On the second Thursday of each month, I host a free gathering of leaders of dozens of the world’s leading job boards, recruitment marketplaces, and organizations that are heavily involved in our industry such as programmatic and other job distributors, job board software providers, employment advertising agencies, recruitment process outsourcing organizations, and others.
Today’s guest was industry veteran and long-time friend, Jonathan Duarte, of GoHire. His company is helping to revolutionize the mobile-first job application process, candidate acquisition, and workforce engagement, with recruiting chatbots for small- and mid-market companies hiring between 50 – 5,000 hourly frontline workers each year.
As has become the custom for these monthly calls, I talked one-on-one for the first 25 or so minutes with Jonathan to help the group better understand his background and thoughts on this month’s topic. Today, the group explored whether AI agents would likely destroy the revenues that job boards earn from selling job postings as well as CV/resume searching. Perhaps this is a spoiler alert, but it seemed to me that the consensus was that job boards were far more likely to evolve by adding agents for employers and candidates than to be replaced by other organizations offering similar agents.
Want an invite for future meetings? Email me at Steven@CollegeRecruiter.com.
[embedded content]
AI-generated summary:
Monthly Roundtable Overview and Upcoming Schedule: Steven Rothberg convened the November meeting of the job board leaders monthly roundtable, noting that the group was founded by Matt Farah of nurses.co.uk and has reached 28 episodes (00:05:09). Steven Rothberg outlined their roles with College Recruiter and the Inside Job Boards and Recruitment Marketplaces podcast, and announced that Ward Christman from HR Tech Advisor will be the guest for the December meeting to discuss build-versus-buy decisions (00:06:23).
Jonathan Duarte’s Background in Job Distribution and Tech Platforms: Jonathan Duarte shared their extensive industry background, starting GoJob in 1996 as an early job board, building a job distribution network across over a thousand boards, and operating AppCast on the GoHire backend (00:09:39). Jonathan Duarte later built Good Hire, a background screening company acquired by Checker for $400 million in cash, and launched GOI, a conversational chatbot for recruiting, which currently supports multi-channel messaging and 22 languages under GoHire (00:13:44).
Integrating Job Boards with ATS via Model Context Protocols: Jonathan Duarte and Steven Rothberg discussed the historical limitation where job boards end at the resume submission into an applicant tracking system. Jonathan Duarte explained that emerging Model Context Protocols and open ATS integrations (such as from Greenhouse and Ashby) allow job boards to bridge this gap, asserting that a pre-screened candidate scheduled for an interview is worth ten times more than a standard applicant (00:17:04). Additionally, Jonathan Duarte noted using Claude Code to rebuild their entire tech stack over six months without bugs (00:19:19).
Candidate Adoption of AI Agents and Application Volume Friction: Discussing candidate behavior, Jonathan Duarte referenced a recent Greenhouse study indicating that 68% of candidates use AI, though agent-driven automated applications (search, match, and apply automatically) remain in the low single digits around 10% (00:20:33). Steven Rothberg and Jonathan Duarte agreed that as these automated volumes increase, recruiters and employers will inevitably introduce friction and blocks to manage the overwhelming influx of unqualified AI-generated applications (00:22:32).
Employer Sourcing Agents and Data Stale-ness Challenges: Jonathan Duarte noted that small and midsize companies are adopting employer-side sourcing agents like Juicebox and PIN to source, engage, and qualify candidates, although tools like PIN have faced blocking issues from LinkedIn (00:26:47). Jonathan Duarte and Steven Rothberg highlighted that scraped databases quickly become outdated within six months, posing significant hurdles for sourcing passive candidates effectively (00:29:21).
Candidate Intent versus Cold Outreach in Recruitment: Jim Durbin and Jonathan Duarte emphasized that job boards provide critical candidate intent, whereas cold outreach databases often yield cold leads that require intensive selling and proper infrastructure. Jim Durbin noted that while medical data sources like NPI provide comprehensive directories, modern recruiters frequently lack basic phone outreach training, making active job seeker intent uniquely valuable (00:33:15).
SMS Spam Filtering and Text Message Stop Mechanisms: Addressing recruiter text messaging, Mike Corso noted receiving frequent medical recruiting texts despite being retired, prompting Jonathan Duarte to explain carrier-level spam rules (00:36:09). Jonathan Duarte clarified that blocking numbers on iOS devices is ineffective, and users must reply with the word stop to trigger global carrier filters that permanently block spamming phone numbers (00:37:10).
