WorkProofWorkProof, for City of Espoo

Show your work.Get the interview.

WorkProof is hiring that sees the work, not the name.

Try a 2-minute challenge

Made for Espoo's talent events for business and tech people. Demo data: all people and companies are invented.

One event in Espoo.

Demo dataEspoo Talent Day, so far.

challenges done
62
people invited
12
first talks
4
decisions made by AI
0
  • Technical
  • Commercial
  • Design and people
  • Handed in
  • Scored by a person
  • Invited to talk
  • A thicker line carries more work
Each line runs from the event to an employer. No names of people on this map.
62 pieces of work went to 10 companies. 34 scored by a person. 12 invitations.
CompanySectorDistrictPeople who took its challengeScored by a personInvited
Halla SupplyWholesaleLeppävaara632
Harjula EngineeringStructural engineeringEspoonlahti300
Hohka MaterialsBattery materialsKera442
Tuulos RoboticsRoboticsKera982
Kide SiliconSemiconductorsOtaniemi2000
Saarnivo GridEnergy softwareOtaniemi872
Kivira LabsPayments softwareKeilaniemi331
Revontuli GamesGamesKeilaniemi441
Kuulas StudioDesign and softwareTapiola331
Vesto ControlsBuilding automationEspoon keskus221

Her CV got no answer.
Her work got 9 of 10.

  1. The company scored her work before it knew who did it.
  2. Her name showed only after she said yes.

What Halla Supply reads

Automate the invoices first. 900 a month at two minutes each is about 30 hours of typing, and about 18 have a typo. The rule is the same every time: one delivery note, one invoice. Not the support chatbot first. Its mistakes go straight to customers, and it needs clean order data anyway. Week 1: read 200 old delivery notes with a text reader and match them to the order system. Week 2: the tool drafts the invoices, and a person approves each one. Risk: a wrong amount or the wrong customer. For the first month a person approves every invoice, and we compare the totals each week. Last spring I built the same flow for [hidden]. I live in [hidden] and studied at [hidden]. [hidden]

What Olena wrote

Automate the invoices first. 900 a month at two minutes each is about 30 hours of typing, and about 18 have a typo. The rule is the same every time: one delivery note, one invoice. Not the support chatbot first. Its mistakes go straight to customers, and it needs clean order data anyway. Week 1: read 200 old delivery notes with a text reader and match them to the order system. Week 2: the tool drafts the invoices, and a person approves each one. Risk: a wrong amount or the wrong customer. For the first month a person approves every invoice, and we compare the totals each week. Last spring I built the same flow for Kotiranta Oy. I live in Matinkylä and studied at Metropolia. Olena Kovalenko

4 details hidden before anyone read this

How an event works.

  1. The challenge opens before the event.
  2. The company scores the work.
  3. The top people get invited.
  4. A first talk happens at the event.
  5. A follow-up is agreed.

People score the work.
The AI never does.

A person reads the work without the name, scores it against 3 to 5 skills and writes why. The name shows only after the person says yes.

Work first. Names last.
People decide.

Try a 2-minute challenge

Built for the City of Espoo challenge, Junction 2026.

Demo data. All people and companies are invented.

Built with Next.js · React · TypeScript · Tailwind CSS · Vitest · Playwright · AI: Claude or Gemini, optional · Voice: Web Speech API or ElevenLabs, optional