Letaido Review: 10+ Jobs Our Marketing Team Hands to It Every Week

Content Marketer at Ahrefs
- What is Letaido?
- 1. We hook Letaido up to the tools we already use
- 2. We build on Ahrefs data our public API can’t reach
- 3. We build, host and publish our own sites
- 4. We build on each other's work in one shared workspace
- 5. We recruit Letaido as a teammate
- 6. We keep our content updated automatically
- 7. We monitor our market without manual checks
- 8. We build the tools we need in under an hour
- 9. We prototype the Ahrefs features we've always wanted
- 10. We turn our data into original research
- What you can do with the $99 plan
- Why choose Letaido over Claude, ChatGPT or Ahrefs MCP?
- Who Letaido is for
- Final thoughts
Every week, the Ahrefs marketing team hands Letaido work we used to do by hand.
It keeps our data-led articles up to date, builds and hosts our sites, runs our research, watches our competitors, and plans out our day from start to finish.
We're lucky enough to have lots of credits to use internally, so I also tested the regular $99 plan + $50 in AI tokens on a fresh workspace, to see what a typical marketer gets for their money.
- What is Letaido?
- 1. We hook Letaido up to the tools we already use
- 2. We build on Ahrefs data our public API can’t reach
- 3. We build, host and publish our own sites
- 4. We build on each other's work in one shared workspace
- 5. We recruit Letaido as a teammate
- 6. We keep our content updated automatically
- 7. We monitor our market without manual checks
- 8. We build the tools we need in under an hour
- 9. We prototype the Ahrefs features we've always wanted
- 10. We turn our data into original research
- What you can do with the $99 plan
- Why choose Letaido over Claude, ChatGPT or Ahrefs MCP?
- Who Letaido is for
- Final thoughts
What is Letaido?
Letaido is an AI marketing platform built by the team at Ahrefs.
Think of it as an AI teammate that connects to your marketing tools and Ahrefs data, then puts AI agents to work; building reports, shipping tools, and automating workflows 24/7, in one shared workspace for your whole team.
Here are 10 of our favorite jobs that we use it for everyday.
1. We hook Letaido up to the tools we already use
Once we connected Letaido to our existing tools, it started to feel like a proper member of the marketing team.
Letaido comes with 70+ integrations, including HubSpot, Google Analytics 4, Notion, Stripe and Google Ads.
You can plug into one tool and get it doing deep work on its own, but a lot of the best marketing work crosses tools.
For instance, a keyword gap lives in Ahrefs, the brief lives in a Linear ticket, and the draft lives in WordPress.
Before Letaido, we’d be the go-between: trawling and analyzing, documenting and building—crying, maybe?
Now we just prompt, click approve, and Letaido handles it.
Here are some of the cool things our team has been doing with those integrations:
Trend monitoring
I use Apify to build a LinkedIn Monitor Feed of popular posts from a select few SEO and AI experts.
Then I get a Slack alert when their posts start heating-up, as well as a summary of trending themes at the end of the week.
I use these posts for content ideas, my own social posts, and fresh blog examples.
Reporting
Ryan Law, our Director of Content Marketing, compares Google Search Console clicks with Ahrefs traffic data across our blog, our SEO and AEO hubs, and our localized articles to build his monthly report of our team's performance.
Planning content
Ryan also turns Ahrefs keyword gaps against competitors into Linear cards, assigned to the writer who’ll cover them.
Managing partners
Igor Gorbenko, our Head of Partnerships, built a database of every influencer partner we’ve worked with, from Google Sheets and Gmail.
The connectors in Letaido mean there’s absolutely no copying in and out of a chat window.
We can just get on and do our work all in one place.
2. We build on Ahrefs data our public API can’t reach
We work at Ahrefs, and Letaido still shows us new ways into our own data.
That’s thanks to the Ahrefs connector.
It is incredibly easy to set up. You just select the connector and log into your Ahrefs account one time.
Letaido’s Ahrefs connector gives us exclusive access to data from tools our API and MCP can’t reach, including:
For instance, my Blog Refresh Engine scores an article’s topic coverage against the top-ranking pages, then suggests fresh examples that also raise that score.
The score comes from AI Content Helper (AICH).
Our public API and MCP include just one AICH endpoint (Topics), while Letaido's connector reaches 35, including competitor selection, document setup, and the re-scoring my Refresh Engine relies on.
3. We build, host and publish our own sites
In Letaido, you can connect your own domain, or your clients’ sites, and Leta (Letaido's main agent) builds and publishes pages straight to them.
