Skip to main content

Search Power Modes — What They Do and When to Use Them

Search Power lets you choose how much data the system uses and which AI model powers your results. Pick the mode that fits your task for the best balance of speed, accuracy, and cost.

Written by Loukas Vouthounis
Updated over 2 months ago

Modes at a glance

1) Lite

Description: No communication data, small AI model

What it uses: Profiles, resumes/CVs, structured fields.

Best for:

  • Quick, lightweight searches

  • Early discovery/shortlisting

    Pros: Fast, predictable, low compute.
    Trade-offs: Less context → may miss signal that only exists in emails/notes.


2) Core

Description: All data, small AI model

What it uses: Everything Core uses plus communications (calls and meetings transcripts, WhatsApp messages ).

Best for:

  • Day-to-day searches where context matters

  • Matching that depends on interview notes, call summaries, or email content

    Pros: Richer context with good speed.
    Trade-offs: Small model may be less nuanced than Executive on complex queries.


3) Executive

Description: All data, leading AI model

What it uses: Full data scope (profiles + communications) with our most capable model.

Best for:

  • High-stakes searches (exec roles, niche skill mixes)

  • Ambiguous or multi-constraint prompts that need stronger reasoning

    Pros: Highest quality matching and explanations.
    Trade-offs: Heavier compute; may take slightly longer.


How to choose (quick rules)

  • Need speed and a clean baseline → Lite

  • Need context from emails/notes but want it snappy → Core

  • Need the best possible match quality and justification → Executive

Heads up: Magic Search is currently in Alpha with a limited testing group. If you’re in the Alpha and don’t see it yet, we’re rolling access out in batches-you’ll be added soon.

Did this answer your question?