Keyword research and content audits used to take 40 hours. Then Ahrefs launched Letaido, an agent workspace on August 12, and the story goes it cut that down to 60 minutes. But the same industry research tells a different story—45% of martech leaders say the AI agents vendors handed them fell short of their promises.
That's what everyone wants to believe—one agent, marketing time freed up overnight
Letaido isn't just a chatbot. It's an AI marketing agent workspace that plans multi-step workflows, runs them, builds dashboards and reports, and keeps tabs on your site and competitors—all on autopilot. The kicker? It has native access to Ahrefs' link analysis and keyword data, so you don't need to bolt on and babysit a separate API.
Ahrefs CEO Dmytro Gerasymenko put it this way: "We want to help marketers hand off that constant research and monitoring to an AI agent instead." Ross Simmonds, CEO of the content agency Foundation, got more specific. 40 hours of keyword research and content audit? Now it's 60 minutes.
Market sentiment's bullish too. Gartner saw 81% of martech leaders already piloting or running AI agents, and AI-powered marketing automation is expected to go from 16% of the mix in 2026 to 36% by 2028. So when a tool like Letaido shows up, the first instinct is: "Great, we're about to get time back."
Then the numbers tell you something else
Here's the rub: those same surveys tell the opposite story. BCG asked CMOs, and only 32% actually restructured marketing operations around agents. 42% use AI just to assist with individual tasks. Multiple agents running campaigns autonomously? That's 8%.
| Source | Number | What it means |
|---|---|---|
| Gartner | 45% | AI agents from vendors missed their promised outcomes |
| KPMG | 49% | Operating costs exceeded benefits; deployments scaled back |
| KPMG | 26% | Tracking AI operating costs in real time |
| TransUnion | 53% | Seeing measurable ROI from AI investment |
KPMG's survey of 2,145 executives across 20 countries stands out. Nearly half scaled back AI agent adoption, and the culprit? Token-based pay-as-you-go pricing. Every long-running task, every external tool call—the costs pile up. And only 26% of companies could actually see that happening in real time. Typeface found something else troubling: campaigns actually took longer after AI adoption, not shorter.
Heads up
That "40 hours to 60 minutes" is one agency CEO's story. No pricing. No scope details. No independent audit methodology. Read it as one data point, not an industry average.
The gap isn't about the tool—it's about how you run it
Look at companies six months into deployment, and the failure pattern's almost never the tool itself. It's operations: low adoption, expectations that don't match reality, data quality tanking over time, workflows falling apart when someone leaves, ROI that you can't actually see. Those are the repeating bruises.
Through that lens, Letaido's design priorities make sense. Always-on infrastructure so work keeps happening after you close the chat. Native data integration so no one has to manage a separate API. Both of those directly target the most common failure points: scope stalling out, and the tax of keeping integrations alive. But that's design intent. Real success depends on what teams discover once they're actually using it.
| How it usually breaks | How Letaido's designed around it | |
|---|---|---|
| Data plumbing | Your team builds and maintains custom APIs | Native Ahrefs data access |
| Task continuity | Close the chat window, work stops | Always-on infrastructure |
| Scope creep | Can't move beyond the first workflow | Dashboard and competitor monitoring built in |
5 things to nail before you adopt an AI marketing agent
- Lock down the cost model first
Per-token or flat fee? Make sure your approval process includes a cost review step. Without real-time visibility, the bill's a shock. - Know what "data integration" actually means
Native hookup vs. your team managing the API from scratch—that difference is the whole long-term ops burden. - Ask if it keeps working after you close the chat
No always-on infrastructure? Then someone has to open the chat window again to restart the work. - Start narrow. One workflow. That's it.
Set a clear entry point: "This task starts here." If you make it optional, people bail back to what they know. - Measure it yourself after four weeks
Track processing time, error rate, volume handled. Verify the "40 to 60" claim actually happens in your house.
The gap between adoption rates (81%) and satisfaction (45% disappointed) comes down to one thing: whether you've got the discipline to track costs and sustain operations. Letaido, or any other agent—this checklist comes first, tool choice comes after.
.png)
