An AI agent that acts on behalf of one party in a hiring negotiation. Employer-side agents screen, score, and negotiate on your behalf. Talent-side agents search and evaluate fit on the candidate's behalf.
Both operate within human-defined boundaries — you set the constraints, the agent works within them. Every decision is surfaced for human review before it becomes binding. Hiring agents don't replace recruiters; they eliminate the work that shouldn't require a recruiter.
Both the employer agent and the talent agent must independently evaluate and accept a match. This is distinct from one-way screening, where only the employer reviews candidates and candidates are passive objects in the process.
Bilateral matching eliminates most mismatched applications before a human reads them. When a match reaches your shortlist, both sides have already said yes — reducing time-to-interview dramatically and improving offer acceptance rates.
A structured job posting on hirinq, defined using the Open Job Protocol (OJP). A hiring queue contains the role spec, constraints (required skills, seniority level, location, salary band), and autonomy parameters that define how far your hiring agent can negotiate without human approval.
A hiring queue is the unit of work in hirinq. Each queue runs bilateral matching continuously until it either fills or is closed. You can have multiple open queues simultaneously, each with its own agent configuration.
An open, MIT-licensed schema for representing a candidate as structured, portable data. Skills, experience, availability, and consent — all in a standard format any platform can adopt, parse, and validate without proprietary transformation.
OTP is the candidate-side data contract in the hirinq exchange. Talent platforms publish OTP profiles; employer agents evaluate them. Because the format is open, a candidate's data isn't locked to a single platform — it travels with them, in their control.
The job-side counterpart to OTP. A structured spec for a role: required skills, seniority level, location, compensation range, and hiring constraints. OJP enables deterministic matching — scoring an OTP profile against an OJP spec produces a structured, auditable result, not a black box ranking.
OJP also enables GDPR-compliant processing. Because the spec is structured, the system knows exactly which data fields are relevant to each role, and can limit data access accordingly.
The bilateral exchange between an employer agent and a talent agent. Each round, both sides evaluate against their respective structured criteria. A negotiation reaches a match state when both sides accept the terms, or a no-match state when constraints on either side cannot be reconciled.
Every negotiation round is logged to the compliance vault with the full context: prompt, constraint check, score, and resulting state. The log is append-only and cannot be modified after the fact — required for EU AI Act Article 12 compliance.
The level of candidate data shared with an employer agent at each stage of matching. hirinq implements a three-tier model:
Metadata — anonymous profile: aggregate skills, seniority bracket, location region. No personally identifiable information. Available at initial matching.
Profile — structured professional data: work history, skills detail, availability. Still pseudonymous. Unlocked when both agents indicate initial interest.
Contact — identifiable data: name, email, LinkedIn. Only unlocked when both parties accept the match and the candidate consents to employer contact. Tiers are governed by candidate consent defined in their OTP profile.
An immutable, append-only log of every negotiation decision made in a hiring queue. Every log entry contains: the prompt sent to the model, the constraint check performed, the score or decision returned, and the resulting state transition.
The compliance vault is the technical implementation of EU AI Act Article 12 transparency obligations for high-risk AI systems. It is retained for 7 years on Pro plans and is exportable on demand. The vault is structured for auditability — not just storage.
The ability to substitute hirinq's default matching model with your own. Your model scores OTP profiles against OJP specs; hirinq handles bilateral clearing, audit logging, and compliance. The interface is minimal: OTP + OJP in, score + rationale out.
Your model stays in your infrastructure — hirinq calls your endpoint. You control data residency, model version, and latency. Compliance vault logging applies regardless of which model you use.
Anthropic's open protocol for exposing tools and context to AI models. hirinq's MCP server gives any compatible assistant — Claude, GPT, Gemini via MCP client — read/write access to queues, pipeline state, and negotiation tools.
With MCP, your AI assistant can report pipeline status, propose shortlists, and trigger negotiation rounds from within a normal conversation — without custom integrations or bespoke APIs. The tool definitions are typed and versioned.
Google's open protocol for structured communication between AI agents across vendor boundaries. hirinq implements A2A so employer and talent agents on different platforms can negotiate without a shared backend, proprietary bridge, or platform intermediary.
A2A defines the message envelope, task lifecycle, and error contract. Both sides speak the same standard, regardless of which AI vendor powers their respective agents.
A curated list of candidates who have passed bilateral matching and are ready for human review. Each candidate on a named shortlist has been accepted by both sides' agents and has consented to employer contact — meaning every name on the list is a qualified lead, not a filtered dump.
Named shortlists are the output of a hirinq hiring queue: the moment the exchange hands off to your team. From this point forward, the process is human-led.