Noa. The hiring team’s agent.
She answers, searches and acts, with 14 families of tools: candidates, roles, interviews, clients, HR, finance, analytics and more. Every action is checked against the user’s permissions.
Hiring, HR, operations and finance. One source of truth.
LinkMatch is not a recruiting tool with add-ons. It is one core that carries eight portals, an explainable match engine, AI agents and 60+ automated routines. All of them read and write the same data.
Recruiters, coordinators, account managers, HR, finance and management. Each sees the screens their permissions open.
Anyone looking for their next role, or being looked for by it. Built for the phone first.
Hiring managers and HR on the requesting side. They approve, interview and decide from inside the portal.
Whoever was hired, from day one. The profile built during hiring carries on with them.
Team leads and direct managers. Only what concerns their team, without the noise.
Agencies and vendors who submit candidates. Every submission is recorded, and candidate ownership is clear.
Anyone who knows someone good. They refer, follow along, and earn a bonus when the placement holds.
System administrators. Organizations, the shared pool, monitoring and control, with a separate audit trail.
Every candidate is scored against every role on eight dimensions. Each organization sets its own weights and knockout conditions, and the engine writes an explanation a non-recruiter can follow.
Two agents with a name and a job, and a row of capabilities working in the background. Anthropic’s Claude is the primary model, with OpenAI as fallback.
She answers, searches and acts, with 14 families of tools: candidates, roles, interviews, clients, HR, finance, analytics and more. Every action is checked against the user’s permissions.
She walks candidates through the career page and the portal: answers questions about roles, explains the process and guides them to apply. In Hebrew and English.
Reminders, syncs, checks and processing that run on a fixed schedule. What stalls gets a nudge, and what is about to expire gets an alert.
A PostgreSQL database with vector search, a job queue for background processing, and tenant isolation enforced at row level too.
The system was built right-to-left from the start: screens, emails, documents and signatures. The AI’s explanations are written in natural Hebrew as well.
Thirty minutes on the live system, with a real role of yours.