That means we can build real commercial products, with user accounts, billing and a postgres database— not just the lightweight artifacts and apps we'd create with Claude or ChatGPT.
Take Ryan’s Conference Spotter, a feed of upcoming UK marketing conferences.
He built it in under a day, and Letaido did more than just the design.
Actually, it built the whole site.
The agent collects event and speaker data and refreshes it every week, on a schedule Ryan set with a single prompt.
It also suggested available domain names, then bought and deployed the one he picked.
Ryan was shocked at how quickly that happened, since he’s used to going to Namecheap and sorting out the DNS himself.
From the first prompt to a live site on its own domain, he never left his Letaido workspace.
Sidenote
You can watch how he built the whole site on the Ahrefs Podcast.
Kamila Olexa, our Product Marketer, found site building in Letaido just as easy.
She set up a test website from one panel, with no GitHub pushes or Vercel deploys.
Now, Letaido has become her go-to tool for site projects.
After these experiences I find myself opening Letaido much more than Cursor

Kamila Olexa, Product Marketer, Ahrefs
Letaido has “self improving agents”.
They keep their memory and context between sessions, meaning they get smarter and more personalized as they learn what we need and how our team works.
They don't just remember preferences (“British English, please”).
They actually learn why we asked for a change and apply it next time.
And when one person builds or corrects something, everyone benefits from it, since everyone in the workspace shares the same files, apps and skills (i.e. saved instructions).
This is hands down one of my favorite things about Letaido.
I get to stand on the shoulders of giants (i.e. my ridiculously smart colleagues) and build on their great ideas, tools, and workflows.
It is like having our own Ahrefs Hive Mind (think: Pluribus, but less creepy).
When I start a new build in the Blog team workspace, Letaido often suggests ideas based on tools my teammates have already made.
For instance, when I asked Letaido for a better way to line edit my vibe written content, it recommended the Google Docs style editor built by Ryan Law.
Now, I use that app every day.
I also love the additional context and information it provides, based on what it’s learned about our products, features, or from discussions elsewhere in the workspace.
For instance, while contextualizing site drops during a Google algorithm update, Letaido brought up and linked to Ryan’s internal Letaido analysis of our own blog’s traffic drop—which looked like the result of an update at first glance, but was really a migration bug.
That’s exactly the kind of example I might write about in a blog post on algorithm updates.
(Letaido) is always updating its memory and context based on the conversations we're having. I feel like the workspace is actually getting smarter over time. It is internalizing our preferences and our workflows without us ever having to actually tell it to do that. It's subtle but the recommendations it makes and the tools it builds are becoming more useful, more customized to how we actually work. The blog workspace feels like quite a personal environment now.

Ryan Law, Director of Content Marketing, Ahrefs
5. We recruit Letaido as a teammate
We think of a Letaido agent as an AI teammate: we give it a role, a goal and instructions.
Then, it wakes up every few hours, decides what to do next and sends us a summary of its work.
We can even watch it cook in Letaido’s browser panel.
Agents help us get both strategic and tactical work done around the clock.
They’re also customizable. Admins can add up to 4,000 characters of “shared rules” that every agent follows, so replies stay consistent with our team’s standards.
My colleague Sam has created a “Head of Organic” agent that wakes up every two hours.
It works on the content strategy for a new site, makes updates and checks Search Console for ways it can improve search performance.
Then, Sam has a daily morning standup with his “Head of Organic” via Letaido’s voice call feature, where he finds out what it’s worked on, and points it in a different direction if needed.
In Sam’s words...
Voice calls are 🤯 when you have agent loops running.

Sam Oh, VP of Marketing, Ahrefs
While Sam chats, a subagent works on his requests in the background.
And he can take his call on the go, since Letaido is also available on mobile.
Here’s an example of a voice call I had with Leta.
In this discussion, Leta is my Data Analyst, taking me through the findings of an analysis I ran on the August and September Spam updates.
6. We keep our content updated automatically
Letaido’s automations were a turning point for me.
Here I am singing their praises in the Ahrefs HQ Slack.
They mean my content keeps updating in the background, so I can get more work done elsewhere.
The scheduled jobs in Letaido essentially pull fresh data, rebuild our data-led articles and send each draft to WordPress for review.
I run six monthly ranking articles this way, including our list of the fastest growing SaaS companies.
During this process, Letaido helps me turn my judgment into a series of repeatable rules.
