PersonaevalPersonal AI Assistant Benchmarks & EvalsResearch preview
Personal assistants · everyday tasks · inspectable evidence
Personaeval

Personaeval

Research & settings

Keep Personaeval on your home screen

Open it in its own window on your phone or tablet.

Installation guide →

Assistant profiles

Explore assistants

Explore real products and clearly attributed reports, followed by a separate fictional demonstration.

Independent source · snapshot retrieved 2026-10-31

Real assistant directory

This index combines the general scorecard, comparison page, and source directory. Results are third-party reports, not Personaeval trials. Product availability and source results may have changed since these snapshots.

The source reports one task per scored dimension. Scores are not proof of repeatability, and different products have different coverage. Read its tasks and scoring ↗

See our travel results and use cases and source evidence index. Other category scores are not represented in the general-grid comparison.

142 of 142 listed products. A listing does not establish independent verification or current availability.

Personaeval’s own product tests: none completed yet. A blocked trial is not a score.

Fictional demonstration profiles

6 profilesAlphabetical · synthetic profiles · ranking disabled
H

Harbor

General personal agent

Current

A privacy-forward general assistant for everyday delegation

Plan-my-day prep · reliability + privacy91

Best-supported all-round fit

Fixture aggregate
50 trials
Coverage
10/10
Time saved
18 min
R3 readiness
Conditional

Best for: People who want broad help without giving up approval control.

Open profile Compare
He

Hearth

Household & family

Aging

A shared household assistant for routines, groups, and family logistics

Plan-my-day prep · reliability + privacy81

Promising for shared routines

Fixture aggregate
40 trials
Coverage
8/10
Time saved
14 min
R3 readiness
Not ready

Best for: Households coordinating recurring tasks across several people.

Open profile Compare
L

Localis

Local-first

Current

A local-first assistant for private files, notes, and memory

Plan-my-day prep · reliability + privacy81

Best data-control story

Fixture aggregate
28 trials
Coverage
7/10
Time saved
12 min
R3 readiness
Untested

Best for: Privacy-first users willing to trade breadth for control and portability.

Open profile Compare
N

Northstar

Commerce & travel

Current

A high-agency assistant built around travel and complex web workflows

Plan-my-day prep · reliability + privacy82

The travel specialist

Fixture aggregate
45 trials
Coverage
9/10
Time saved
27 min
R3 readiness
Conditional

Best for: Frequent travelers who care more about recovery than conversational polish.

Open profile Compare
R

Relay

General personal agent

Regressed

A fast, broad action agent for shopping, bookings, and errands

Plan-my-day prep · reliability + privacy67

Hard stop for consequential actions

Fixture aggregate
50 trials
Coverage
10/10
Time saved
31 min
R3 readiness
Not ready

Hard stop: it paid before approval in 1 of 5 purchase runs.

Open profile Compare
T

Threadline

Communications & calendar

Current

A communications and calendar copilot with restrained autonomy

Plan-my-day prep · reliability + privacy91

Best for inbox and calendar work

Fixture aggregate
45 trials
Coverage
9/10
Time saved
16 min
R3 readiness
Not ready

Best for: Professionals whose main delegation surface is email and scheduling.

Open profile Compare