AI-Powered Workflow Automation: Multilingual Translation and Publishing
Samuel Hallé
Co-Founder & CIO
Nesto manages over $73 million in mortgages. Its marketing team was facing a major bottleneck in content translation — with 30 articles pending, high costs, and inconsistent brand tone. This challenge was slowing down the publication of bilingual content, which is essential to the company’s digital strategy.
Translation speed for a 1,500-word English-to-French text
Based on Hologram analytics data
Vision
Inspired by the concept of the “Iron Man suit”, Nesto aimed to equip its marketing team with AI tools designed to augment their capabilities.The goal: to create an ecosystem of intelligent agents working collaboratively to automate linguistic tasks while preserving the company’s unique brand voice.
Solution
Developed using the HALO platform (designed by Hologram), the solution established a specialized, interconnected, and fully orchestrated multi-agent infrastructure.
Key Components :
- Orchestrator Agent – Triggers and coordinates tasks through n8n workflows.
- Glossary & Brand Style Agent – Enforces Nesto’s linguistic tone and brand guidelines.
- Initial Translation Agent – Performs the first-pass translation from source content.
- Linguistic Reviewer Agent – Ensures consistency and fluency of the text.
- Final Editor Agent – Generates the formatted draft directly within the CMS (WordPress or Google Docs).
Technologies Used
- OpenAI AgentKit
- MCP Servers
- CMS REST API
- n8n (automation and orchestration)
Results
- Translation time reduced from 2 days to 2 minutes
- Significant reduction in cost and manual workload
- Enhanced consistency in brand voice across all content
- Data protection and scalability ensured by internal architecture
.avif)
