Passer au contenu principal

Digs transforme les plans d'étage en données de construction lisibles par l'IA

Découvrez comment Digs a fait passer à l’échelle l’annotation de plans de haute qualité pour des processus de construction assistés par l’IA.

septembre 1, 2026 | États-Unis

White house icon centered inside a blue cloud-like shape on a white background

Executive Summary

As Digs expanded its AI-powered collaboration platform for home builders and homeowners, it needed a reliable way to turn complex floor plans and blueprint images into structured data its AI systems could use.

By partnering with Uber AI Solutions (UAIS), Digs established a flexible annotation operation that adapted as blueprint-labeling requirements evolved. The engagement maintained high annotation quality, expanded support across more detailed construction elements, improved operational visibility, and created a repeatable foundation for continued AI product development.

Résultats clés

99.5%+

annotation quality maintained

1,000+

tâches supplémentaires étiquetées pour l'identification du nom de la pièce

La mise en place de flux de travail de construction basés sur l'IA repose sur une annotation de données sophistiquée.

Digs développe une plateforme d'intelligence artificielle qui aide les constructeurs et les propriétaires à organiser les documents de construction, à collaborer sur les plans, à automatiser les transferts de propriété et à conserver un dossier numérique du logement. Pour que ces processus fonctionnent de manière fiable, la plateforme devait d'abord être capable de comprendre les plans et schémas d'aménagement résidentiels, qui varient considérablement en termes de disposition, de symboles, de mesures, d'orientation et de détails architecturaux.

Barre de navigation avec des boutons à bascule, un menu déroulant intitulé 'Aller à...', et des options de sélection de l’affichage 2D/3D.

As Digs introduced new AI capabilities, its data annotation needs became more sophisticated. Work that began with room segmentation expanded into room naming, measurements, wall interpretation, and other blueprint elements. The challenge was not simply labeling more images. Digs needed a data annotation operation capable of adapting as the product evolved while also preserving downstream consistency builders and homeowners depended on.

Finding an annotation partner that could adapt as the product evolved


Digs partnered with Uber AI Solutions (UAIS). The UAIS team quickly aligned with Digs’ existing technical environment and evolving workflows, rather than forcing them into a rigid operational structure. UAIS assembled a dedicated annotation team and worked within Digs’ established review process. As requirements changed, workflows were adjusted incrementally. Adjusted workflows supported rotated layouts, updated segmentation rules, room naming, measurements, and increasingly specialized blueprint elements without rebuilding the operation from scratch.

The collaboration also addressed an operational visibility gap. Because review and ingestion processes existed across separate systems, blocked work and emerging issues could be difficult to identify quickly. UAIS introduced dashboard reporting and automated Slack updates that surfaced queue status, active operations, and blocked items, giving both teams greater day-to-day visibility while reducing manual coordination.

Maintenir une précision des annotations de plans de 99.5%+ alors que les exigences en matière d’IA se sont accrues

The engagement established a stable annotation operation capable of supporting changing blueprint requirements while sustaining greater than 99.5% quality across core workflows.

Reporting visibility made it easier to identify workflow issues and coordinate production, while the flexible operating model allowed Digs to introduce new annotation tasks as its AI product evolved. For example, the new workflow was able to easily absorb an add-on project of over 1,000 tasks for room name identification.

Interface de plan d'étage numérique avec menu de navigation, outils de dessin et commentaires du projet sur la droite.

Forte de son expérience opérationnelle dans la gestion de programmes d’annotation complexes et l’adaptation à des exigences évolutives, UAIS a permis à Digs d’étendre l’utilisation de données de construction structurées sans devoir créer un nouveau processus interne chaque fois que les exigences d’annotation évoluaient.

A scalable foundation for AI-powered construction workflows


AI systems for residential construction depend on accurate interpretation of highly variable visual documents. Building those capabilities requires annotation operations that can evolve alongside the product itself.

By partnering with Uber AI Solutions, Digs established a reliable blueprint annotation workflow capable of supporting changing requirements, maintaining quality, and scaling operationally without introducing unnecessary complexity. Now, Digs is building AI-powered workflows for the modern era of home construction and ownership.

Découvrez comment Uber AI Solutions peut s'associer à votre entreprise pour faire progresser votre IA.

Les témoignages sont le reflet de résultats individuels. Les résultats individuels ne constituent pas une garantie de résultat pour un client donné, et l'expérience client peut varier d'un commerçant à l'autre.