An engine, by an agency, for agencies.
Every agency has the same conversation at the same time on a Wednesday afternoon: "This draft isn't quite us." And the same answer: "OK, rewrite it." The hours that follow — the diff between a draft that's almost-on-brand and one that's actually-on-brand — those hours are the agency tax. We built Solvgent because that tax doesn't have to be paid by senior strategists at four in the afternoon, again, on a Wednesday.
Solvgent is built by itligt, a small operating company that ran an agency for long enough to learn what the bottleneck actually is. Not generation. Not scheduling. Coherence. A draft that sounds like the brand on the first pass, not the third.
This is a note about what we believe, how it works, and where it goes next.
The agency content problem.
Most content tools assume the bottleneck is generation. Get something on the page, fast. They're solving for blank-page anxiety — a real problem for solo creators, but not the operative one inside a fifteen-brand agency. There, the bottleneck is coherence. Twenty drafts arrive on Monday; six are usable; the other fourteen need a rewrite from someone who knows the brand's voice better than the writer did. That rewrite cycle is where the margin lives, and dies.
The conventional answer is "more training, better prompts, sharper briefs." The honest answer is: those things don't compound. Every reviewer correction sits in someone's head, or in a Notion doc nobody reopens, or on a Slack thread that scrolls past quitting time. Never in the next draft. The agency pays for the same lesson twice in November and once more in January.
A brand brain is what happens when those corrections become substrate.1 Not a folder of style notes. Not a fine-tuned model. A working memory that gets denser every Wednesday afternoon — and that the Wednesday-afternoon-version of you doesn't have to maintain. It maintains itself by being used.
What we believe.
Every brand is a distinct system. Treating brands as variants of one template loses the texture that makes the brand the brand. A brand brain that converges to a single house style is a brand brain that has failed.
Corrections are the most expensive content an agency produces. A reviewer's note — "let's not say 'leverage'" — takes thirty seconds to write and represents five years of taste. It should never be re-learned.
Approval workflows are not friction — they are signal. Each approval is a vote of confidence; each rejection is a free lesson, paid for in calendar time the agency was going to spend anyway. Build the system that learns from both, not just the rejections.
The right metric isn't volume — it's the rewrite rate. If you're publishing more drafts but rewriting fewer of them, something is working. The rewrite-rate curve is the only graph we put on the wall.
How it actually works.
Each brand on Solvgent has its own brain — a private learning surface that watches every approval, every rejection, every rewritten line. The first week it's polite. The second week it starts catching the things the reviewer would have caught. By the end of the month it's catching things the reviewer would not have thought to flag, because the model of the brand is denser than the model of the reviewer's checklist.
Below: six pilot brands, six brains. Look at them. They are deliberately not the same shape.
Six brands. Six brains.
No two agents end up alike. The brain you see in twelve months won't be the brain it started — it's been shaped by every approval, every correction, every reviewer's note.
Each orb above is a real pilot brand's brain at its current density. The number of rings is the depth of the voice model; the scattered dots are individual lessons the brain has learned and will not relearn; the small amber traveler is the most recent correction, captured this week. If the orbs all looked the same, we would have failed. They don't.
Behind the orbs sits a working content engine — generation, quality review, scheduling, publishing, analytics — but the engine is the boring part. The orb is the product.
Where this goes.
The next twelve months are about depth, not breadth. We are running a small pilot cohort — fewer than twenty agencies — and building the brand-brain dashboard in public with them. Every feature ships to the cohort first. Every correction that flows through the system is studied. Nothing about the roadmap is determined by what's easy to build; it's determined by what the cohort's Wednesday-afternoon problems actually are.
We are not interested in being the biggest content tool. We are interested in being the one that agency partners forward to each other, the way they forward a good essay.
Every brand, its brain.
The agent that solves social content production.