Search for "best AI marketing tools" and you'll find lists of forty, fifty, a hundred tools, most of them with an affiliate link and a suspiciously identical description.
This isn't that. This is what survived daily use in a studio that ships brand, content, web, performance and social work every week. For each tool: what we use it for, what it's genuinely great at, and where it lets us down. No sponsorships, no affiliate links.
One principle before the list. A tool is only as good as the brief you give it and the edit you do after. None of these replace taste. They replace hours.
Writing, research and thinking: Claude
What we use it for: first drafts, research summaries, restructuring messy documents, interrogating a brief, rewriting copy for different audiences, building scripts and automations, and as a sparring partner when we're stuck.
What it's great at: long, nuanced documents. Following a tone-of-voice guide closely. Reasoning through a problem rather than just producing text. It's also become a serious technical colleague: much of the plumbing behind our own operations is written and maintained with it.
Where it falls short: like every language model, it can be confidently wrong, so facts get checked. Left without guidance, its prose drifts towards polite and generic. Give it your voice, your examples and your opinions, or you'll get everyone else's.
Alternatives we rate: ChatGPT and Gemini are both excellent. Honestly, the gap between the top assistants matters less than how well you brief them.
Video: Higgsfield
What we use it for: concept films, motion tests, animating stills for social, mood pieces for pitches, and short-form content where a full shoot would be overkill.
What it's great at: cinematic camera moves and fast iteration. You can test ten visual ideas in the time it takes to book one location.
Where it falls short: consistency across shots, hands, text in frame, and anything that must show a real product or person exactly as they are. It's a sketchbook and an effects department, not a replacement for a real shoot when accuracy matters.
Our rule: we never present AI-generated people or footage as real in client communications. That's a hard line.
Images: Magnific
What we use it for: upscaling and enhancing images, rescuing low-resolution assets clients send us, adding detail to renders, and generating textures and backgrounds.
What it's great at: making an image sharper and richer without making it look plastic, when you keep the settings restrained.
Where it falls short: push it too hard and it "hallucinates" detail that wasn't there: extra texture, changed faces, altered product features. Always compare against the original.
SEO and keyword research: OpenSEO
What we use it for: keyword research, difficulty checks, competitor gaps, rank tracking and site audits. The research behind this very Journal came from it.
What it's great at: fast, credit-based research without a heavyweight enterprise subscription. Good data for deciding what's worth writing before anyone writes it.
Where it falls short: search-volume numbers for small markets (Greece, for instance) can be thin or missing. Data tells you where demand is. It can't tell you what to say that's worth reading.
Automation: Make and Airtable
What we use them for: connecting everything else. Form submissions into a CRM, reports assembled automatically, content calendars that update themselves, alerts when an ad account misbehaves.
What they're great at: turning a dozen manual copy-paste jobs into a workflow that runs while you sleep. Airtable is a database people actually enjoy using. Make is the glue.
Where they fall short: automations break silently when an app changes. Someone has to own them, check them and fix them. Undocumented automations are tomorrow's mystery outage.
Meetings: Granola
What we use it for: notes during client calls and internal meetings. It enhances our own rough notes with the transcript rather than replacing them.
What it's great at: catching the detail you missed while you were busy actually listening. Action lists that write themselves.
Where it falls short: it records what was said, not what was meant. The judgement of what matters in a meeting is still yours. And always tell people when a call is being transcribed.
Ads: the AI already inside Google and Meta
What we use it for: Performance Max on Google and Advantage+ on Meta, where the platforms' own AI handles targeting, placements and bidding.
What it's great at: finding buyers you wouldn't have thought to target, at scale, faster than any manual setup.
Where it falls short: it's a black box that optimises towards whatever you tell it success looks like. Feed it bad conversion tracking or weak creative and it will efficiently spend your money on the wrong thing. The strategy, the creative and the measurement are still very much a human job.
Email: Brevo and MailerLite
What we use them for: newsletters, automated sequences and transactional email for us and for clients.
What they're great at: solid deliverability and built-in AI helpers for subject lines and drafts.
Where they fall short: the AI features are fine for a first draft and not much more. The thing that makes people open an email is still having something worth saying.
What's not on the list
We've tried plenty of "all-in-one AI marketing platforms" that promise to plan, write, design and post everything for you. None made the cut. They tend to be average at everything, and average is exactly what a brand can't afford.
The real tool is the workflow
If there's one lesson from two years of building AI into a studio, it's this: no single tool is the advantage. The advantage is a workflow where each tool does the one thing it's best at, and humans make the decisions in between.
That's what we set up for clients in our "Make it AI-ready" work: the right tools, connected properly, with your team trained to use them. See how we work or talk to us.