Forge Studio: where we ship what we sell

Forge Studio is the venture arm of Forge Together. It exists for one reason: agencies that build products understand the work better than agencies that only advise on it. Every Studio product started as a problem we hit doing client work. We solved it for ourselves first, then turned the solution into software other people could use.

This page covers what is live, what is in build, how the two sides feed each other, how the build process works, and where it goes next. If you came from Forge AI Polish or word of mouth, the rest of the catalogue is below. If you are weighing up Forge Together as a marketing partner, this is the proof that the methodology is more than a deck.

Forge AI Polish

Live

Forge AI Polish is a content quality gate for AI-assisted writing. You paste in a draft, it scans for 51 patterns across 12 content types, scores the work, and rewrites the lines that read as obviously AI-written. Polish is a workflow tool for people who use AI in their process and still need the output to read like a person wrote it. Detection and humanising are different jobs.

We built Polish because we kept hitting the same editorial pain on agency content. Drafts came back from writers and from Claude carrying the same tells: triplet framing, em dashes everywhere, contrastive openings, conclusion mirroring. Catching them by eye took longer than writing the piece. Polish encodes those checks. It runs in seconds, scores out of 100, and shows you the exact lines to fix.

Live at polish.forgetogether.agency. Free as a guest, with a 3-day Pro trial on signup. Pro is £9 per month for unlimited scans, all 51 patterns, full rewrites, and exports. Used inside Forge Together on every piece of client content we publish.

Try Forge AI Polish

Messaging Audit

In build

A 60-second website messaging diagnostic. Give it a URL, it scores the homepage against an 8-dimension framework and returns specific findings with rewrite suggestions for the lines pulling the score down. £15 per audit, no subscription. It encodes the same positioning diagnostic we run with clients, scaled to a single page and a flat fee.

Built, security-reviewed, user-tested at 8.1 out of 10 with the core ICP. Currently in pre-launch polish. Messaging Diff sits next in the same lane: scores you against named competitors instead of in isolation.

Graftwork

Code-complete, pre-deploy

Graftwork is the all-in-one trades hub: quoting, jobs, invoicing, compliance, payment chasing. Built for UK plumbers, electricians, builders, and landscapers, the segment that loses 8 to 12 hours a week to admin and £112bn a year to late payments. It is the first proper Studio spinout, with its own brand and domain at graftwork.io, designed to operate independently of Forge Together. Code-complete across six modules and 100-plus API endpoints, with Stripe, Resend, Twilio, PostHog, and Sentry wired in. Currently pre-deploy.

Forge Engine v2

In build

A multi-tenant content engine for B2B teams and agencies. Seven-stage pipeline, composite quality scoring, a 17-sector source library covering 400-plus domains, and an AI trope blacklist of 160-plus phrases. The pipeline came out of the production system Forge Together uses on retainer client content. Forge Engine v2 wraps it in a SaaS shell with workspaces, billing, brand voice configs, SEO tooling, and publishing connectors. 39 of 42 build issues complete. 12-week target from mid-April 2026.

Forge Social Graphics

In build

Forge Social Graphics works like an AI creative director. You write a brief, the system makes the compositional decisions (layout, hierarchy, spacing, emphasis), and Puppeteer renders the result at exact canvas dimensions. The brand spec is the prompt context, so the output stays inside the visual rules without looking like everyone else's Canva post. Building against Credentially's brand first, with Naq next.

The flywheel is simple. Forge Together does retained client work in healthtech, fintech, and regulated B2B. That work surfaces the same problems across clients. Editorial drift in AI drafts is the one we hit first. Weak homepage messaging is second. Social production speed is the third. Forge Studio takes the productisable ones and builds them into software, which then feeds back into the agency. Retained clients get tooling an agency without a build arm cannot match.

Polish is the cleanest example. It started as an internal QA pass. Every piece of content we shipped to Naq, Credentially, OneID, Cymphony, and Wilson Partners ran through the same review: catch the AI tells, fix them, then publish. We turned that workflow into a product and opened it up. Our own content still runs through the live tool. The same logic applies to Messaging Audit, where the diagnostic Forge Together delivers in a £5k positioning workshop is being scaled to a £15 self-serve audit.

For clients, the practical effect is better tooling and a lower cost per outcome. You get an agency that has shipped the things it talks about. You get production tools drawn from real engagements rather than vendor demos. And you get a team that builds software for a living, which means we know what is coming next in the AI stack instead of reading about it on Substack.

The build process is deliberately boring. Plan, then test-driven development, then systematic debug, then verify. Every product follows the same loop. Linear holds the issue list. Anything risky goes into a worktree. Every change closes through a pull request, and tests pass before the next sprint starts. Polish has 1,230 passing tests across 86 files. Forge Armoury, our client-first sales enablement platform on the way to enterprise SaaS, sits at 542 passing tests with zero TypeScript errors. The discipline is the difference between a demo and a product.

The toolchain is opinionated and consistent across builds. Claude Code (Sonnet) for the heavy lifting, Supabase for database and auth, Vercel for deployment, Stripe for billing. Edge functions over server bloat. Migrations are versioned, and secrets live in 1Password. The sameness lets us reuse patterns across products and ship faster than a one-off stack would.

For a marketing buyer, the point is simple. The same rigour applied to a SaaS product gets applied to your campaigns. Every piece of client content runs the same review loop: research first, write, edit, score against an anti-AI rubric, only ship when both writing and brand alignment hit 98 out of 100. Most agencies cannot describe their QA process. We can show you ours, in the form of the tool that runs it.

The pipeline is real. Messaging Audit ships next, followed by Messaging Diff and ClaimGuard in the same £15-to-£19 pay-per-use lane. Forge Engine v2 lands as a multi-tenant SaaS later this year. Forge Armoury continues to evolve as the platform asset. Credentially is live on it. Naq follows, then Wilson Partners.

Some Studio products will spin out properly. Graftwork already has its own brand and domain and will operate independently of Forge Together. Others will stay close to the agency. We do not commit to a fixed launch roadmap because the next product is usually the one we just hit on a client engagement.

If you are an operator with a problem we should be building for, or an investor watching the spinouts, drop us a line.

Two ways in. If you want to see what Studio actually feels like, try Forge AI Polish on a piece of writing you were about to publish. The free guest scan takes about 60 seconds and shows you the score, the pattern count, and the conversion path before you decide whether to sign up.

If you want to talk about the relationship between Forge Together and Forge Studio, whether your business fits either side, or whether one of the in-build products solves something you are working on, get in touch. We will not sell you anything you do not need.