measurable baseline. Question 02 is where that baseline comes from. Skip it, or answer "a lot," and even a genuinely build-worthy problem gets routed to a lighter lane until someone captures the real number. State the figure and the gate can route with confidence.
Your problem, in committee form.
Press Shape the initiative and the gate rewrites your four answers into a short, structured brief — the version a scoring committee can read without you in the room. It's fully editable; fix anything that's off before you send it through.
Every brief ends with one line that does the real work:
BASELINE: the before-state, in numbers, that any future result will be measured against. If you gave a real figure, it's captured here. If you didn't, the line reads NOT YET MEASURABLE — and that's not a failure, it's the gate being honest that there's nothing yet to measure a build against.
Think of the BASELINE as the contract. Nothing gets built on top of a problem whose before-state nobody wrote down — because without it, no one can ever prove the thing worked.
Four lanes, and the honest one wins.
Press Send it through the gate and Wolfgate weighs the brief against everything already in the register, then assigns one of four lanes. The lane is a recommendation with reasoning attached — never a bare label.
The root cause is a missing process, template, or bit of communication. The cheapest honest answer costs almost nothing — fix it, measure, revisit only if the pain survives.
Drafting or synthesis work that assistant tooling the organization likely already licenses can cover. Training beats building.
A department-owned redesign, possibly AI-assisted, is the proportionate response. Measure before and after.
High-volume, measurable, and beyond what off-the-shelf enablement covers. Advances to scoring — with the baseline captured before anything ships.
The verdict card also carries a baseline quality tag and, when the problem overlaps something already in the register, a dedup flag naming the initiative it resembles — so two teams don't quietly build the same thing twice.
A real, measurable before-state is on record. The verdict can be trusted and compared.
Some numbers, but soft or partial. Verify against system data before it advances far.
No measurable figure. A governed build is downgraded until the baseline is captured — no baseline, no build.
One click, one file that names itself.
Every verdict can be exported as a decision record — a single Markdown file that captures the whole decision so it survives outside the browser.
Press Decision record (.md) and the file downloads already named for you: a timestamp and the department, so records sort themselves in a folder. Inside is the shaped brief, the lane and its reasoning, the baseline quality, any dedup note, and the recommended next steps — everything a reviewer needs, nothing they don't.
From the same row you can jump to the register to see your submission logged alongside the rest of the portfolio, or run another one to start a fresh initiative without reloading.
The one rule the gate won't bend.
No baseline, no build.
Wolfgate will happily route a problem to a process fix, to enablement, or to a workflow redesign on judgement alone. But it will not advance anything to a governed build without a measurable before-state — no matter how promising the idea sounds.
This is on purpose. A build with no baseline is a build no one can ever prove worked; it's how portfolios fill with expensive things nobody can defend. By holding the line here, the gate keeps the scoring queue full of decisions that can actually be measured — and spares good ideas from being judged on vibes.
Your submissions stay yours. And it works even offline.
Two things worth knowing before you start: where your submissions go, and what happens if your connection drops.
That banner means your verdict came from the gate built-in fallback logic rather than the full AI model — still a useful read, but worth running the initiative again once you are back online for the sharper one.
