Most of the AI conversation in SEO right now is about the output. Can it write the post. Can it write the brief. Can it replace the person who writes the brief.
Erika Braeger is pointed somewhere else. She is Manager of Organic Growth Strategy at Ten Speed (@tenspeed), a B2B SaaS content agency that does organic, AEO, digital PR and design. She has been in marketing a little over 10 years, almost all of it agency side. And the thing she is automating is not the work. It is the two hours in front of the work.
The pre-work is what actually burns the day
Ask an agency strategist what takes the longest and they will not say "the analysis." They will say the part before it.
Here is how Erika put it:
"The thing with innovation is when you're asked to do an analysis for a client, there's often a lot of pre work involved. You're grabbing data from multiple systems. You're organizing it into a spreadsheet. And that pre-work takes time, it takes brain power before you can actually begin the analysis, before you can make a decision, before you can craft a narrative for the client, and before you can help the client execute."
That is four things stacked behind a spreadsheet. Analysis, decision, narrative, execution. None of them can start until somebody finishes exporting.
So that is what she builds against:
"So the innovation that I work on for my strategy team is Claude skills that take away that pre-work so the strategist has all of that time back to focus on what really matters, which is the analysis, building relationships with the client, growing their skills, crafting the narrative."
She said the same thing again later in the interview when I asked what her biggest win of the past year was. She did not name a client result. She named the skills.
"Through all of my agency experience, what I consistently ran into is it's the pre-work that can burn a lot of time, a lot of energy. And if someone else is helping you with pre-work, whether it's someone on the team or you're using an AI skill in Claude, you're able to then get back the time to invest into you know running your analysis, crafting the narrative. Making a strategic decision and helping the client execute."
The example she gave is the boring one, which is why I believe it. Conversations scattered across four systems:
"Claude can summarize conversations across your system. So instead of jumping from like place to place to place to read conversations, you can have Claude go out, grab them, summarize them, and that saves you a lot of time and effort that you would have spent going from to three, four different places."
Nobody is going to put that in a case study. It is also probably the single highest-leverage thing an agency can automate this year.
The line she will not cross
I asked her what pieces of the puzzle are best left un-Claude-ified. Her answer was fast:
"I would say everything that comes after that. I don't want to take away the thinking part. I want to allow the strategist to arrive at the thinking part with more ease, without burning energy, on grabbing data, organizing it out of multiple systems, organizing it in the spreadsheet."
Arrive at the thinking part with more ease. That is the whole strategy in one sentence.
It is a different bet than most agencies are making. A lot of shops are pointing AI at the deliverable and hoping the margin shows up. She is pointing it at the setup so the human has more gas in the tank when the actual judgment call arrives.
And she is not starry-eyed about how far that goes. My previous guest Mike Montague passed forward a question about superintelligence, whether we get there and what changes. Erika:
"Possibly. I still find even with all of my Claude skills or Claude usage, I still need to run a thorough QA. There's still a QA element, so I question like when that could possibly happen and how much we could possibly hand over. I'm hesitant."
Someone who builds skills full time telling you she still QAs everything is worth more than a hundred LinkedIn posts about agents.
She was a teacher, and it shows
The part of this conversation I did not expect was the management.
Erika listed off her day, advising on strategy, deliverable quality, spotting technical and content and UX and CRO opportunities, and then landed on this:
"Most importantly, a big part of my role is helping people grow, helping them build confidence and trust their instincts."
Then she told me where that came from.
"I was a teacher in a past life and my students didn't always have the answers to everything. Sometimes they knew it, they weren't sure whether or not to say it, so I do something very similar with my team."
Here is the mechanism. A client is overhauling their navigation. The strategist on the account is plugged into the client, plugged into the industry, and already knows what works in a navigation redesign. They come to Erika for a gut check.
"I don't just sign off on the gut check. I keep asking curious questions to coax the answer out of them because they come to me if they're not sure. But ultimately I want to show them that they did know the answer, and then they can reapply this confidence the next time."
Read that again if you manage anyone. Signing off is faster. Signing off also guarantees they come back next time.
Her reason is that SEO moves too fast for the answer to be the deliverable. The deliverable is a person who can act without you.
I asked if this was really about imposter syndrome, the thing sitting in the back of every consultant's head.
