What challenges did Bancolombia face before AI implementation?

Traditional marketing campaign ideation and visual asset creation required many meetings and revision rounds, taking weeks or months and slowing time to market.

What was the impact of the AI solution?

The article says the process produces visuals in hours instead of days or weeks and helps reduce early-stage time and resources. It also helps teams react to trends and publish faster with more consistent messaging.

Source

ExploringEvidence: Medium50/100

Bancolombia streamlines marketing campaign visual asset generation with Amazon Bedrock (Amazon Nova)

Use case typeMarketing analyticsUpdated Dec 23, 2025

Bancolombia is experimenting with Amazon Nova and Amazon Bedrock to generate visual assets for marketing campaigns. The workflow uses automated retrieval of prior campaign images and metaprompting to generate prompts for Amazon Nova Canvas.

AI-generated summary. Generated from indexed evidence or computed site data. Underlying records pass applicable deterministic checks; verify important details with source links where available. How content is created.

Organization: Bancolombia
Industry: Finance
Location: Colombia
Published: December 2025

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

Normalized claim

Visual asset production time: 75% decrease

AWS Machine Learning BlogDec 23, 2025Blog postInferred claimMedium evidence strength

producing visuals in hours vs days/weeks

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

Customer: Bancolombia
Provider: AWS
Maturity: Exploring
Linked source: AWS Machine Learning Blog

Bancolombia is experimenting with Amazon Nova and Amazon Bedrock to generate visual assets for marketing campaigns

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Marketing analytics
2Content generation
3Workflow automation

The solution uses Amazon Bedrock with Amazon Nova Pro and Amazon Nova Canvas to generate reference prompts and campaign images.
A multimodal/vector-based image retrieval step surfaces related visuals from previous campaigns.
The workflow is implemented with AWS Lambda, Amazon API Gateway, IAM, AWS KMS and Amazon Cognito.

Technologies

Amazon Bedrock Amazon Nova Amazon Nova Pro Amazon Nova Canvas AWS Lambda Amazon API Gateway AWS KMS Amazon Cognito AWS Identity and Access Management

The article says the process produces visuals in hours instead of days or weeks and helps reduce early-stage time and resources.
It also helps teams react to trends and publish faster with more consistent messaging.

Architecture

Reference architecture uses Amazon Bedrock to invoke Amazon Nova Pro and Amazon Nova Canvas via APIs; AWS Lambda functions handle metaprompt generation, image retrieval and storage; Amazon API Gateway exposes APIs; Amazon Cognito manages users; AWS KMS encrypts data; and IAM is used for least-privilege access.

Sources & evidence1

Evidence: Medium50/100Evidence strength