Skill assessments & hiring workflows

People. Potential. Possibility.

Better hiring.

In your flo.

Build your hiring process as one connected path: real-work assessments, the details you need on file, rules you set, and a person deciding where it matters.

Get started free See how a flo works
EVALFLOREALPOTENTIAL.Less guesswork. More good hires.
  • 11

    assessment types that look like the job, from reviewing a code change to fixing a broken server.

  • 0

    seats counted, on any plan. Invite every reviewer you need.

  • 1

    unit of billing: a candidate run. Candidates never pay.

What EvalFlo is

Assess.Collect.Decide.

EvalFlo is where a hiring team assesses skills with tasks that look like the job, collects the applications and details it needs, and decides with rules it set and people it trusts. All of it in one connected process, called a flo.

Employers build the flo. Candidates move through it once and never pay. The team reviews the evidence and decides.

See the product

What a flo is

Your hiring process,
as one connected path.

Every stage is a node. Put the ones you need in the order that fits the role, and a candidate moves through them as one run.

See every stage
A1

Code review task. Show them a real code change and ask whether they would approve it. Fifteen minutes, and the highest signal per minute in the catalogue.

Assessments built around real work

Look beyond
the answer.

See what they can do.
Understand how they think.

Any model solves a closed-book puzzle instantly, so a score on one measures access to a model. The scarce skill now is telling right from plausible: the query that runs and returns the wrong number, the diff that reads well and swallows an exception. EvalFlo ships eleven assessment types built around that: review a code change and say whether you would ship it, write the prompt that has to get an AI to do the job, untangle messy data, fix a broken server with real commands.

Every assessment type

Code review task

Show a diff, ask if they would ship it

Messy-data SQL

The trap is in the data, not the syntax

Prompting / delegation

They write the instructions, not the answer

Auto-graded coding

Classic DSA, graded instantly

Connected history

Optional past session logs. Never required, never a gate

MCQ / objective test

Timed, auto graded

Debug puzzle

Find the bug against the clock

No-code pipeline puzzle

Build a system from blocks

Browser mini-games

Quick puzzles that quietly measure skill

Simulated terminal

Fix a broken system with real commands

Broken-system triage

Work out what is wrong first

How it works

Three decisions.
Then it runs.

  1. 01

    Choose the stages

    Start with how candidates get in. Add the assessments the role actually needs, the details you want on file, and where a person should look.

  2. 02

    Set the rules

    Branch on any score or countable event. Send a candidate to review, to the next task, or to an outcome. Every automated decision is logged with the threshold that made it.

  3. 03

    Review the work

    Scores sit beside the work behind them: the edits, the pastes, the sequence of events. Invite every reviewer you need. Seats are never counted.

Integrity, honestly

Signals for a reviewer.
Never a verdict.

We count what a browser can actually see, keep camera findings for human eyes only, and say plainly what no tier detects.

What we count

  • Paste count, largest paste, total pasted characters
  • Tab switches, focus lost count and seconds
  • Fullscreen exits, devtools opened, device changes
  • Cross-candidate similarity on the event sequence

What only a human sees

  • Camera snapshots and the contact sheet
  • Event-triggered clips
  • Face-missing findings
  • Speech segments and durations

What no tier detects

  • A phone sitting next to the keyboard
  • A second laptop
  • A person off camera
  • Remote desktop, without a native client we have not built

Pricing that counts runs, not people

Bring every reviewer.
Pay for the runs.

One subscription with candidate runs included. Free starts at 25 runs a month. Nothing to top up, nothing that expires.

See the plans
  • SeatsNo plan counts users, on any tier including free. When reviewers cost money the first thing a cost-conscious admin does is delete reviewers and lean harder on auto-rejection.
  • Candidate accessCandidates never pay to sit an assessment an employer sent them. Ever.
  • Your dataAn organization can always export its own results, on any plan, including after a downgrade.

Two paths. A shared belief in potential.

People make
the possibilities.

A few good questions.

No. EvalFlo is an evidence layer. Scores sit beside the work behind them, integrity findings arrive as tiers with the evidence attached, and a person on your team makes the call. You can wire automated rules on scores and countable events if you choose to; we warn where that is risky and log every decision either way.

Free to start, no card, no seats

Your next great team.
Your kind of flo.

Create your account, bring every reviewer you need, and run your first candidates on the free plan. Upgrade when the volume asks for it.

Create your account