HrFlow.ai
HRConnect AI agents to HrFlow.ai, an AI talent data platform for parsing and searching HR data. Agents parse resumes, index profiles and jobs, and search across sources so candidate and job data becomes structured and searchable.
What This Integration Enables
HrFlow.ai is not an applicant tracking system and it is a mistake to shop for it as one. There is no pipeline, no stage, no hire or reject. It is the data layer underneath those things: the place where a resume stops being a document and becomes a record other systems can query. That distinction is why it belongs in a flow rather than in a recruiter's browser tab.
The workspace is organized around two containers. A Source holds candidate Profiles, and a Board holds Jobs. Agents list both, then parse a resume from a public file URL straight into a Source or retrieve the parsing object attached to an existing Profile. They can also index a structured Profile directly, which is the right path when candidate data already exists in an applicant tracking system or a form and only needs to become searchable. Jobs are indexed into Boards the same way. Search Profiles and Search Jobs run keyword, name, and pagination filters scoped to the Sources or Boards you name.
Underneath sits a set of text intelligence operations that do the work no keyword search can. Parse Text extracts over fifty HR data points from a raw block of text and returns named entities with their positions, which lets an agent read a job description or a candidate summary the way it reads a resume. Find Synonyms returns semantically related words with a similarity score, and Embed Text produces cross lingual vectors for semantic matching and clustering. Those three operations are what let a flow reason about talent data instead of merely storing it.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Resume Intake Into a Searchable Source
Resumes arrive as attachments in a shared mailbox and land in cloud storage. The agent picks up each file from Google Drive, takes its public URL, and runs Parse Resume into the appropriate Source. HrFlow.ai extracts contact details, experiences, education, skills, languages, and certifications, and returns the parsing object together with the created Profile. The agent writes the Profile key back to the candidate record in the applicant tracking system so the two stay linked, and posts a short summary of what was extracted to the sourcing channel in Slack. A folder of documents becomes a queryable pool without anyone typing a field.
Requirements to Candidate Shortlist
A hiring manager writes the role requirements in prose. The agent runs Parse Text on that prose to lift out the skills, job titles, and seniority signals as structured entities, then indexes the role into a Board with Index Job so it can be matched and reported on later. It runs Find Synonyms on each required skill to widen the vocabulary, presents the expanded term set for approval, and only then runs Search Profiles across the named Sources. Matching profiles are pulled with Get Profile for their full structured data and written to a shortlist sheet with the parse evidence beside each name. The manager sees why each candidate surfaced.
Keeping an External System's Data Honest
Candidate records already live in an applicant tracking system, entered by recruiters over several years with inconsistent field discipline. Rather than migrate, the agent reads each record and runs Index Profile to mirror it into a Source with a stable reference, so it becomes searchable without changing the source system. Where a resume is attached it also runs Parse Resume and compares the parsed fields against what the recruiter typed. Disagreements, a seniority label that the resume does not support or a skill list that omits half the person's history, are collected into a review queue rather than silently overwritten. The mirror improves search without pretending the original data was clean.
Human-in-Loop Highlight
The riskiest thing an agent does here is invisible. When Find Synonyms expands "project manager" into a term set and that set feeds Search Profiles, every candidate whose resume used a word the expansion missed is excluded, and nothing in the result tells anyone they were there. A bad hire is visible and correctable. A pool that quietly never contained the right person is neither. So the agent treats the search vocabulary as a rule that needs approval, not an output to review afterward. Before a standing sourcing search runs, it posts the proposed terms to the sourcing lead: "Expanding [required skill] to these 9 terms with similarity scores attached. These 4 related terms scored below the cutoff and will be excluded, so profiles using only those words will not appear. Approve the set, add a term, or lower the cutoff." The lead owns the vocabulary. Parsing, indexing, and retrieval run on their own, because they add records rather than deciding which ones are never seen. This is the human-in-the-loop placed one step earlier than usual, on the rule instead of the result.
Agent Capabilities
12 actionsSources and Boards
2- List Sources Retrieves the Sources in your HrFlow.ai workspace. A Source is a container that holds candidate Profiles, for example a resume database, a job board integration, or an applicant tracking connector. Results are paginated and filterable by name. Use a Source key with the profile parsing, indexing, and search operations.
- List Boards Retrieves the Boards in your HrFlow.ai workspace. A Board is a container that holds Jobs, for example a careers page feed or a job board integration. Results are paginated and filterable by name. Use a Board key with the job indexing and search operations.
Profiles
5- Parse Resume Parses a resume document from a public file URL and creates a candidate Profile in the chosen Source. The document is fetched and uploaded to HrFlow.ai, which extracts contact details, experiences, education, skills, languages, and certifications. Supported formats include PDF, DOC, DOCX, ODT, RTF, PPT, PPTX, and common image types. Returns the extracted parsing object with the created Profile.
- Get Resume Parsing Retrieves the resume parsing object attached to an existing Profile, returning the structured data extracted from the candidate's resume. Identify the Profile by its key or by the custom reference assigned when it was created. Used to compare a parse against what a recruiter typed by hand.
- Index Profile Indexes a structured candidate Profile directly into a Source without parsing a document. Use it when candidate data already exists in an applicant tracking system, a form, or another platform and only needs to become searchable. The Profile object must include an info block with at least a full name.
- Get Profile Retrieves a single indexed Profile from a Source with its full structured data: info, experiences, education, skills, languages, tags, and metadata. Identify the Profile by key or custom reference.
- Search Profiles Searches indexed Profiles across one or more Sources with keyword, name, and pagination filters. Returns matching Profiles with their key, reference, and summary info. Scope the search by naming Source keys. Results are paginated and sortable by creation date.
Jobs
2- Index Job Indexes a structured Job into a Board so it can be searched and scored against candidate Profiles. The Job object must include a name, and may include a reference, URL, summary, sections, skills, languages, tags, and location details.
- Search Jobs Searches indexed Jobs across one or more Boards with keyword, name, and pagination filters. Returns matching Jobs with their key, reference, and summary details. Scope the search by naming Board keys.
Text Intelligence
3- Parse Text Extracts over fifty HR data points from a raw block of text such as a resume, a job description, or a candidate summary. Detects named entities including skills, job titles, companies, schools, locations, emails, and phone numbers, and returns them with their positions in the text. Choose a faster model or a more accurate one, and optionally shape the output as a Job or Profile object.
- Find Synonyms Finds words semantically related to a given word using HrFlow.ai's HR-native language model, returning a ranked list with similarity scores between 0 and 1. Used for query expansion before a search and for building skill taxonomies. Because the expansion decides which profiles a search can reach, the resulting term set is routed for approval.
- Embed Text Generates an HR-native, cross lingual embedding vector for a block of text using HrFlow.ai's encoder models. The vector powers semantic search, matching, clustering, and retrieval across more than forty languages. Choose a higher dimension model for accuracy or a lower dimension model for speed.
Frequently Asked Questions
What can FlowRunner do with HrFlow.ai?
FlowRunner agents can run List Sources, List Boards, and Parse Resume in HrFlow.ai, plus 9 more actions.
Does connecting HrFlow.ai to FlowRunner require OAuth?
No. HrFlow.ai connects to FlowRunner with an API key, no OAuth flow required.
Can HrFlow.ai trigger a FlowRunner workflow automatically?
HrFlow.ai doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with HrFlow.ai
$100 in credits. No card required. Connect in minutes.