Each month, the automation checks the search growth of around 2,500 companies, applies those rules (and many others) on its own, rewrites the article and sends it to WordPress for me to review.
SQ’s Data Refresh Hub does the same for our other data posts and saves him more than 20 hours a month.
7. We monitor our market without manual checks
Letaido monitors our market on a schedule and sends us alerts, so we hear about changes without checking for them.
For instance, Ryan set up a competitor monitoring system that allows him to keep tabs on what our competitors are publishing, and save ideas for posts we should create.
It pulls in our competitors’ new posts daily, lets us triage them with hotkeys, flags our closest existing content, and summarizes their strategy.
This is now an app anyone can use in the App Marketplace.
Instead of trawling competitors’ blogs when we remember to, we get a daily feed of what they’ve published, with our closest matching posts already lined up next to it.
8. We build the tools we need in under an hour
Building tools in Letaido is fast.
You describe what you want in plain English, and a working app appears in your workspace, often within seconds.
Tweaking it is just as quick.
Take The Webinar Analyzer built by our Product Marketer, Constance Tan.
It took her over a day to build in Lovable and another half day in Claude Code.
When she rebuilt it in Letaido, it took an hour.
Letaido just gets lots of little parts right in the first iteration: working buttons, working UI, charts with working graphs and tables, etc. Building something with Letaido gets most of the things I envision correctly at this point that I haven't needed to explore other platforms.

Constance Tan, Product Marketer, Ahrefs
Our tools materialize as fast as our ideas now.
On the blog team, this is a huge benefit: with AI eating a lot of our clicks, we can’t afford to move slowly.
It’s also helping us win back some of that traffic; we can embed tools into content that give readers a reason to click and stick around.
For example, Ryan created a tongue-in-cheek free llms.txt generator that, when run, spits out the main findings of our llms.txt study—i.e. that 97% of these files never get read.
It does a great job of building interest around our research.
9. We prototype the Ahrefs features we've always wanted
Letaido has turned our team into an informal R&D lab for Ahrefs.
With the Ahrefs connector giving us tons of data, we can build the features we wish existed, then hand the prototypes to our product team to build into the platform.
For instance, my colleague Mateusz Makosiewicz used Letaido to build a custom report tracking pages we had created or sponsored to influence AI answers, and it’s since been added to Brand Radar.
He also built a fact-checker for AI answers: it takes what assistants say about a brand (from Brand Radar), compares it to a list of approved facts, and flags anything that doesn’t match.
That’s now an app you can use in Letaido’s App Marketplace.
I’ve used Ahrefs data via the Letaido connector to build my own feature idea, too.
While writing a guide on AI search ROI, I wanted to put a dollar value on our AI visibility, so I worked with Letaido to develop a calculation for mention-adjusted impressions.
I floated the idea in our Slack #feature-suggestions channel, and it picked up tons of upvotes.
One came from our Customer Success Leader, Kellsey Yova, who had already built a prototype of her own in Letaido:
10. We turn our data into original research
Letaido also lets us turn our own data into research that nobody else can publish.
When Google rolled out two spam updates this year, I used it to measure how many sites each update actually affected.
To show what makes this possible, I’ll compare it with two popular alternatives, Claude Cowork and ChatGPT Work.
1. The data
Letaido pulled Site Explorer traffic data for 25,000 sites.
As I mentioned earlier, Letaido’s Ahrefs connector lets us access more data than Ahrefs’ MCP server or an API pull.
Claude Cowork and ChatGPT Work both reach Ahrefs through its MCP server, which offers around 28 Site Explorer reports to Letaido’s 104.
Three of the reports this study relied on, including AI content level and archived page snapshots, aren’t available through the MCP at all.
Even for the reports the MCP does cover, we’d have to pay for every lookup in API units.
Pulling a year of traffic data for 25,000 sites would use nearly a quarter of our Enterprise team’s monthly API allowance, on a single study.
Sidenote
A team on Ahrefs’ Standard plan would need around four months of API units to cover the same pull, and a team on Advanced would need nearly two.
In Letaido, the same pull doesn’t touch the Ahrefs API allowance.
2. The time it took
Research like this used to mean days, if not weeks, with our Data Scientist.
Now we run small and medium-sized studies ourselves, while our Data Scientist leads our large studies and anything that needs exclusive data.
For small studies, Letaido pitches an idea based on the endpoints we can access and runs the analysis once we’ve workshopped it, which is how we can now add information gain to almost every article—like SQ’s recent one below.
For my spam update analysis, Letaido ran the pull in the background and built a live dashboard in 45 minutes for $6-$7 in credits.
