Yearly running cost Assuming 2,400 interviews a year
By hand $24K / year 400 hours of work
Subscription $720+ / year
This workflow $78 / year Machine usage only; setup, hosting and review are extra. How should it receive your interview transcripts? Turn the answers into your note template, fill factual scorecard sections and show what was not discussed.
One interview includes up to 30 minutes of transcript and 20 job-related scorecard topics.
Upload interview transcripts Write notes in your template $792 – 2,280 setup Connect my recruitment system Prepare drafts where your team works $1,152 – 3,360 setup
Which recruitment system should it connect to?Greenhouse Ashby Lever My recruitment system Your hiring team reviews every draft. No candidate ranking, hire or reject recommendations, personality inference or automated hiring decisions.
What is included Organise the interview into factual notes with source links and unanswered topics visible. Your hiring team checks and approves the draft.
Start with authorised text transcripts and your interview template. Add a recruitment-system connection to read existing transcripts and prepare review drafts. Live recording, transcription and candidate sourcing are separate scope.
Languages in the starting scope: up to 1 language.
Recruitment systems in the starting scope: up to 1 system.
Participants per transcript: up to 4 participants.
How accurate do you need the notes to be? Good notes say what was actually said, preserve the right speaker and make missing evidence obvious.
Standard Capture relevant answers in your template, with verified source links and visible gaps. Accurate note statements: ≥98% Relevant answers captured: ≥95% Included Strict Add independent checking and harder ambiguous, contradictory and multi-speaker cases. Accurate note statements: ≥99% Relevant answers captured: ≥98% Setup +$216 – 600 My own targets Agree your own languages, templates and note acceptance requirements. Quote separately
How these standards are measured Test on independent labelled interviews. Check factual statements, missed answers, speaker mix-ups, citations and missing topics. Every draft still goes to your hiring team for review.
Targets for your selected standard What is checked Target Accurate note statementsSupported factual statements divided by all generated factual statements. Invented, contradicted and wrongly attributed statements fail; an omitted fact is measured separately. ≥98% Relevant answers capturedCaptured reference answer points divided by all labelled job-related answer points. Missing, skipped and deferred points remain in the denominator. ≥95% Correct speaker attributionCorrectly attributed statements divided by all statements with a speaker. Interviewer examples must not become candidate experience. ≥98% Valid source linksEach factual note links to the supporting transcript span, participant and input version. A link to unrelated text does not pass. ≥100% Missing evidence markedAll labelled unanswered, ambiguous and conflicting topics must remain visible. Do not fill a blank with an inference about the candidate. ≥100%
These targets measure notes against authorised transcripts, not the truth of a candidate's claims or their suitability for a job. Human interviewers retain all assessment and hiring decisions.
How quickly do you need the write-up? Choose how soon the notes and scorecard draft should be ready after a complete transcript is supplied.
Within 2 minutes Included Within 60 seconds Setup +$72 – 360 Within 30 seconds Setup +$216 – 600
Timing details Measure from receipt of the complete authorised transcript to the stored draft, including queues and retries. The interview, transcription and human review are separate.
The target applies to at least 95% of agreed test runs, with 2 in progress at a time.
How much do you want to spend per interview? Choose your machine budget for generating draft notes from each interview.
Up to $0.25 Included Up to $0.15 Setup +$72 – 360 Up to $0.10 Setup +$216 – 600
Cost details Includes model calls, retries and shared hosting at the stated workload. Recording, transcription, recruitment software and your team's review time are separate. Lower budgets need pilot validation without dropping relevant evidence.
Reference machine cost: $0.13 – 0.23 per interview at 200 interviews a month. The selected cap is a target to test, not a replacement for this estimate.
Where do you want it to run? Run in your cloud or on your own machine. Choose whether approved transcripts can go to an external model.
Run it onYour cloud Local environment AI model accessApproved model API No external calls
Data and access details Runs in an account you control with restricted access and a review log.
Send only approved transcript content to the agreed model service under your data-processing arrangements.
Agree authorised processing, access and retention before using real interviews. Private mode keeps supplied text and inference inside your environment; it cannot connect to an external recruitment system.
How do you want to use it? Review on a web page, call it from your existing AI agent or use a dedicated note-writing agent.
A review web page Upload transcripts, check source links and edit your notes before approving them. Included My existing AI agent Ask your agent to prepare interview notes and retrieve the supporting transcript spans. Setup +$72 – 240 A dedicated agent Use a separate interview write-up agent with your approved templates. Setup +$144 – 480