"I would say yes and no. When I started in SEO, I certainly didn't know everything. I went in for a lot of gut checks, and I will say there were gut checks where I went in, like knowing the answer in the back of my mind. There's ones where I went in to collaborate, there's ones where I went in with the answer in the back of my mind, and I would say more so when the answer was in the back of my mind, it was imposter syndrome."
That is a clean diagnostic. If you already know and you are asking anyway, that is not collaboration. That is permission-seeking, and a good manager will make you say the answer out loud.
I also asked her what surprised her in 2025 or 2026. She thought about it and said:
"Can't think of one right now, actually. After many years in agencies, less things surprise you over time, I would say."
Honestly the best answer anyone has given me to that question all year.
SEO got wider and there is no opting out
I asked how she explains the value of SEO to leads now.
"It has really, I would say, expanded a lot over the past two and a half, three years. With Chat GPT got to be on the rise, I would say, like end of 2023. We have to think about so much more. We always had to, but now it's become a must-do. It wasn't something that could be kicked to like the nice to have part of the list. Now it's the must-have part of the list. So we're thinking about digital PR, we're thinking about design, we're thinking about video, we're thinking about all of these other elements on whether it's on page or off page that matter, that impact how you show up across the web."
That matches how I have started framing it, which is mine and not hers, so take it as mine. I used to draw SEO and LLMs as two circles in a Venn diagram. I do not anymore. I slid the LLM window completely over the top of SEO. There is nothing we do in SEO that you cannot also look at through the framework of LLM optimization at the same time.
What that costs you is the ability to work alone. In what I call the golden age of SEO, 2010 to 2020, you could pull so much organic traffic that you could afford to never talk to the digital PR team, never coordinate with social, and never speak to the client.
I did exactly that. Three or four projects where I did a company's SEO and never met or spoke to a single human on their side. Everything through email, handed the account, did the work, sent it to content, done.
That systemology is dead. There is too much value locked up in the humans on the client side to go get it through a ticket queue.
The brief is the product
So I asked Erika the version of the question that actually matters for an agency in 2026. Anybody can buy Claude now. A small business owner pays the monthly fee and has the same model you do. What are you adding on top of baseline?
She was straight about her lane first, that she runs strategy and client relationships and there is a separate side of the house handling production. Then she answered anyway, and the answer was the brief.
"It kind of depends on team bandwidth and how they choose to execute. But ultimately, it's the brief that sets the content up for success, ensuring that it has subject matter expertise, that we conducted our deep research on the product, we conducted our deep research on the ICP and understand where this piece lands in that moment of the journey."
Some Ten Speed clients get fully produced content. Some get briefs. Some of the ones who get briefs run an AI-assisted process with editing on their end. Erika's point is that the model does not change the requirement:
"I think the human element is important because a human has to strategize what the topic should be, how it's connected to the greater picture, and then how it's connected to the ICP at that moment of the funnel, and that lens also needs to be applied in the editing process. Whether a human wrote that content end-to-end or whether if it was an AI-assisted writing process, we still need that expertise review, expert review layer."
Expert review layer, either way. Not because a human wrote it. Because somebody has to check that it lands where it was supposed to land.
The SME input is not a nice extra either. It is built into the brief itself:
"A lot of that happens early on in the client engagement where we're getting to know the client's team, their industry, we're getting to know the product and we're understanding like how the product is used by the ICP, what keeps them up at night, what gives them gray hairs, what does their tech stack look like, all of these different moving pieces that draw like the ICP to that product. And SME is built into our briefs, so we are tapping experts on the client's team to inject that expertise into the contents so that the piece that comes out is ultimately more aligned with their needs and their challenges."
One more data point that surprised me. I assumed she was fighting clients who wanted to cut the agency out and just prompt it themselves. She is not:
"I haven't had a situation come up where a client is set on using an LLM to write their content."
Humans first, machine readable second
I threw her my favorite loaded question. Are we writing for humans, or are we writing for bots who talk to other bots who are being prompted by humans?
"It's a little bit of both. You need to write content for humans first. However, you need to keep in mind that in order to get the content in front of humans, it needs to be machine readable so that the bots that parsed the content can present it to the correct humans at the correct time."
That is the cleanest version of this I have heard. Machine readability is not the goal. It is the delivery mechanism. You write for the person, then you structure it so the parser can find the right person to hand it to.