On the $99 subscription, that would use 12-14% of the $50 in monthly credits.
It’s a not-insignificant cost, but I think it’s worthwhile for an analysis of 25,000 sites at this depth.
3. Reuse and sharing
Letaido keeps the scripts, raw data and results in the workspace, so when Google ships the next update, I can ask it to rerun the study, without spending the API units that a rerun in Claude Cowork or ChatGPT Work would.
The results live in an interactive dashboard on our own site, which I can link from my blogs and share with the team.
What you can do with the $99 plan
I tested the regular $99 plan + $50 in AI tokens on a fresh workspace, to see what a typical marketer gets for their money in Letaido.
Sidenote
Ahrefs users on Lite or higher can access the regular plan (+$50 AI tokens) free for 1 month.
During my testing, I worked on four projects.
Here's what I built and how much time it saved me:
Output | What I use it for | Time saved vs doing it by hand |
|---|---|---|
Slack SEO report | Keeping an eye on my article portfolio without opening Ahrefs | [~1 hour a week of pulling and formatting data] |
Content gap dashboard | Planning new articles and pitching them to writers | [~1–2 days of keyword research and SERP checks] |
Barnacle SEO opportunity finder | A free tool embedded in my article | [weeks; I couldn't have built this myself at all] |
Backlink benchmark tool | A free tool to go with a blog post | [~1 day of pulling backlink data for ranking pages and working out medians; building the tool itself was beyond me] |
How fast I built and what it cost in $ credits
The Barnacle tool on Opus took most of my budget, at about $18, but my three Luna builds came to less than $5 between them.
In total I spent about $22 across the four builds, leaving roughly $28 of my $50 in AI credits.
What I built | Model | What it cost | How long it took |
|---|---|---|---|
Slack SEO report | Luna | <$1 | <30 minutes |
Content gap dashboard | Luna | ~$2 | < 20 minutes |
Backlink benchmark tool | Luna | ~$1.50 | < 20 minutes |
Barnacle SEO Finder: first working version (30 industries) | Opus 5.5 | ~$8 | ~45 mins |
Barnacle SEO Finder: expansion to 195 industries | Opus 5.5 | ~$10 | ~45 mins |
Pros
Connecting Ahrefs and Slack is genuinely easy for a non-technical marketer.
Letaido uses Ahrefs data without being asked.
It challenges weak ideas before building them, and suggests simpler ways to get what you're after.
Chatting with GPT-6 Luna is free, and it still gets you real, usable output.
Cons
You pick the initial model yourself. I’d love Letaido to choose dynamically based on task complexity. It does a little of this already with subagents: on the Barnacle build, Opus was the orchestrator, and handed the task (e.g. bulk page labelling) to a cheaper model.
What used credits fast vs slowly (and what's free)
Not everything you do in Letaido uses credits at the same rate: some things burn through them fast, others slowly, and some not at all.
Fast: long, multi-step builds on a premium model like Opus.
Slow: Scripted automations that make one small model call per run, such as summarizing what they pulled. My LinkedIn Monitor tool is an example of this; the only part that uses credits is a single AI call that writes a one-sentence summary of each post, so a whole day of monitoring costs very little.
Free: chat usage on GPT-6 Luna or Qwen. The back and forth with Letaido in a chat window (asking questions, planning, re-prompting) doesn't come out of your credits.
Unused credits also carry over to the next month, so nothing goes to waste.
You can see a running cost of your usage at the top of every chat, and each request is logged and priced in real dollars in Controls > LLM Requests, so you can see exactly where your credits went.
Why choose Letaido over Claude, ChatGPT or Ahrefs MCP?
Claude and ChatGPT are great all-rounders, and both can plug into Ahrefs through its MCP server.
But here’s what I think Letaido does better for marketing work.
More Ahrefs data
Letaido’s Ahrefs connector gives us ~525 actions across 17 tools, including ones that aren’t in the public API or MCP that Claude and ChatGPT connect to, like AI Content Helper and Content Explorer.
In my experience, that’s what makes it the best option for SEO work.
Build, host and run sites all in one place
Letaido builds your site based on one prompt, and it goes live from your workspace, with no GitHub, Vercel or separate hosting account to set up.
Claude can only share lightweight interactive pages.
ChatGPT can now publish sites too, but Letaido goes further in a few ways.
First, the domain.
Letaido lets you buy one and connect it without leaving the workspace, while ChatGPT only connects domains you already own.
Second, connected data.