Enterprise Readiness and Maintenance of AI Agents: Jonathan Duarte and Steven Rothberg discussed the operational challenges of AI agents, with Jonathan Duarte noting that while agents are simple to build, maintaining them for enterprise-grade stability is difficult because they frequently go off the rails (00:38:04). Steven Rothberg compared current AI capabilities to a reliable intern who requires guidance when processes shift, anticipating continuous improvements in upcoming months (00:39:06).
Intent Signals and Dormant User Reactivation: Joe Stubblebine shared that Lensa processes roughly 10 million unique monthly visitors and leverages hashed email signals to reawaken about 30,000 dormant users daily, pushing a fire hose of 6.6 million intent signals monthly to ATSs and job boards (00:40:03). Joe Stubblebine explained that passing real-time activity data allows organizations to optimize ad targeting via Data Management Platforms like Live Ramp and re-engage job seekers efficiently without relying solely on net-new user acquisition budgets. Steven Rothberg noted that this intent-driven approach reduces market noise, while Joe Stubblebine confirmed their data operations comply with regulations by operating strictly within the United States (00:42:00) (00:43:43).
Infrastructure Costs and Pricing Models for AI Agents: Responding to an inquiry from Mike Corso regarding startup costs, Jonathan Duarte explained that expenses for deploying AI agents and data tools are driven primarily by infrastructure setup and usage-based API calls rather than seat licenses (00:46:37). Jonathan Duarte noted that contact databases like Moltzet offer flat monthly rates starting around 27 to 500 dollars depending on usage scale (00:47:33).
Outbound Targeting Versus Traditional Job Postings: Jeff Taylor discussed the fundamental shift from traditional job postings to outbound targeting, noting that automated apply tools generate thousands of unqualified resumes per position and clutter applicant tracking systems with candidates who often do not realize they applied (00:50:20). Jeff Taylor described Boomband’s ingestion of 3 billion rows of data, monthly refreshes, and technographic mapping of company tech stacks against employee profiles to power outbound recruitment (00:51:34).
Boomand’s AI Agent Implementation and Future Outlook: Jeff Taylor detailed how Boomband combines an AI coach and agent to ingest candidate resumes and generate pre-published dossiers, while scout agents build prospect lists and automated outreach sequences for employers (00:55:06). Concluding the session, Jonathan Duarte emphasized that the future of recruiting will depend on successfully integrating technology, data, and candidate engagement (00:57:04).
Limitations of Current Matching Systems and AI Interviewers: Jonathan Duarte stated that current matching systems operate like playing 52 card pickup and lack defensibility. Furthermore, they argued that current AI interviewers should only be classified as screeners because true interviews require a bidirectional conversation where someone can answer questions about the company and the role. Jonathan Duarte predicted that advanced AI interviewers will emerge within 18 months to three years, and emphasized that all participants must inform candidates when AI is being utilized and how it functions (00:59:09).
AI Apply and Job Seeker Technology Developments: Karin Lash shared that they are working with AI apply to develop technology that creates more meaningful choices for job seekers, such as preventing the submission of 300 applications overnight (01:00:18). They noted that this work responds to partner requests for custom questions and knockout questions. Additionally, they explained that the platform filters out irrelevant jobs, while acknowledging that job seekers can be aspirational regarding the work they can perform (01:01:10).
Exploring Deeper Funnel Pricing Models Beyond Cost Per Application: Karin Lash discussed exploring deeper funnel conversations regarding cost per interview, referencing Jonathan Duarte’s earlier comment about a 10x increase in value. While noting that the standard payment model is currently Cost Per Application, Karin Lash stated that they are challenging this model to move deeper into the funnel (01:01:10). Steven Rothberg praised Karin Lash as an industry veteran with a strong, trusted reputation (01:02:08).
The Long-Term Viability of Job Postings and Resume Searching: Steven Rothberg asked whether AI agents will eliminate job postings and resume searching. Jonathan Duarte responded that such a change is not happening anytime soon, citing the enduring longevity of job postings over the last 20 years. Jonathan Duarte suggested that industry professionals should refortify their relationships with companies by going deeper. Steven Rothberg concluded the meeting and announced that the next session will take place a month later, featuring guest Ward Christman with HR Tech Adviser to discuss “build, buy or partner” (01:02:08).