What happened
On October 28, 2026, Firecrawl announced Directory, a new Alexandria collection aimed at recruiting work. Firecrawl describes Alexandria as a library that combines data providers, specialized indexes, and the live web for agents. Directory contains more than 240 million professional profiles, including employment histories and work contact information.
The collection is available to agents in ChatGPT and Claude through the Firecrawl plugin. Firecrawl says its providers include Apollo, FullEnrich, and Data Legion. An agent can search people using job title, company, location, and keywords, then retrieve available information on a person’s current role, former employers, professional background, and work contact details. The company positions this as a way to search for relevant experience rather than only matching a title.
The product examples are intentionally conversational. A hiring team can ask for senior backend engineers in New York who have worked on infrastructure at scale, for example. Firecrawl says that one such free search returned 142 matches. Users can also provide a job description and ask an agent to organize relevant profiles, compare candidates against stated requirements, and refine the pool by geography, technical background, or similar past roles.
Directory is not presented as a self-contained assessment system. After finding a potential candidate, the agent can use Firecrawl’s web-search capability to look for public technical writing, conference talks, open-source contributions, interviews, and projects. Firecrawl frames that additional material as context for understanding work beyond job titles and for preparing better interview questions. It explicitly says the product is not meant to automate hiring decisions or replace conversations with candidates.
Searches in Directory are free. Credits apply when a user retrieves detailed professional information, including available contact details, and each account receives 1,000 free credits per month. Firecrawl also provides a terminal-agent setup path: users can install its skills with npx -y firecrawl-cli@latest init --all --browser, then restart their agent so it loads them. The source links to plugin installations for both ChatGPT and Claude.
What the archive adds
This is a new product event within a known Firecrawl data-library story. The earlier note on Firecrawl’s free Government & Legal Index covered another Alexandria collection exposed to ChatGPT and Claude. Directory extends that same distribution model from government and legal material to professional-profile research.
Trend signal
The launch supports Model answers become task interfaces. The notable product surface is not a standalone people-search page alone: Firecrawl places a bounded search-and-retrieval capability inside assistant conversations, where a role description can become a search, a comparison, and a working shortlist.
Connections to other events
The Government & Legal Index launch established that Firecrawl was packaging specialized external data for use through ChatGPT and Claude. Directory applies the same approach to a different task: finding and researching prospective candidates. In both cases, the assistant is a front end for a scoped data collection rather than merely a generator of prose.
Why it matters
For teams building with AI, the useful change is the compression of a multi-tool research workflow. A recruiter or hiring manager can express the target profile in natural language, alter constraints as they inspect results, and ask the same agent to organize a shortlist. Technical teams can use the same pattern for talent mapping, sourcing specialists for a project, or identifying people with particular operational experience.
The implementation boundary still matters. Free search is not the same as unrestricted profile enrichment: detailed information and available contact details consume credits. Teams should therefore define which fields are actually needed before letting an agent retrieve them, and should treat output as research input for review rather than as a hiring recommendation.
What happens next
- Firecrawl adds further specialized Alexandria collections that can be called from assistant conversations. (6 months, 65%). Watch for: new category pages or provider announcements.
- Teams begin measuring recruiting-agent workflows by shortlist quality and reviewer time rather than search volume. (12 months, 55%). Watch for: case studies with screening or response metrics.
Limits and uncertainty
The announcement is a vendor description and does not independently establish the completeness, freshness, accuracy, geographic coverage, or permissible-use conditions of the 240 million profiles. It does not state how frequently individual records are updated, what contact fields are available in practice, or how providers’ underlying data differs. The source gives an example result count, but not a benchmark against conventional recruiting databases or evidence that the system improves hiring outcomes. It also does not resolve privacy, consent, or local employment-law questions that may arise when a team uses professional data for outreach or screening.
Skill to build now
Run a small, reviewable sourcing experiment. Write a role brief with must-have experience, preferred company types, location constraints, and explicit exclusion criteria. Ask the agent for a limited shortlist, then have a human verify each person’s relevant work from public artifacts before any outreach. Record how many profiles were useful, how many details required credits, and where the agent’s interpretation of the brief was too broad. This yields a practical quality bar before the workflow is connected to a larger recruiting process.
Reader takeaway
Directory gives ChatGPT and Claude agents a new Firecrawl-backed people-search surface: more than 240 million professional profiles, free discovery searches, and credit-based detail retrieval. Its near-term value is not automated hiring; it is faster, conversational creation of a candidate-research starting point that people still need to check.
Open question
How accurate, current, and jurisdictionally usable will Directory’s profile and contact data prove across the markets where teams actually recruit?