Skip to content

Open decisions surface first

Styleguide

Living brand book. Draft direction, pick samples here, then we lock DESIGN.md.

Component library, pull from here onlyPainterly-sky · single accent

01

Heroes, WebGL shaders

One hand-written WebGL hero background everywhere: CloudBackground — a domain-warped fbm cloud field drifting on a lighter-blue sky. The homepage runs it at text-hero scale; every sub-page gets the same sky via ShaderHero (the old Digital Aurora is retired). Paper text lifted by a drop-shadow + the field's built-in bottom-left scrim, with a static CSS cloudscape fallback for first paint / prefers-reduced-motion / no-WebGL / mobile.

ShaderHero · the shared cloud skydrifting clouds, lighter blue

The execution gap

Close the execution gap on people decisions.

01b

Blocks (patterns)

Composable section building blocks (src/components/patterns/*). Pages compose these instead of hand-rolling layout, so spacing/colour/structure change in one place. Colour via the shared accent system (lib/accents.ts) — a single hue (indigo), with teal reserved for the 'after'/outcome meaning. Composite section blocks (own their <Section>): SplitStatement (editorial split), CtaSection (closing CTA, optional mesh bookend).

Section surfacespaper · elev · ink · mesh, the rhythm vocabulary
paper
elev
ink
mesh
SectionHeading, accent marker + eyebrow + headline + action

How Everday works

One core, productized.

Steps, numerals in the single accent (indigo)

  1. 01

    The core

    Two intelligence engines and a knowledge graph.

  2. 02

    Playbooks

    Click-and-go implementations of common people-decisions.

  3. 03

    Managed services

    When you want the outcome guaranteed.

MarkerList, on paper
  • Skills Intelligence engine
  • Market Intelligence engine
  • Knowledge graph

MarkerList, onDark

  • Skills Intelligence engine
  • Market Intelligence engine
  • Knowledge graph
CardGrid + Card, float on the elev surface

Skill-based hiring

Before

Manual, slow, hard to defend.

After

Ranked, evidence-backed, in minutes.

Bench utilization

Before

Manual, slow, hard to defend.

After

Ranked, evidence-backed, in minutes.

Skill taxonomy

Before

Manual, slow, hard to defend.

After

Ranked, evidence-backed, in minutes.

Eyebrow accents — one hue (indigo), teal for the outcome

The lineup

How it works

The problem

The outcome

Steps with role labels (You=indigo / We=teal — the allowed role distinction)

  1. You

    Own the relationship

    You sell and stay the face of the work.

  2. We

    Power the engine

    Our core and agents do the AI lift.

  3. Together

    Co-implement

    We pair on the first pilot end-to-end.

NumberedList, a vertical process path (tone-coloured numerals, optional meta)

  1. 01

    Name a client

    Pick one active client with a real people-decision problem.

  2. 02

    Run a paid, scoped pilot

    We co-implement a playbook end-to-end.

    4–8 weeks

  3. 03

    Formalize the partnership

    The partnership rides on proof, not a signature.

DefinitionList, term + description

  • A pilot is 4–8 weeks.

    Scoped, paid, with a real outcome.

  • Playbooks are pre-productized.

    Click-and-go, wired to your ATS/HRIS.

DefinitionList, term only (checklist)

  • A defined deliverable, not a platform to operate.

  • Agents running inside your existing tools.

  • Deterministic, traceable, explainable evaluations.

01c

Showcase elements

Non-text visuals (src/components/showcase/*) that break the text wall and give pages a text↔visual rhythm. OUTCOME visuals (Transformation / Comparison / PullQuote / StatBand / FlowDiagram) suit every page, they show the result, not a tool. PRODUCT visuals (ProductFrame / OutputMock, a UI you'd log into) are /platform ONLY; the rest of the site owns the outcome, not a dashboard.

Transformation, the outcome shift (old way struck through → result)
  • Pilots that produce slide-decksDecisions in production
  • A defensible guessEvidence behind every call
  • Months of bespoke buildLive in 4–8 weeks
Comparison, the wedge made visual (Everday column highlighted)

  • Software vendors

    Hand you a tool, then walk away.

