BooAI Technical Bid Writing

Problem It Solves

Writing technical bid proposals is often hard not because of length, but because the AI may miss scoring items, ignore the evaluator’s logic, pad chapters without structure, or leak process notes into the final document. This skill turns long proposal drafting into scoring-table-driven heading generation and chapter-level concurrent writing, targeting technical bid documents up to around 300,000 words.

How It Works

• Input conversion: It first uses doc2txt.py to convert .doc, .docx, .xlsx, and text-layer .pdf files into .txt or .md files. Scanned PDFs trigger a hint to use OCR or a vision-capable model.
• Requirement analysis: It extracts project basics, core requirement tags, hard thresholds, evaluation rules, terminology, and project type into a traceable requirements report.
• Heading generation: L1, L2, and L3 headings are generated from the scoring table item by item. The workflow aims to cover scoring points, filter out items requiring manual entry such as company-specific numbers, credentials, or proof documents, and produce a scoring-item coverage table for review.
• Word-count allocation: Chapter quotas are estimated from the actual number of L3 headings and an empirical per-point word count, using T = H₃ × P_avg × A.
• Concurrent writing: Each chapter is written by an independent agent. Completion reports are kept in separate files, chapter bodies should contain only proposal content, and the merge stage checks word count, terminology consistency, and scoring coverage.

Scope and Caveats

This skill does not replace real qualifications, personnel data, performance records, or supporting evidence. Scanned documents, complex drawings, and manually entered items still require OCR, vision models, or human completion. The scoring table is the primary source of truth; project-type templates only help with industry phrasing and should not override the evaluator’s scoring structure.