tesseraLAB

An independent human judgment lab making AI better at craft.

Start a partnership Research
PLINTH
Internal tools builder
RIPPLEKIT
Interactive animation
FACETTE
3D design
POSTMARQUE
Email design
CADENZA
AI development
LATTICEWORK
Website builder
ARGOT
AI agents
ANYFORM
AI 3D generation
BOUGIE
No-code apps
BUILDHOUSE
Backend automation
FLORIN
Creative collaboration
FLUTTERBOARD
Mobile app builder
FRESCO
Website builder
FRESCO COMMERCE
Ecommerce platform
GANTRY
Developer infrastructure
GENERAL WORKSHOP
Craft education
GLIDEPATH
No-code apps
GLORIETTE
AI design tool
HELIOGEN
AI video generation
IDEOGRAPH
AI image generation
INSTANTLY
AI eCommerce builder
JITTERBUG
Motion design
KAJIBA
Course platform
KEEL (POSTKIT)
Email marketing
KITTIWAKE
AI graphic design
LOTTIEHOUSE
Animation platform
LOVABEL
AI app builder
LOVART STUDIO
AI art generation
LUMINA
AI image platform
MAGICPATH
AI workflows
PAGEDECK
Presentation tools
PALETTA
Colour systems

Creative judgment keeps advancing. We make sure it stays ahead.

Start a partnership

Meet the Tessera Lab ecosystem

We build the evaluation layer for creative AI, helping model teams and tool builders understand quality the way working practitioners do.

Open evaluations

Judgment Arena

Research & findings

Taste Benchmark

Closed evaluations

Human Signal Sets

Craft AI agents

Co—Reviewers

Bringing craft into generative models

Tessera Lab unites closed evaluations, open comparisons, and published research inside one human judgment framework.

Paired blind reviews

Models are judged side by side through structured, bias-controlled comparison protocols.

Expert preference models

We capture and analyse high-signal preference data from vetted working practitioners.

Craft scoring systems

Qualitative judgment is translated into structured rubrics aligned to studio standards.

Multimodal model evals

We assess image, video, layout, and interactive output where metrics miss the craft.

Built on working studio practice

With real commercial pressure

The Atlas of Human Judgment

312Kvetted independent creatives
437distinct craft disciplines
$186Mpaid out to reviewers
23×higher review throughput per panel
64models evaluated

Work with the field's most demanding critics

Start a partnership

The people who set the standard.

Start a partnership

Advancing next-gen craft research.

Human judgment is the new training signal.

Generative craft isn't good enough yet. Human taste keeps moving. We exist to make sure it stays in front.

Output is cheap. Now judgment is the whole game.

Move past leaderboard scores and surface the qualitative signal that shapes what people actually choose.

Start a partnership