AI marketing agency
AI-native means it’s how our whole team works, every day. We’re an AI marketing agency that gets better as model capabilities increase. We build our own tools, and our AI works from our extensive experience and know-how, written down. What hasn’t changed is who makes the calls: the strategist you work with.
Four things that set us apart
AI is built into our culture
Everyone here uses AI daily and has a genuine interest in the technology. We share what’s new in the industry each week, turn our best ways of working into skills the whole team uses, and show off the cool things we’re doing.
We build our own tools
An AI visibility tracker, a landing page platform and a keyword and content platform to name a few, all built in-house. Model agnostic, so each job gets the model that does it best and we’re not tied to one provider.
Context from your whole account
Before any task, the agents we use know the ins and outs of your business: your goals, brand voice, past decisions, account history and live data.
Strategists make the calls
AI does the legwork but there’s still stuff it isn’t good at. Your strategist decides what matters and checks the work before it reaches you.
A team that works this way every day
We adopted AI early because we believed it would change how agencies work, and that the ones who got good at it first would pull ahead. Everyone here uses it daily, and because we’re a small agency, when something better comes along, we can change how we work quickly.
What that looks like in practice
- Chat & RankOur weekly catch-up: what’s happening in AI, what we’re building, and tips on using it better.
- Hands-on workshopsWe build and test new workflows together, on real client work.
- Our know-how, written into skillsThe way our best strategists do a job, written down so our AI follows the same steps right across the agency.How a skill gets built ↓
- AI fluency in hiringIt’s part of how we choose who joins the team.
How a skill gets built and tested
Skills are one of the most important ways we get high-quality, consistent work out of AI models. Each one captures how one of our strategists does a particular job, so the model does the task their way, every time, instead of the generic way it would on its own. For repeatable tasks, this is how one gets built.
- The strategist who does the job best writes down how they do it: the steps, the judgement calls and the mistakes to avoid.
- We work with the models to turn that into a skill, and refine it on sample jobs until the output is consistently up to standard.
- Our most experienced strategists sign it off before the team uses it.
- On our biggest skills, a new version runs alongside the current one on finished work, and ships only if a strategist rates it as good or better.
- When a strategist corrects an output, we fix the skill as well as the work, and every change is versioned.
We build our own tools
As AI models have become more capable at writing code, we’ve started building our own tools and owning the stack. They’re built for the problems we see on client accounts, or the ‘wish list’ features we could never get off the shelf, and we can improve them as soon as we need to.
In useSince Feb 2026
AI visibility tracker
Off-the-shelf AI visibility tools get expensive at the volume we track, so we built our own. Some of our clients have it included in their service.
It tracks prompts across the major engines (ChatGPT, AI Mode, AI Overviews, Gemini and Perplexity) and records the answers they give back. It shows your share of voice amongst competitors, and the sites AI trusts for your competitors but not for you.
It feeds your end-of-month report, so over time you can see the trend and whether our work is moving the needle in AI answers.
In useSince Jul 2026
Landing page platform
We used Unbounce for many years. We’ve now moved off it and own the entire stack: no drag-and-drop page builder, just pages built with Claude on a platform we control.
We build pages on your own domain from templates refined over years of lead-generation work, and every page strictly adheres to your branding and design guidelines.
Every lead is stored and delivered straight into your CRM of choice, uptime checks run every five minutes, and A/B tests on new variants keep pushing conversion rates up. Pages typically go live within days and changes can literally take minutes.
Coming nextWe’re looking into AI that analyses recordings of how visitors use your page, finds the bottlenecks and pain points, and proactively recommends improvements.
In useSince 2025
Keyword and content platform
The main driver of our SEO campaigns. Our tool clusters thousands of keywords by live search intent, creates metadata, maps them to pages and plans site structure. You review the keyword clusters through a shared link.
When AI-assisted content fits the plan, it researches what’s already ranking well, writes the brief that fits the use case and drafts from your own material. Automatic checks and a second AI model read and critique every draft before a strategist does.