She learned that on a food blog, which I love. Her only real in-house role was her family's blog, Ethnic Spoon.
"My family has a food blog, Ethnic Spoon, and that's where I really learned SEO, they had an audit done, and I helped execute the audit, and then I really enjoyed what I was doing. It was like very granular work, updating meta descriptions, updating titles, title tags, going through each blog and reorganizing the content to make it more machine readable as well as human friendly."
Meta descriptions and title tags on a recipe blog, and the same principle now runs her B2B SaaS strategy team. Since then it has been agencies:
"I have mostly been in B2B SaaS. I would say 90% I've been in B2B SaaS."
Would a friend understand this
The failure mode in B2B SaaS content is writing for the people who built the product.
"You want to make sure that you're delivering how the product works, who it's for, in language that humans can understand, but humans outside of the organization. If you're writing too much in technical language or internal language, it's not going to land."
Her gut check for that is dead simple and you can use it today:
"I like to take a step back when I'm working on B2B SaaS content and say, if I were to show this to a friend, would they understand? Or you can always gut check with someone else on the team that's not involved in the account, like, hey, does this make sense to you?"
And the reason she needs it:
"I use this a lot when I was a copywriter, because I would get so far into learning about a company, learning about a product that I didn't realize how technical I was being."
That is the trap. You get good at the account and you stop being able to hear it.
Never rush a website migration
I asked for a red line. Something people believe that they should absolutely stop doing. No hesitation:
"Don't rush a website migration, I will say. Never rush a website migration."
I offered up the usual carnage, forgotten redirects and whole sections of the site left behind, plus my own suspicion that half the time the migration should not have been the answer in the first place. She agreed and gave the fix, which is a question, not a checklist:
"I've run into that as well, like changes from one CMS to another, but when I'm working on a website migration with a client, one of the first questions I ask is like, what are your goals? What are the business goals behind this migration to make sure that they're not moving from WordPress to Webflow or vice versa because somebody likes it over the other."
Ask the business goal before you scope the move. A surprising number of replatforms do not survive that question.
On CMS preference she is agnostic, which is the correct answer for an agency:
"I wouldn't say that we push or prefer any CMS. Our team will learn your CMS if we're publishing content to your site. I started in WordPress years ago and I got most comfortable with WordPress through all of their iterations. So that's my preferred CMS because I know which plugins work well with it and I know all of the other little quirks as I'm making recommendations."
"It's less about a preference and more about knowing how they all operate so you can make good recommendations."
What AI actually cites
Ten Speed just shipped their first research report, and it is aimed at the part of the funnel everybody is guessing about.
"It's taking a look at how your own content that you can control will dominate AI citations at the consideration stage of the funnel. It'll talk about what type of pages are leading citations, what's happening with community-driven sources, and what's happening with comparison format queries."
That is the useful finding buried in there. Your own content still does the work. The reflex right now is to assume the model only cites Reddit and listicles and there is nothing you can do about it, so you throw up your hands and go commission a thousand listicles. This says otherwise.
Read it here: What AI Actually Cites for B2B Evaluation-Stage Prompts.
Her open problem
Every guest passes a question forward to the next one. Erika's is the thing she has not solved:
"Maybe centralizing all of my information. So now I need to pay attention to Reddit and I need to pay attention to G2 and I need to pay attention to YouTube. So making sure I'm keeping track of what's happening with a client's visibility, a client's mentions across all of these systems is a challenge."
She has a hypothesis about the shape of the answer, and it is the same shape as everything else she said:
"The best way to to execute that, I would say, is with a tool that's going out and taking all of these steps for you, similar to what I'm doing with the Claude skills. So what if we're able to take out all of this pre-work and just have everything pulled together into one place so you can start your analysis?"
Kill the pre-work. Keep the analysis. She is consistent about this whether the subject is a spreadsheet, a strategist, or a monitoring problem nobody has a good tool for yet.
If you have solved cross-platform visibility tracking across G2, Reddit and YouTube, I want to hear it, and so does she.
Listen to the whole thing
Erika Braeger is Manager of Organic Growth Strategy at Ten Speed. Find her on LinkedIn.
Watch on YouTube
Listen on Castos
Full recap and show notes at Unscripted SEO
More interviews at unscriptedseo.com.
Jeremy Rivera is a freelance SEO consultant at JeremyRiveraSEO.com and the founder of SEO Arcade.