Startup Monitoring Platform - Audit and Improvement Work
A consulting and implementation engagement focused on reviewing, modernizing, and extending an existing startup monitoring platform and its AI-generated reporting workflow.
About the Project
We were brought in to improve a startup monitoring platform that had originally been built by another team using Lovable. Our role was not to create the product from scratch, but to review the existing implementation, understand how it worked in production, and evolve it safely.
That included analyzing the code used to generate reports, investigating the Firebase environment and trigger flow, reviewing the report that the existing code was producing, and studying the project charter materials that informed the new version.
Based on that analysis, we implemented a new reporting layer that combined the strongest parts of the old report with the newer project-charter structure, while modernizing the document design and making the layout more resilient to variable-length AI-generated content.
We also shipped targeted product improvements on top of the platform itself, including restoring a missing chart from the previous report version, adding new Tally inputs for manager considerations and an optional transcription file, and creating an editing page where users could revise textual PDF fields and regenerate a new report version from those edits.
Key Features
- Technical review of an existing Lovable-built platform and report-generation code
- Firebase production analysis to understand generation triggers and deployment behavior
- Assessment of legacy report output to preserve useful sections and improve weak points
- Implementation of a redesigned report combining Tally inputs with new AI-generated sections
- Restoration of the missing chart so the new report matched the expected visual completeness
- New Tally fields for mandatory manager notes and optional DOCX transcription upload
- Expanded AI context using manager observations and transcription content
- New PDF text-editing page with regeneration flow for edited report versions
- Structured outputs for more predictable AI responses and easier formatting of tables and paragraphs
- Refinement of AI calls for diagnosis summary, expected results, scope, macro timeline, assumptions, and constraints
- Visual redesign of the generated document for more modern presentation and variable text sizing