Also in use
Custom connectors
We use the connectors the platforms provide, and build our own MCP servers where they fall short. nimbl scrape, for example, gives AI a page as a browser renders it, JavaScript included, so it can read content most AI tools miss.
Technical audit engine
87 automated checks across a full site crawl, run alongside Screaming Frog exports. Our technical SEO strategists decide what matters and in what order.
AI-drafted reporting
Pulls the month's numbers, gathers the insights and checks every number against its source. Your strategist writes the analysis.
Link-building finder
Finds unlinked mentions of your brand and the sites AI already cites for the questions we track, and scores each opportunity by checking the page itself. AI drafts the outreach; nothing is sent without our approval and yours.
Model agnostic
Claude is our daily driver. We also use ChatGPT and Gemini, depending on the use case. The tools we’ve built aren’t tied to one model or company, so when a better one arrives, or one model pulls clearly ahead, we can move to it quickly and keep our work on the best model available.
What's planned
Rolling out
nimbl Hub
Our own CRM, built in-house. Connecting it to our AI comes next, so every workflow starts from one up-to-date record of your account.
Next
Our agency brain
A version-controlled knowledge base for our team and our AI, so every strategist and every workflow works from the same, current playbook.
Next
Live reporting dashboards
Built in-house to replace Looker Studio (formerly Google Data Studio). Your results in one place, always current, and because we build it ourselves, we can add whatever view you need.
Next
Always-on alerts
Early warning on budget pacing and on drops in traffic or conversions.
Context from your whole account
Anyone can use the same AI models we do. What makes the output good is what the AI knows: context. So before we run a task on your account, a report, an audit, an ad or a page, our AI has access to all the relevant context about your business it needs for the job.
AI context file
A living document for each client: your goals, brand voice, current strategy, the decisions we've made together and notes for each channel we run. Our AI workflows read it first.
Your account history
What we've done and why, pulled from our task lists, client conversations and the ad platforms' change logs, so reports and recommendations reflect the work that was done.
Live data
Connections to your Google Ads, GA4, Search Console, Meta, TikTok and more, read-only by default, so it works from current numbers.
Your material and feedback
Your website, product information and anything your team shares with us. When you correct something, it goes into your context file, so it holds next time.
Where AI still falls short
Given the right context, AI can know your business well. What it still struggles with is nuance: knowing when a dip is just noise, the claim your brand would never make, the call that comes from years of experience. That’s why a strategist makes every call.
Strategists make the calls
AI is fast and tireless, but it can be confidently wrong (yes, ours too). So we give AI the work we know it does well, with the right context and tools, and a strategist checks every output before it reaches you.
Who does what: some examples
SEO and AEO/GEO
Clusters thousands of keywords by live search intent, runs the technical checks, tracks rankings and AI answers, and finds and scores outreach opportunities
What to prioritise, which pages to build or fix, and what will move the needle
Content (when AI-assisted)
Researches what’s already ranking, briefs for the page’s purpose and format, and drafts from your own material under strict guidelines. Automatic checks and a second AI model read every draft
The insight the AI’s training data doesn’t have, the angle, the edit and the sign-off
Paid media
Pulls performance and insights across Google Ads, Meta and other ad platforms, and drafts negative keyword lists, ads and assets for approval
Strategy, budgets, what goes live, and how far to trust the platforms’ own automation
Landing pages
Builds the page on our platform from your brand guidelines, configures tracking, ensures lead delivery
The offer, the design direction and the examples to build from, and the go-live
Reporting
Gathers the data and the record of what we did, and checks every number against its source
Why the numbers moved, what it means, and what to change
- Plan (Strategist)Your strategist picks the page or pages from your custom roadmap and keyword research. Our tool orders each by expected traffic impact so we can prioritise accordingly.
- Brief (AI)Studies the content already ranking for the topic, pulls in the live 'people also ask' questions Google serves, briefs for the page's format and follows any custom instructions a strategist has set for the piece.
- Gather (AI)Pulls the relevant context from your own material via RAG, so the draft works from your facts and contains unique insights.
- Draft (AI)Writes to the brief and your voice, following strict writing rules. It never invents an offer, a figure or a policy the brief doesn't support.