  • Consultancies

    Rebuild from scratch, every time.

  • Everday

    Deliver the outcome on a core that already works.

PullQuote, a display-scale outcome line (narrow, centered)
We own the AI risk. You get the outcome.
The Everday model
ProductFrame + OutputMock, PRODUCT UI · /platform ONLY
everday · skill-based hiring

Ranked shortlist

Senior Backend Engineer

  • A. de Vries

    GoPostgresDistributed systems
    94%

  • S. Bakker

    GoKubernetes
    89%

  • M. Janssen

    Java → GoAPI design
    82%

  • L. Visser

    PythonData pipelines
    76%

Every score traces to evidence, deterministic, not a guess.

FlowDiagram, explain the engine / a process visually

  1. Intelligence engines

    Skills + Market

  2. Knowledge graph

    Compounds

  3. Agents

    Deterministic

  4. Decision

    Evidence-backed

StatBand, proof points (real, cleared facts only)
4–8 wks
pilot to outcome
ECHT
partnership live
Traceable
every evaluation

01d

New blocks, Tier 1

Layout + structural upgrades on the indigo→teal system. BentoGrid breaks the rigid 3-equal-column rhythm with mixed-size tiles. ComparisonTable is the structured 'why us' matrix (the table upgrade of the prose-only Comparison, both kept). CountUpStatBand animates numeric proof points on scroll (non-numeric values fade). FeatureMockupSplit pairs a claim with a generic visual slot, alternating sides. Background pattern primitives texture dark sections (currentColor-tinted, deterministic, server-rendered).

BentoGrid, mixed-size tiles (sm / lg / wide / tall)float on the elev surface

The core

One engine, productized.

Two intelligence engines and a knowledge graph that compounds with every decision.

The outcome

Decisions, not slide-decks.

Ranked, evidence-backed, traceable.

Speed

Live in 4–8 weeks.

Inside your stack

Runs in your ATS / HRIS.

Proof

ECHT partnership, live in production.

The model proven with a real partner, not a pilot that ends in a deck.

ComparisonTable, capability matrix (Everday column highlighted indigo)
CapabilitySoftware vendorsConsultanciesEverday
Delivers an outcome, not a toolNoYesYes
Live in 4–8 weeksYesNoYes
Runs on a core that already worksNoNoYes
Evidence behind every decisionPartialManualTraceable
Owns the AI riskNoNoYes
CountUpStatBand, numeric figures count up on scroll; 'ECHT' fades in
0%
evaluations traceable
4–0 weeks
pilot to outcome
ECHT
partnership live
FeatureMockupSplit, claim beside a visual slot (reverse alternates sides)

Skill-based hiring

Every applicant, scored against the job.

Ranked shortlists in minutes, with the evidence behind each one, deterministic, not a black box.
See the playbook
everday · skill-based hiring

Ranked shortlist

Senior Backend Engineer

  • A. de Vries

    GoPostgresDistributed systems
    94%

  • S. Bakker

    GoKubernetes
    89%

  • M. Janssen

    Java → GoAPI design
    82%

  • L. Visser

    PythonData pipelines
    76%

Every score traces to evidence, deterministic, not a guess.

The outcome

You get the decision. We own the risk.

The engine room stays ours. What lands on your desk is a defensible call.

  1. Engines

    Skills + Market

  2. Graph

    Compounds

  3. Decision

    Evidence-backed

Background patterns over a sky image (tone='ink' = light-on-dark; the grid stays, the flat navy is gone)
DiagonalLines
IntersectingLines
GridEllipsis
NoiseTexture

01e

New blocks, Tier 2

Richer interactive + narrative blocks. CaseStudyCard structures a cleared story (challenge → what we did → outcome + quote). BeforeAfterSlider is a draggable, keyboard-accessible (arrow keys, Home/End) content comparison. LogoMarquee auto-scrolls cleared names (pauses on hover, static under reduced-motion). FaqAccordion animates height via the grid-rows 0fr→1fr trick. ProcessTimeline is a vertical numbered path with a connecting rail (distinct from NumberedList). EngagementSelector repurposes a pricing-tier layout for the three deployment shapes (one featured).