ChatGPT’s connected apps only work on private team sites, where each teammate sees their own data. You can’t use them to show your data to the public.
A Letaido public site can show data pulled from your connected tools to anyone with the link, no login required, which is why it can support self-updating content like the brand search demand data in Ryan’s Conference Spotter.
Third, logins and payments.
Letaido has both built in, via Stripe.
OpenAI’s docs also don’t describe a ready-made way to take payments on a Site, and its built-in sign-in option is “Sign in with ChatGPT”, which uses visitors’ ChatGPT accounts.
Finally, the price.
Hosting is included in Letaido’s $99 plan, whereas ChatGPT’s site hosting is free until the end of 2026 and then charged by usage above your plan’s allowance from 1 January 2027.
Any model, for any job
Claude only offers Anthropic’s models, and ChatGPT only offers OpenAI’s.
In Letaido, you can pick a model per job (say, Opus to write code and another model to review it), and add image, video, voice and music models too.
I like using a cheaper model for simple jobs and saving the premium ones for complex builds.
Automations that don’t burn AI credits
Agentic features and automations in Claude and ChatGPT count against your usage allowance.
In Letaido, mechanical jobs, like pulling data and posting it to Slack, can run as plain scripts instead, which keep costs down.
One price for the whole team
I think this is one of the best things about Letaido.
The regular $99 Letaido plan covers unlimited users.
With Claude and ChatGPT, every new seat adds $20 to $25 a month.
More on that below.
Letaido is one price, however big the team gets
The regular $99 Letaido plan covers unlimited users, so adding teammates costs nothing extra.
Team size | Letaido | ChatGPT | Claude |
|---|---|---|---|
1 person | $99 | $20 (Plus) | $20 (Pro) |
5 people | $99 | $125 ($100 annual) | $125 ($100 annual) |
20 people | $99 | $500 ($400 annual) | $500 ($400 annual) |
That’s not quite like-for-like, since those plans come with their own usage allowances, and Letaido’s $99 includes $50 in AI credits.
Still, for a team of five or more, I think the maths is hard to ignore.
Sources: Letaido pricing, ChatGPT Business, Claude pricing summary. Per-seat prices are for monthly billing, Claude Team needs at least two seats.
Who Letaido is for
I think Letaido pays off most when you have marketing work that repeats, and you’d rather not do it by hand.
You don’t need a team for that. One person with the right jobs can get plenty out of it.
Who Letaido suits
These are the four groups I think Letaido suits best:
In-house marketers who want to automate recurring work, like reports, monitoring and content updates.
Agencies managing lots of clients, who can run many sites, workflows and skills from one workspace and get basic client work done fast. I can see this being a big unlock for them.
Creators and solopreneurs who can’t afford a team yet.
Local businesses who want to focus on their craft, not on building a website or managing a Google Business Profile. A plumber should be plumbing.
You don’t need an Ahrefs account
You don't have to have an Ahrefs plan to use Letaido. It works on its own. A free Ahrefs account can be connected, but it only gives free-plan data.
That said, I’d say the Ahrefs data is where it really shines.
Who it doesn’t suit
Letaido does a lot, but there are two cases where I think a different tool is the better call:
Anyone who only wants an occasional chat assistant for writing or brainstorming. A free or $20 chat assistant covers that.
Developers who want the code in their own repo and a local setup. I’d say Claude Code or Cursor suit that better.
Final thoughts
Pick one job that repeats, run it in Letaido for a month, and see what changes.
Ahrefs users on Lite or higher can do that for free, so bring a colleague along: Letaido gets more useful the more of your team is in it, because everyone’s builds, skills and fixes end up helping everyone else.
If it works, the plan covers unlimited users, so the rest of your team can join at no extra cost.

Louise is a Content Marketer at Ahrefs. Over the past ten years, she has held senior content positions at SaaS brands: Pi Datametrics, BuzzSumo, and Cision. During this time, she has published hundreds of blog posts, spearheaded industry-leading research, and developed expert-led webinar programs.
Louise first learned the ropes of SEO during her time at enterprise SEO company, Pi. Within weeks of joining, she had published a piece on cannibalization that caught the attention of the biggest name in SEO at the time: Rand Fishkin. Since then, she has written on topics ranging from manipulative link building, to content syndication, and search demand forecasting. In her spare time, she has deepened her practical understanding of SEO by building websites for two local businesses.
You can read Louise's words in industry publications like Search Engine Watch, The Drum, The Telegraph, and Spin Sucks.