- Check (Automated checks)Flags any number that isn't in your material, and checks for banned phrases (ours and yours), keyword stuffing and stock AI openers. Anything it finds is flagged for a rewrite.
- Second read (Second AI model)A model from a different AI company reads the draft and flags anything generic, repeated, off-voice or against the original brief. It flags problems for the strategist; it doesn't rewrite the draft.
- Final edit (Strategist)Your strategist does the last stage of editing and signs it off, so human input is always part of the process.
- Your review (You)The content is pushed to Google Drive for your review, with commenting access.
- Learn (You + strategist)Your feedback is collated and stored in your writing rules and the relevant context libraries, so the same issues don't surface again and each piece is better than the last.
A workflow in action
How a page of AI-assisted content gets made
Not every client needs it. When it fits, this is how it runs.
Who does each step
- Strategist
- AI
- Automated checks
- Second AI model
- You
- You + strategist
A page of content
Plan
Your strategist picks the page or pages from your custom roadmap and keyword research. Our tool orders each by expected traffic impact so we can prioritise accordingly.
The rules we work by
A strategist owns every output.
AI can’t be held accountable; people can. Every piece of work has a named owner, and very little goes out without their input or changes.
Every number traces to its source.
In AI-drafted reports and content, every figure is checked against where it came from. If we can’t point to it, it doesn’t go in.
We tell our AI to challenge us.
Our workflows instruct the AI to question our assumptions and flag what’s missing, and never to invent, embellish or tell us what we want to hear.
Read-only by default.
AI can look at your accounts. A person approves every change to them.
We don't use AI to inflate output.
The time AI saves on data pulls and first drafts goes into deeper analysis of your account.
Your data stays confidential.
AI doesn’t change our obligations: the same confidentiality terms cover everything it works with. Access is per person, and each client’s work runs from their own data.
Common questions
What does an AI marketing agency do?
An AI marketing agency does the same work as any digital marketing agency (SEO, paid media, content and reporting), with AI built into how that work gets done. At nimbl, AI handles the data pulls, technical checks and first drafts, using tools we’ve built ourselves and context from your whole account. Your strategist decides what matters, and very little goes out without their input or changes.
Will AI replace the people on my account?
No. You work with the same strategists, directly. AI takes the legwork, so more of their time goes into thinking about your account and coming up with ideas only a human can.
If AI does the heavy lifting, do I get more of your time?
You get better use of it. In 1865 the economist William Stanley Jevons noticed that as steam engines used coal more efficiently, Britain burned more coal, not less: when something gets cheaper to do, people do more of it. We apply that to thinking rather than output. The time AI saves on data pulls and first drafts goes into deeper analysis and sharper recommendations on the work you already have with us. Your scope stays the same; the thinking behind it goes deeper.
Do you use AI to write content?
When it fits the plan; not every client needs AI-assisted content. When we use it, the draft is researched, briefed and written from your own material under rules that ban invented offers and figures, checked automatically and by a second AI model, then edited and signed off by a strategist. Nothing is published without review.
Which AI models do you use?
Mostly Claude day to day, plus ChatGPT and Gemini where they do a job better. We review that choice as the models change.
Can we see the tools ourselves?
Where it’s useful, yes. You review your keyword clusters and site structure through a shared link, you see every landing page on a preview link before it goes live, and your AI visibility results can be shared as a read-only report link alongside your monthly report. Running the tools stays with us, and your strategist walks you through what matters.
Will Google penalise AI-assisted content?
Google’s guidance targets unhelpful, low-quality content however it’s made. We write for people, check every fact against your material and measure the results. Just ask us for examples of high-performing AI-assisted content we’ve done; we’ve got plenty!
Is this your AI SEO service?
No. That’s how we get you seen in ChatGPT, Gemini and Google’s AI answers: see AI SEO services. This page is about how we work.
Work with an agency that's ahead on AI.
The difference shows up on your account: how much of your strategist’s time goes into strategy instead of legwork, how fast a new page or a change goes live, and whether anyone is watching what AI says about your brand. Talk to a strategist about what that looks like for you.