CaseStudyCard, structured story (ECHT is the one cleared proof point)
ECHT

Platform partnership

ECHT

The challenge
People-decisions made on gut feel, defensible only after the fact.
What we did
Deployed a playbook on our core, wired into the existing stack, in weeks.
The outcome
Evidence-backed decisions in production, the model proven, not piloted.
“We own the AI risk. You get the outcome.”
The Everday model
BeforeAfterSlider, drag the handle (or arrow keys) to compare
After

With Everday

A ranked, evidence-backed shortlist.

Minutes · traceable · explainable.

Before

The old way

A stack of CVs and a gut call.

Slow · hard to defend · no trail.

Reveal After versus Before
LogoMarquee, on paper (cleared names only)
  • ECHT
  • Your stack
  • Your ATS
  • Your HRIS

LogoMarquee, onDark

  • ECHT
  • Your stack
  • Your ATS
  • Your HRIS
FaqAccordion, grid-rows height trick (single-open by default)

  • A pilot is 4–8 weeks, scoped, paid, with a real outcome at the end, not a slide-deck.

  • No. Agents run inside your existing ATS / HRIS. You keep your stack; we power the decisions.

  • Yes, every score traces back to evidence. Deterministic and explainable, never a black box.
ProcessTimeline, vertical numbered path with a connecting rail + meta chips
  1. 01

    Name a problem

    Pick one real people-decision with a measurable outcome.

    Week 0
  2. 02

    Run a scoped pilot

    We co-implement a playbook on our core, wired into your stack.

    4–8 weeksPaid
  3. 03

    Go to production

    The decision ships, evidence-backed and in your tools, not a deck.

    Live
EngagementSelector, three deployment shapes (one featured) · NOT prices

You run it

We hand you the engine.

We deploy a playbook on our core and you operate it day-to-day, with our support on tap.

How it works
Playbook + access + onboarding.
Typical fit
Teams with the capacity to run it in-house.
Talk to us
Most chosen

We deliver

You get the outcome.

We run the engine end-to-end and hand you the decision. We own the AI risk.

How it works
Managed service on our core.
Typical fit
Teams who want the result guaranteed.
Book a pilot

Co-built

We build it together.

We pair with your team on the first pilot, then hand over a running playbook.

How it works
Joint build + handover.
Typical fit
Teams building internal capability.
Talk to us

02

AnimatedGridPattern

The product-side visual language (used on /platform). Static grid + a pool of squares pulsing in/out at random delays. No fade mask, present, slightly hard, with the dynamism in the animation.

text-(--color-ink)/2244×44 cells, 28 squares

03

Two paths, sky cards

The homepage TwoPaths split. Each direction is its own full-bleed painterly-sky card (sunset-golden for companies, sunset-pink for service providers) with white copy over a soft dark gradient. The old audience colour-coding (gold vs purple, data-tone) is gone — both cards carry the single accent system; the sky image does the differentiation.

For companies

Bring us your hardest people decision. We build the agents into your stack and leave them running with you.

For service providers

Bring us your hardest people decision. We build the agents into your stack and leave them running with you.

01

Type pairing, LOCKED

Geist (UI / body) + Space Grotesk (the modern grotesque display face — reads 'engineered / AI'). Body stays Geist; the display var is set by next/font as --font-display.

Geist + Space Grotesk

Close the execution gap on people decisions.

Ranked, evidence-backed shortlists inside your ATS. Deterministic, traceable, explainable, not a black box.

02

Display treatment, LOCKED

Treatment 3: a single structural word in the accent color. One word, never two. Per-locale, the natural stress phrase varies and lives in the YAML.

Treatment 3, single-word accent (LOCKED)

Close the execution gap on people decisions.

03

Hero layout, LOCKED

Left-aligned, container-content width. Eyebrow → display-1 → sub → primary CTA + ghost secondary. Background pending, picked above.

Layout 1, left-aligned (LOCKED)

The execution gap

Close the execution gap on people decisions.

Ranked, evidence-backed shortlists inside your ATS. Deterministic, traceable, explainable, not a black box.

