ContxtIQ Screening Copilot

From a broad idea to a sharp wedge

ContxtIQ started as a question: what would AI support look like if it actually lived inside a live scenario, not after it? The premise was simple: most AI tools summarize after the fact. We wanted to be present during the moment, where decisions are actually made.

That framing was too broad to ship, so we narrowed: script-based roles where there's a known structure to guide, grade, and extend. Recruiting became our first industry poke because the team had felt the pain firsthand. I'd sat through final-round loops with designers who clearly didn't have the skills they'd claimed; the same story played out with engineers and sales hires. The signal from a screening call wasn't surviving into the interview.

Technical screens, non-technical screeners

Screens exist to catch skill gaps before the loop. But the person screening an engineer or designer is usually a recruiter. They can hear a technical answer without being able to tell whether it's any good. The usual workaround is putting a technical person on the call.

• Can't grade the answer A recruiter can tell whether a candidate sounded confident, not whether the answer was technically right. The rubric lists topics, not what a strong answer sounds like.
• Missed follow-ups Knowing what to probe takes domain knowledge. Without it, a vague answer goes unchallenged and claimed experience goes untested.
• Context stays in the recruiter's head Hiring managers downstream inherit a short summary instead of evidence. Weak signal surfaces as a surprise in round 3.

A copilot that stays in the call with you

ContxtIQ is a desktop app that runs alongside the screening call. The recruiter picks a script template, starts a session, and works through questions one at a time. Behind the scenes, live transcription feeds a reasoning loop that grades answers, suggests follow-ups, and stays open for ad-hoc questions, all scoped to the candidate in front of you.

Primary Flow

Start a session, work the script, close the loop

The whole product bends around the screening hour. Open the app, pick a template, and run the call. Reasoning and follow-ups assemble themselves as you go.

Prototype to production, in one hand-off loop

I ran this end-to-end as CXO, not just as a design artifact but all the way into the production repo. The build stayed tight because each stage handed the living prototype forward instead of writing it down on paper.

Sharpen the wedge, then widen it

Near-term: deepen the recruiting experience with the design partner, tighten the grading reasoning, and turn session artifacts into something a hiring manager can actually pick up. Further out: the same live-copilot pattern extends to any script-based role with a structured conversation and a signal worth preserving, like sales discovery, support triage, or clinical intake.