What challenges did Vionlabs face before AI implementation?

Audio-visual models alone could not detect important plot details or textual nuances in dialogue. Vionlabs wanted to improve metadata generation for streaming content and speed integration of new tools without building proprietary LLMs.

What was the impact of the AI solution?

Deployment of new multimodal capabilities took weeks instead of the six-to-nine months typically required to train embedding models. Revenue scaled without significantly impacting costs.

Source

GCPEvidence: Medium50/100

Vionlabs strengthens multimodal content discovery with Gemini Enterprise Agent Platform

Use case typeContent discovery assistantUpdated Jun 29, 2026

Vionlabs added text as a third modality to its multimodal content intelligence pipeline so it could better understand plot and textual nuances in video libraries. The company used Llama 3.1 models on Gemini Enterprise Agent Platform, integrated with BigQuery, to speed tool adoption and training-job iteration without building its own text LLM. The result was faster deployment, new frame-level indexing plans, and revenue growth without significant cost increase.

Organization: Vionlabs
Industry: Tech & Comms
Location: Sweden
Published: June 2026

Planned next steps

The approach enabled new frame-level indexing plans.

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Deployment time for new multimodal capabilities: 80% decrease

Google Cloud Customer StoryJun 29, 2026Customer storyInferred claimMedium evidence strength

taking only a few weeks instead of the six to nine months typically required for training their own embedding models

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Vionlabs
Provider: GCP
Maturity: Unknown
Linked source: Google Cloud Customer Story

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 3 of 3

1Content discovery assistant
2AI agents
3Multimodal analytics

Used Llama 3.1 405B and 70B on Gemini Enterprise Agent Platform as the text modality.
Used hosted APIs on the platform and BigQuery for data integration and training-job tracking.
Focused on integrating text into existing multimodal content analysis workflows and planning frame-level indexing.

Technologies

Gemini Enterprise Agent Platform Llama 3.1 BigQuery Cloud Run Kubernetes Engine Dataflow TensorFlow

Deployment of new multimodal capabilities took weeks instead of the six-to-nine months typically required to train embedding models.
Revenue scaled without significantly impacting costs.

Architecture

Vionlabs uses Llama 3.1 models on Gemini Enterprise Agent Platform together with BigQuery, Cloud Run, Kubernetes Engine, Dataflow, and TensorFlow to add text as a third modality to its multimodal content analysis workflow and speed the launch of new capabilities.

Sources & evidence1

Evidence: Medium50/100Evidence strength