04

Color tokens, indigo → teal (2026-06)

Disciplined system: ONE primary hue (indigo, the accent everywhere), ONE supporting accent (teal — doubles as positive/after and the on-dark accent), one warm accent (amber — INK TEXT ONLY, sun glyph + occasional fills), functional green/red. The old purple/sky/peach/gold rainbow AND the per-audience colour flip (data-tone business=gold / partners=purple) are both retired — one accent, no audience coding. Deprecated token names remain only as aliases → these values. WCAG AA verified below.

Surface

--color-paper

#FFFFFF

--color-paper-elev

#F5F6F8

--color-ink

#1D2332

on paper: 14.1:1 · AAA

--color-ink-mute

#6B738A

on paper: 4.6:1 · AA

--color-ink-faint

#9CA3B4

--color-rule

#E8EAF0

--color-rule-soft

#F0F2F5

Primary · indigo (CTAs, links, accents, numerals)

--color-indigo

#4338CA

on paper: 8:1 · AAA

--color-indigo-hover

#3730A3

--color-indigo-soft

#E8E7FB

Supporting · teal (secondary accent + positive/“after”)

--color-teal

#0EA5A5

on paper: 3.2:1 · AA UI / large only

--color-teal-hover

#0C8A8A

--color-teal-soft

#DFF6F5

Audience · amber (companies), INK TEXT ON AMBER ONLY

--color-amber

#F59E0B

on ink: 8.6:1 · AAA (ink-on-amber)

--color-amber-hover

#DB8B08

--color-amber-soft

#FFF4DD

Functional only (states, before/after, not decorative)

--color-positive

#0EA5A5

--color-negative

#E5484D

WCAG AA verified. Indigo #4338CA on paper 8.0:1 (AAA, text + UI). White on indigo 8.0:1 (AAA). Ink #1D2332 on amber #F59E0B 8.6:1 (AAA), amber is fills/CTAs only, never text (amber-on-paper ≈ 2.0:1, fails). Teal #0EA5A5 on paper 3.2:1, AA for UI/large/icons + the “after” bar, paired with ink text for prose. Ink-mute #6B738A on paper 4.6:1 (AA). The accent-text token resolves to indigo on light surfaces and teal on dark, so coloured text always clears AA.

05

Type scale

Display sizes (text-hero / text-mega / display-1 / display-2) in Space Grotesk, 600–700 weight, tightened tracking. Body sizes (h2 / h3 / body / small / eyebrow) in Geist. Each token's family + weight is set on the .text-* class in globals.css (Tailwind v4 doesn't apply font-family from the --text-* sub-props).

text-hero · Space Grotesk 700 · clamp(48→104px) · homepage hero only

See it on one of your real decisions.

text-mega · Space Grotesk 600 · clamp(40→64px) · hero / tile headline

Mega headline

text-display-1 · Space Grotesk 600 · clamp(34→52px)

Display 1

text-display-2 · Space Grotesk 600 · clamp(28→40px)

Display 2

text-h2 · Geist · clamp(24→34px)

Heading 2

text-h3 · Geist · clamp(18→24px)

Heading 3

text-body · Geist · clamp(16→19px) · 1.65 line-height

Every applicant scored against your job spec and required skills. Ranked shortlists in minutes, with the evidence behind every one.

text-small · Geist · 14px

Meta, captions, footnotes.

text-eyebrow · Geist · 12px · +10% tracking

Eyebrow / section label

06

Spacing scale

4px base. Semantic tokens (section/block/stack) scale up on tablet+.

4px
8px
12px
16px
24px
32px
48px
64px
96px
128px

07

Primitives

Bespoke layer. Focus-visible rings, keyboard interaction, 44px+ hit targets.

Buttons
Text link

Read more about our deterministic engine and the evidence behind every decision.

Badges
NeutralAccent

08

Motion samples

Three-motion budget. Reduced-motion compresses everything to instant.

1 · Page enter

Page enter

Opacity 0→1, Y 16→0

500ms · ease-out-quart

2 · Section reveal

Stagger 40ms · Y 8→0 · ease-out-quart

3 · Interactive feedback

Hover: color shift (150ms). Press: scale 0.985. Focus-visible: instant ring.