Skip to content
Cupivec

Product

Matching that shows its work.

Cupivec represents people as structured, multi-dimensional profiles, compares them against criteria you define and explains every result, so that people stay in charge of every placement.

Request a walkthrough

Capabilities

Precise enough to trust. Clear enough to question.

Every part of Cupivec is designed so the people using it, and the people it describes, can see what's going on.

01

Multi-dimensional profiles

Profiles are structured vectors with named dimensions you can inspect. Matching works on dimensions, not buzzwords, so two people with the same job title can still show very different fits.

  • Named, inspectable dimensions
  • Source-linked attributes
  • Profiles update as people grow
02

Explainable fit

Each result carries a per-dimension breakdown, plain-language reasons and visible gaps. Reviewers can see why a match surfaced, and so can the person it describes.

  • Per-dimension fit breakdown
  • Plain-language reasons and gaps
  • Shareable with the person matched
03

Your criteria, your weights

Define required and preferred criteria, then weight them to suit your context. Every change is versioned, so you can always tell which configuration produced which result.

  • Required vs. preferred criteria
  • Versioned weight configurations
  • No hidden global standard
04

Humans make the call

Results arrive as suggestions with evidence, never as automatic outcomes. Review steps and decision logs keep accountability clear and auditable.

  • Suggestions, not verdicts
  • Built-in review steps
  • Decision audit trail

How matching works

Five stages. Every one of them inspectable.

Data flows in one direction, from consented inputs to a human decision, and each stage leaves a record you can review.

  1. 01

    Consented data

    CVs, assessments, preferences and logistics, each shared with permission.

  2. 02

    Profile vector

    Structured into named dimensions, each traceable to its source.

  3. 03

    Your criteria

    Required and preferred criteria, weighted by your team.

  4. 04

    Explained fit

    Per-dimension breakdown, reasons and gaps for every match.

  5. 05

    Human decision

    A reviewer decides, and that decision is logged with its evidence.

API

Built to sit inside your product.

Cupivec is infrastructure, not another dashboard. Send profiles and openings over a JSON REST API and get explained matches back for your own interface.

  • POST/v1/profilesCreate or update a consented profile
  • POST/v1/openingsDescribe a role, place or program
  • POST/v1/matchesGet explained fit for a set of profiles
  • HOOKmatch.completedReceive asynchronous results

These are illustrative examples. The full API reference is shared during onboarding.

match.json
POST /v1/matches{  "opening": "opn_platform_eng_lisbon",  "criteria_version": 3,  "criteria": {    "skills":     { "weight": 0.35, "require": ["go", "kubernetes"] },    "experience": { "weight": 0.25, "min_years": 4, "prefer": ["on_call"] },    "work_style": { "weight": 0.20, "prefer": ["async_first"] },    "logistics":  { "weight": 0.20, "location": "Lisbon", "mode": "hybrid" }  },  "profiles": ["prf_3091", "prf_1187", "prf_5520"],  "explain": true}
200 · application/json
HTTP/1.1 200 OK{  "opening": "opn_platform_eng_lisbon",  "results": [    {      "profile": "prf_3091",      "fit": 92,      "breakdown": {        "skills": 96, "experience": 84, "work_style": 90, "logistics": 96      },      "reasons": [        "covers 2 of 2 required skills",        "based in Lisbon, open to hybrid"      ],      "gaps": ["less on-call experience than preferred"],      "decision": "pending_review"    }  ]}

Data & privacy

Consent, minimization and deletion are part of the data model.

Matching people means handling personal data, so we treat privacy as an engineering constraint rather than a settings page.

Consent first

Profiles are built only from sources a person has agreed to share, and that consent is stored with the data.

Data minimization

We ask for the dimensions a match needs and nothing more. Fewer fields mean less risk.

Purpose limitation

Data shared for one matching purpose isn't reused for another without fresh consent.

No protected attributes as criteria

Characteristics such as age, gender, ethnicity or religion can't be used as matching criteria.

Retention & deletion

You set retention windows per data type, and deletion requests propagate through every profile.

Audit trail

Configurations, results and human decisions are logged so every outcome can be reviewed later.

Read our privacy policy for how this website handles data.

FAQ

Common questions.

See explainable matching with your own criteria.

We'll walk you through a scoped pilot covering your use case, your data sources and the criteria your team cares about.

Contact us