For one month, the Automattic for Agencies team ran an experiment alongside other product teams at Automattic. The question was simple: what happens when you strip away the overhead and let small, paired teams move at full speed entirely focused on one project?
We called it Radical Speed Month. People paired up, focused on a single passion project, and worked to ship as quickly as possible.
The Automattic for Agencies team took that challenge and ran with it. In 30 days, three projects went from idea to feature. Each one tackles a pain point that agencies feel every day: winning new business, producing on-brand collateral at scale, and getting answers out of your account without digging through dashboards.
This is a look at what we built and where it’s headed. These are experiments today, but the Automattic for Agencies product team is already folding them into the roadmap. The goal is to put all three in the hands of our agency partners very soon.
Amplify: From client pitch to partner
Win more business by knowing exactly what to fix on both your own agency site and client sites.
The problem
A prospective client lands on an agency’s homepage and bounces in 12 seconds because the trust signals are buried. The same week, someone asks an AI agent “who should I hire to redesign my law firm site,” and that agency never surfaces, because the model can’t parse the site properly. Both failures cost business. Neither shows up in a traditional performance audit.
That same pattern repeats one layer down. Every client site an agency builds gets judged by those exact two audiences: the human visitor deciding whether to trust it, and the AI tool deciding whether to recommend it.
What Amplify does
Point Amplify at any homepage and it returns a scored, prioritized report from two angles: how a prospective client perceives the site, and how AI tools read and rank it. Every flagged issue ships with an AI-ready prompt the agency can paste in to apply the fix.
Two scores out of 100:
- Human Mode: What a prospective client sees: trust signals, contact and conversion, mobile experience, design, SEO, content quality, accessibility, and audience resonance. We measure 60 signals across eight criteria to make sure your site is building trust with its viewers.
- AI Mode: What models see: technical health, structured data, AEO readiness, E-E-A-T signals, content freshness, entity clarity, content specificity, and llms.txt. We measure 49 signals across eight criteria to make sure your site is speaking the language AI agents need to rank it.
Every signal traces back to a credible, published source: Google Search Central, Schema.org, the Laws of UX, WCAG, and conversion research from Nielsen Norman Group and others. The result is a branded PDF with scores on the cover, a full rubric breakdown, and a prioritized issue list ranked critical to low. Each issue pairs the concrete problem with a recommended fix, and the report closes with paste-ready prompts to hand your AI tooling of choice.
Explore a sample report
Imprint: An agentic design studio inside Automattic for Agencies
Drop in a brief and your content. Walk away with on-brand collateral.
The problem
Every time an agency needs a blog cover, social media images, or a one-pager, a designer produces it by hand. A blog cover takes about twenty minutes. A one-pager takes a day or two with feedback rounds. Multiply that across the pitches, client work, and promotion an agency ships every week, and the team may spend hundreds of hours on routine design work every quarter.
Imprint does not replace an agency’s creatives. It hands them back their time. The studio speeds up the standard, repeatable pieces so designers focus on the high-craft, high-judgment work that actually needs them. Imprint clears the routine bottleneck.
What Imprint does
Imprint is a native team of agents built into Automattic for Agencies that produces on-brand collateral on demand. Give it a brief and your content, and the studio returns a first on-brand draft near-instantly, along with a matching promo set: social stories, email headers, and more.
Two agents anchor the first version. June designs one-pagers and Iris designs social graphics. An in-product chat lets you ask for changes after the first draft, the same way you’d direct a designer. For template-able collateral, that’s much faster than the human design-request workflow. The hand-finishing, brand-judgment calls, and custom client work stay with people, which is exactly where those hours should go.
Automattic for Agencies MCP: Your entire agency account, delivered to you
Connect the AI client you already use, then just ask.
The problem
Answers about your agency account live across different areas of the dashboard. Getting a quick answer often means knowing exactly where to click and maneuvering through various interfaces. As you advance your AI workflows, we want to make sure we meet you where you’re working and get the answers you need in seconds.
What the Automattic for Agencies MCP does
MCP (the Model Context Protocol) is an open standard that lets AI clients call into external tools. The Automattic for Agencies MCP server exposes our existing Automattic for Agencies read APIs as MCP tools. Pick your AI client of choice (Claude, Cursor, ChatGPT, Codex, etc.), opt in, choose which tools to expose, and authorize via OAuth. Once connected, you can ask AI anything about your Automattic for Agencies account.
Our MCP includes 22 read-only tools across seven categories:
- Agency: profile, tier progress, and team member list.
- Sites: managed sites, site details, and site tags.
- Site health & operations: site health, health overview, security status, and installed plugins.
- Migrations: incentive status, site migration list, and migration status.
- Commissions: earnings summary, site eligibility, and client referrals.
- Billing & invoices: billing summary, invoices, invoice details, and upcoming renewals.
- Knowledge base: search and retrieve articles.
What the experiment taught us
Three projects, one month, and a clear set of lessons that outlast the experiment itself.
- Time-boxing sharpens decisions. A fixed deadline forced focus on what could realistically ship and kept scope debates from dragging on.
- Existing infrastructure is a force multiplier. Reusing proven AI, MCP, and OAuth foundations made building real products in weeks possible.
- AI accelerates execution, not ownership. Agents were excellent for scaffolding and repetitive work, but architecture, debugging, and product judgment still came from people.
- Small teams move faster with tight feedback cycles. Daily standups, focused overlap hours, and small pull requests kept momentum high.
- Constraints made the work better. Building inside Automattic for Agencies’s real design system produced cleaner, more focused results than an unconstrained, stand-alone prototypes.
- Internal customer feedback was the best signal we got. When people kept asking for more, it meant we’d built something worth wanting more of.
It also reconnected people across the company who might never have crossed paths. Amplify exists because two people in different parts of Automattic found each other mid-experiment and discovered related work that would otherwise have stayed siloed.
What’s next
Radical Speed Month proved a point. When two people pair up on a project they care about and you clear the usual blockers, they ship useful products fast. Teams chose their own passion projects, and that ownership turned three ideas into three working tools in 30 days.
The Automattic for Agencies product team is now exploring refinements to all three, with the goal of releasing them to our agency partners very soon.
Interested in becoming an early adopter and trying these tools to elevate your agency? Let us know.
Automattic 