Documentation

Documentation·How Jedah works

Everything you need to start scoring interviews with confidence.

Quickstart

You can run your first scored interview in about ten minutes. Here's the full path from install to export.

Install Jedah from the Zoom Marketplace.

Open the Zoom Marketplace listing and click Add. Admins can install team-wide for an org; individual interviewers can install just for themselves. If you're on a managed Zoom plan, your IT admin will need to approve the app first.

Install scope Admin install · org-wide, requires Zoom admin approval
User install · just you, available in your next meeting
Open the Apps panel in your next Zoom meeting.

Click Apps in the Zoom toolbar and select Jedah from your installed apps. The side panel opens to the right of the meeting window.

Zoom toolbar Mute · Video · Participants · Chat · Apps ← · More
Paste the candidate's resume or LinkedIn URL.

Jedah analyzes their background while you talk — flagging gaps, surfacing strengths, and pre-populating context for the rubric. You can also upload a PDF resume or paste raw text.
```
{
  "input": "https://linkedin.com/in/candidate-name",
  "context": "Senior PM role · B2B SaaS",
  "rubric": "lou-adler-10"
}
```
Conduct the interview as normal.

Live transcript appears in the side panel. The rubric populates in real time as Jedah scores responses against the Lou Adler 10-Factor and Dennis Wood frameworks. Confidence scores and source quotes are visible for every dimension so you can sanity-check the model's reasoning.

Live rubric Outcomes ····· 4 / 5 · high confidence
Energy ······· 3 / 5 · medium confidence
Fit ·········· — · awaiting signal

Export the rubric as XLSX.

One click exports a color-coded scoring sheet structured for ATS handoff. Each dimension includes the score, the supporting quote, and the model's confidence level. Done.

jedah-evaluation-2026-06-08.xlsx
  ├─ Rubric          (color-coded scores)
  ├─ Source quotes   (per dimension)
  ├─ Transcript      (if opted in)
  └─ Metadata        (interviewer, candidate, role)

How scoring works

Jedah scores against two complementary frameworks. They're independently respected and, together, give you both a performance read and a potential read on every candidate.

Lou Adler 10-Factor

The standard performance-based hiring rubric used by thousands of recruiting teams. It evaluates candidates across ten dimensions — outcomes, energy, fit, recognition, leadership, team skills, problem-solving, planning, technical competence, and motivation — anchored to real work products rather than personality traits. Jedah scores each dimension against quotes from the live interview, surfacing the exact language that drove every score so you can audit the reasoning.

Dennis Wood Human Potential Framework (HPF)

A 7-dimension model for assessing growth potential rather than just current performance. HPF looks at cognitive agility, learning velocity, self-awareness, resilience, drive, interpersonal range, and values alignment — useful when the role you're hiring for will look different in twelve months than it does today. Jedah operationalizes HPF for live interviews so you don't need a separate assessment battery.

Both frameworks are credited to their original authors — Lou Adler's Performance-Based Hiring methodology and Dennis Wood's Human Potential Framework. Jedah operationalizes them inside the interview itself; it does not replace the underlying theory or claim authorship of either model.

The AI-answer detector

Candidates reading from ChatGPT mid-interview is now a common reality. Jedah flags it.

Jedah runs a 5-signal heuristic across the live transcript to detect responses that appear to be read from an AI assistant rather than spoken from memory or experience. When the heuristic fires, the interviewer sees a discreet flag in the side panel — not an accusation, just a prompt to dig deeper with a follow-up question. We deliberately don't publish the specific signals to avoid teaching evasion techniques.

Privacy by design

Candidate trust is non-negotiable. Three commitments anchor everything we build.

Audio is processed in transit, never stored at rest. Read the data flow →
Transcripts retained only on interviewer opt-in, with a 90-day default lifecycle. Retention policy →
Candidates notified via in-meeting banner before analysis begins. Consent model →

Integrations

Live today and on the roadmap.

Zoom

Live

Greenhouse

Coming soon

Lever

Coming soon

For developers

Architecture & API access

Jedah does not currently offer a public API. We're working with design-partner customers to define the surface area before opening it — if you have a specific integration use case, write to support@jedah.ai and we'll loop you into that conversation.

Internal architecture, for buyer due-diligence: Jedah's backend is Node/Express hosted on Fly.io with regional failover. Real-time audio is captured via Zoom's RTMS (Real-Time Media Streams) feed; scoring is performed by Anthropic Claude with structured prompts per rubric dimension. All inference happens server-side, no model weights or candidate data leave our infrastructure. SOC 2 Type II is in progress.

Help & contact

Need a human? Head to /support or email support@jedah.ai — we respond within one business day.