Growy already writes
like a category leader.
The interface hasn't caught up yet.

What this is based on.

Your manifesto says the best argument wins and problems get named early. So let me name the limit of this one first.

Your verbal brand is sharper
than your visual system.

"Not a chatbot. A co-worker that never drops the ball." · "Your knowledge exists. It's just stuck." · "Ego is expensive. We don't carry it."

That is confident, compressed, unmistakably yours. It is doing the work a brand is supposed to do. The product surfaces don't yet carry that same conviction. Not because the design is bad, but because the rules haven't been written down. Below are four places where the absence of a rule is visible on a page you show prospects.

That's the whole engagement in one line. Don't invent a new brand. Encode the one you already have.

Four, with evidence.

Ranked by effort against impact.

An audit that lists everything wrong is a complaint. This is the order I'd actually work in, and the reasoning is visible so you can overrule it on day one with information I don't have.

Items 1 to 4 are roughly two days of work and would be visible on your live marketing site inside week one. That is deliberate: I'd rather the first thing you see be small and shipped than large and pending.

The dashboard, re-systematised.

No new colours, no new logo, no new type family. Every change below comes from applying one rule consistently. This is what "amplify rather than replace" looks like in practice.

The node builder.

Your homepage runs a live pipeline: Understand request · Pull context from brain · Run workflow · Deliver outcome, with states Running and Queued. That's the canvas in miniature, and it's the surface where the visual language matters most, because it's the one thing a prospect watches rather than reads.

Courses, the surface nobody is defending.

"Total courses · 42 · Available to learners" is on your dashboard. There is no Courses page in your navigation, no Courses section in the product story, and no mention of it in the manifesto. It is either a real part of the product that the brand has abandoned, or a KPI that outlived its feature. Both are worth ten minutes in session one.

Rules, in the form a codebase reads.

The usual failure mode of brand work is a beautiful file that engineering never applies. I hand over the decisions in the form your codebase already consumes, so "apply the brand" becomes a merge instead of a project. Below is the shape of that handover, written as standards rather than suggestions.

That last panel is the one most brand handovers leave out, and it's the one that caused F-04. A shared word list is a brand asset. When conflict means exactly one thing, every future screen inherits the meaning for free, and nobody has to invent a synonym under deadline.

/* growy · foundation, illustrative structure */
--num-format: grouped;   /* 2,367 not 2367 */
--num-figures: tabular-nums;  /* digits align on refresh */
--num-precision-count: 0;   /* counts are integers */

/* KPI cards answer one question: better or worse? */
--kpi-subtitle-role: delta;  /* not status, not audience */
--kpi-delta-window: 7d;

/* semantic surface scale, extends what exists */
--surface-base       --surface-raised    --surface-sunken
--text-primary       --text-secondary   --text-muted
--state-positive     --state-attention  --state-critical

I've shipped this pattern before. My PulseWork system runs a semantic token layer with native light-dark() theming; a beauty commerce build I did runs a full Material 3 semantic architecture, pairing primary, container, on-primary and fixed-dim across roughly 300 lines. Both are in code, in production, and linked in the portfolio.

The everyday output, from the same rules.

The role asks for decks, one-pagers and marketing graphics alongside the product work. Every line of copy below is yours, taken from your published pages. Nothing here invents a new voice, because you do not need one.

01 · Marketing graphic16 : 9 · light and dark

The argument is the highlighted row. Four steps run themselves; the fifth, Human approval · Your call, is the one picked out. That is the objection every buyer has about autonomous agents, answered in the layout rather than in a sentence. The integration chips down the right do the same job for "five systems, one run" — naming Outlook, Dynamics, Word and HubSpot is more convincing than the number five.

The run itself is a constructed example, not your data. I have not seen a real execution log, and the same rule from section 01 applies here: better to say so than to imply access I do not have. Swap in a real run and nothing about the layout changes.

02 · Deck cover16 : 9 · title slide

Your homepage headline, set rather than rewritten. The second line drops to grey so the sentence breaks where the meaning breaks, and the same device carries across every piece in the set. Mono holds the data — eyebrow, date, confidentiality, integration chips — and the grotesk holds the prose. Two roles, no exceptions, which is the whole of the type rule from section 08.

Every piece exists in both grounds because that is what makes it a system rather than an artefact. The grid, the type scale, the position of the accent and the role of the mono are fixed; only the surface changes. That is the difference between handing you four files and handing you something your team can extend on Monday without asking me.

What I'd ship, and when.

Your form asks about two weeks; the role description says three. I've planned to be useful by day four and materially useful by day ten, so the third week is refinement rather than rescue.

Back-to-back, and separately.

The brief asks for perception sessions with the three of you. I'd run them individually before anyone hears anyone else's answer, because a brand perception gap between founders is the most useful thing an outside designer can find in week one, and it disappears the moment you're all in the same room being agreeable.

Then weekly for thirty minutes, with work in progress on screen rather than a finished presentation. Your manifesto says problems get named early; a weekly session where I show you something half-built is how that stops being a value and starts being a habit.