AI-Powered Automation for Project Management.

The primary objective was unicrew’s own operational efficiency: automate a critical, time-consuming, internal administrative task. unicrew engineered an AI agent that interprets a simple text input, autonomously navigates unicrew’s internal project management system, identifies personnel with delinquent time logs for the preceding week, and issues automated notifications to both the individuals and their respective managers via Google Chat. It served as a strategic proof of concept, to validate AI-driven automation of internal workflows on unicrew’s own back office.

unicrew’s own operations were losing significant time to routine administrative processes, pulling people off core strategic work. The primary goal was to mitigate that inefficiency through intelligent automation.

Key technical obstacles included the inherent difficulties in automating interactions with secure, multi-factor authentication systems like Google sign-in, the unpredictable behavior of some AI models (which required meticulous prompt engineering), and the robust security protocols of platforms like Google Chat, which presented considerable challenges for programmatic user searches and messaging.

The stability of the entire automation process was also contingent on the reliability of front-end UI selectors, which can be prone to change.

• AI-powered process orchestration: leveraged the LangChain framework to orchestrate a sequence of complex, multi-step tasks automatically, from data retrieval to final notification.

• Intelligent web navigation: used advanced browser automation libraries so the AI agent could autonomously navigate and interact with unicrew’s web-based project management tool.

• Scalable AI model deployment: hosted and managed advanced AI models on AWS Bedrock to ensure robust performance and provide a scalable foundation for future AI initiatives.

• Automated workflow execution: designed an intelligent workflow that seamlessly processes inputs and triggers a cascade of automated actions, creating a fully autonomous operational sequence.

To address these challenges, unicrew developed a proof of concept (PoC) centered on a bespoke AI web agent. The agent was engineered to execute a precise workflow: navigate to the target web resource, extract and aggregate the necessary data, analyze it to compile a report of non-compliant users, and trigger the notification protocol through integrated communication channels.

The technical foundation for this solution was built using Python and its associated libraries, with the core AI model hosted on AWS Bedrock for scalable and reliable performance.