Sample Work | Self-Initiated Project

Overview

This sample demonstrates how I use AI as a controlled work tool for information processing, operational reporting, and drafting while keeping factual verification and final decisions under human review. I created an AI-assisted workflow that transforms 30 fictional weekly business records from five different source types into a structured Weekly Operations Brief. The project shows the full process from raw information and initial AI output through error identification, prompt refinement, source-level verification, human judgment, and final business documentation. The emphasis was not on accepting AI-generated content automatically, but on creating a traceable workflow in which generated information could be checked, corrected, and improved before use.

Scenario

BrightDesk Creative is a fictional small remote creative-services business managing several client projects together with internal administrative work.

During one reporting week, operational information is received through meeting notes, internal messages, client emails, task-tracker updates, and administrative or finance notes.

These updates contain overlapping information, changing statuses, tentative dates, missing owners, client questions, resolved issues, and later information that supersedes earlier records.

The company needs a clear Weekly Operations Brief without introducing unsupported assumptions or losing important uncertainty.

BrightDesk Creative, its clients, projects, employees, messages, and all business information used in this sample are entirely fictional and were created solely for this self-initiated portfolio project.

Objective

The objective was to design and test an AI-assisted workflow that could:

• consolidate information from multiple operational sources;
• identify current task statuses and important developments;
• distinguish completed, in-progress, in-review, and blocked work;
• identify relevant dates, actions, questions, and administrative items;
• preserve tentative or missing information rather than converting it into unsupported facts;
• recognize when later evidence clearly supersedes older information;
• produce concise professional business wording;
• and require human verification before the final report was accepted.

The Raw Input Pack remained the source of truth throughout the project.