Everyone is selling AI visibility now. Here's the map.

Hi everyone,

Two things happened in the same seven days last week. Symphony Talent launched a full AI search visibility offering and Indeed paused the automated mode of its Apply For Me feature after employers pushed back on paying for bot-generated applications.

One company entering a category and another retreating from automation might look unrelated, but they're the same story. The volume era is breaking under its own weight and the visibility era just got its first crowded shelf.

The demand side says the same thing. A recent People Brand Collective poll asked employer brand leaders which topic is most pressing for their organization right now, and nearly half picked AI search (AEO and GEO) over measurement, advocacy and everything else on the list.

I've spent the past year inside this shift, talking with hundreds of heads of talent acquisition, agency leaders, vendors and live clients about how employers show up when candidates ask AI where to work. So this week I want to give you the map as I see it, names included.

1. Monitoring and optimization software. Tools that track how your brand appears across ChatGPT, Gemini, Claude and Perplexity, then tell you what to fix. This is where we started and still excel at CollabWORK, paired with the XML feed work underneath it to make your jobs more discoverables to LLMs. Symphony Talent's new AI Search Visibility Solutions enters here too, wrapped in their full-funnel stack. Major vendors building this validates the category more than it threatens anyone in it.

2. Audits and services. Agencies and consultancies selling the diagnostic and the fix. Talivity runs a four to six week audit showing where you're invisible and who's winning instead, then implements. And nearly every agency we talk with is weighing the same question right now: build this capability, buy it or resell it.

3. The source play. If you can't change what the models cite, pay the places they already cite. Built In has earned its position here, showing up in 500 million AI search answers a month and selling employers presence inside that reach. And the same logic is driving a wave of Reddit specialists, full agencies now built exclusively around Reddit strategy, because ChatGPT, Google AI Overviews and Perplexity all count Reddit among their top 10 sources for answering questions.

4. Direct pipes into the models. This lane got its biggest validation yet in June when OpenAI added native job search to ChatGPT, pulling live listings from Appcast, Indeed and Upwork straight into the chat. Vendors like SonicJobs are building native plugin integrations of their own.

5. The career site comeback. What is old is new again, as an advisor put it to me this week. Suddenly everyone has a career site product, because the career site is the one property employers fully control in the citation economy. HireClix launched JobFlow AEO this spring to structure career site content for AI citation, and platforms like Happydance now pitch themselves as the meaning layer between your ATS and the models. The logic is simple: if the models are going to quote somebody, make your own site the most quotable thing they can find.

We track Walmart as a public demo account in our platform, which turned out to be timely after the Google for Jobs audit that made the rounds a few weeks ago.

Across just 10 candidate prompts, the models cited 85 unique pages from Walmart's career site alone. A single Reddit thread comparing Walmart and Amazon got referenced 33 times. And bls.gov outranked most of Walmart's own properties, with the careers page landing seventh among top sources.

That's what all five lanes are really selling access to. The models pull from everywhere, most of it isn't yours and almost none of it is the page you spent the budget on.

We took a hybrid path, and honestly the market pushed us there.

One of our first agency partners told us early on that the tool needed more data, more prompts, more actionable insight. We rebuilt the product around that feedback. Then another customer told us we were showing too much. Both were right, and together they taught me the data isn't the product. What people are buying is someone to tell them what matters for their individual situation.

So we do the monitoring and the optimizing, and we also do the work ourselves. We post on Reddit. We optimize and enrich XML feeds every day. We run our own newsletters and communities, and we hand customers what we learn from all of it as we go. This week a marketing leader extended with us and told me that being early to these learnings, and being able to share them with his content and sales teams, mattered as much as the metrics.

That's the honest state of this market. Nobody has years of benchmarks. The vendors worth working with are the ones learning in public and showing you their homework.

An advisor of mine has a word for how young categories like this one shake out: co-opetition. The monitoring tools need the agencies' client relationships. The agencies need infrastructure they can't justify building alone. The source plays, the pipes and the career site rebuilds all need someone measuring whether any of it works. Everyone in those five lanes is a competitor on the press release and a potential partner on the whiteboard, and the companies that figure out which is which will define this category. We're taking that path ourselves.

So here's my advice, even though I sell in this market: do your research. Ask every vendor or partner what's under the hood, where their data comes from and who interprets it for you. The press releases all say the same sentence. The plumbing and approach is where they differ.

And if you want to see what we're seeing, or you're an agency thinking through build versus buy versus partner, just reply. This is the conversation I'm having every day right now and I'd love to move the industry forward together.

More next week.

Best,
Summer Delaney
CollabWORK Founder and CEO

• Anthropic is adding watermarks to Claude's output — New Claude models now embed an invisible mark in generated text that survives copy and paste, part of EU AI Act compliance that OpenAI, Google and Meta have also signed onto. The nuance that matters for recruiting: a detected mark means content may have been processed by Claude, not written by it. Paste an executive's op-ed in for a proofread and it can carry the mark anyway. If your employer brand team uses AI anywhere in the content process, write the disclosure policy now, before someone runs your CEO's byline through a detector and writes the headline for you.

• Q3 AI Traffic Trends Report — Research from Adobe Digital Insights found AI-referred visitors to travel sites are 21% more engaged, spend 70% longer per visit and bounce 41% less than everyone else. AI traffic to retail sites now converts 54% better than non-AI sources, a full reversal from a year ago.

• Quality of Hire: a short history of a metric we still can't agree on — Gerry Crispin traces 114 years of trying to measure whether we hired the right person, from Taylor's shovels to the ISO standard he helped write. His closing warning: quality of hire measurement is only as sound as the authenticity of what enters the funnel, and AI-generated applications are degrading that input at scale. The volume era isn't just burying recruiters. It's corrupting the metrics we use to judge the work.

• Why people are dropping out of the workforce — June saw the biggest one-month drop in prime age labor force participation since 1976 outside the pandemic, roughly 720,000 people leaving work or the search in a single month. One laid off program manager logged more than 300 hours of applications and heard back on 15 roles before giving up entirely. When discovery is broken on both sides, people don't change tactics. They exit.

CollabWORK is hiring visibility infrastructure. We help employers get found, distributed, and re-engaged across the channels that now define how candidates actually discover jobs: AI search, professional communities and their own talent networks.

Three products, one thesis.

• Employer AI Discoverability monitors how you show up across ChatGPT, Gemini, Perplexity and other LLMs, and optimizes your job descriptions and XML feeds so you can actually be surfaced by the models candidates are searching.

• Community Distribution places roles inside 500+ vetted newsletters and communities including Morning Brew and 6AM City, reaching over 14 million professionals.

• Talent Community Activation turns the dormant ATS and CRM data you've already paid for into a living pipeline.

The companies winning at hiring right now aren't the ones with the biggest job boards budget. They're the ones showing up where great candidates actually are.

Find time to learn more. You can also find us on LinkedIn and X.