Glossary: AIRA Search key terms

AIRA Search introduces several terms that are specific to how the feature works. This glossary covers all of them.

An AI-generated profile built for each candidate in your database. It aggregates and structures data from notes, call transcripts (via Notetaker), uploaded resumes, LinkedIn data, Crunchbase enrichment, custom fields, and more into a format that AIRA can read and rank against. The Talent Graph is what enables qualitative ranking. Without it, AIRA can only match on structured fields. It updates automatically when new data is added to a candidate's profile.

A criterion from your search query that cannot be matched against a structured field. For example, "closed $1M+ deals," "scaled a team from scratch," or "strong communicator with executive presence." These criteria are used to rank candidates, not filter them. Each qualitative signal appears as a 🧠 chip in the interpretation bar.

A criterion that maps directly to a structured, indexed field in Recruiterflow: job title, location, company, years of experience, star rating, stage, and so on. These are hard filters: candidates either match or they don't. Quantitative filters run first and do not invoke the LLM. They appear as standard chips (without ✨ or 🧠) in the interpretation bar.

A structured filter that is AI-derived from the AIRA Talent Graph, not a raw indexed field. Examples include Career Stage, Promotion Velocity, Months in Function, Months in Industry, Inferred Skill Strength, Median Tenure, and Tenure Across Companies. These are structured like quantitative filters (hard-filter behavior) but the values are computed by AIRA, not entered by a user. They appear with a ✨ badge in the interpretation bar.

A percentage (0–100) representing how well a candidate's AIRA Talent Graph content aligns with the qualitative ranking signals in your query. The score reflects fit against your specific query, it is not a global quality score. The same candidate may score differently on different queries.

The evidence column in the NLS results table. For each candidate, AIRA Insight shows specific evidence from their Talent Graph explaining why they ranked highly for your query referencing roles, companies, dates, or achievements. For quantitative-only queries, it shows the matched field value. AIRA Insight never fabricates evidence; if insufficient data exists, it shows "Limited profile data — insights could not be generated."

The set of chips that appear in the AIRA chat side sheet after a query is submitted, showing exactly how AIRA parsed your input. Each chip represents one extracted criterion - either a quantitative filter, an AIRA Intelligence filter, or a qualitative ranking signal. You can edit or remove any chip before or after results load.

The persistent horizontal bar at the top of the NLS results table showing all active filter chips. Clicking "Edit" on any chip opens the side sheet for that filter.

The right-anchored panel that opens when a query is submitted. It shows AIRA's real-time reasoning, any clarifying questions, the final interpretation confirmation, and a full conversation history for the current search session. After results load, it can be minimised but remains accessible for conversational edits.