Doc2Memo

Doc2Memo helps teams build stronger technical memories for public tenders in Spain. Starting from the tender documents (PCA and PPT), it analyzes requirements with AI and produces a structured, editable, and exportable technical proposal.

From raw tender docs to a submission-ready technical memory in minutes, not days.

Sample documents: docs/samples/*.

For teams

Doc2Memo is designed to reduce time spent on repetitive drafting while keeping proposal quality high.

• Upload your PCA and PPT files (pdf, md, txt).
• Let asynchronous AI analysis extract criteria, specifications, and strategic insights.
• Generate a technical memory in dynamic sections aligned with judgment-based evaluation criteria.
• Regenerate individual sections when you want a better draft.
• Export the final output as PDF or Markdown.

Quick 5-minute flow

1. Create a tender and upload PCA + PPT.
2. Wait for background analysis to complete.
3. Generate the technical memory.
4. Review sections, regenerate weak parts, and export.

Why it is useful

• Prioritizes scoring opportunities from judgment criteria.
• Speeds up first drafts without losing structure.
• Keeps the process traceable with operational metrics.

Key capabilities

• Tender creation and document ingestion for PCA/PPT.
• Queue-based analysis pipeline with AI agents.
• Criteria classification (judgment and automatic) with score normalization.
• Dynamic section generation (one queue job per section).
• Per-section quality gate, retries, and optional style editor pass.
• Internal operational metrics dashboard at technical-memories/operational-metrics.

Output quality controls

• Dynamic sections are built from judgment-based criteria groups.
• Each section runs through a quality gate before completion.
• Failed quality checks can be retried automatically.
• Optional style editing improves readability while preserving facts.

Architecture at a glance

1. A tender is created and PCA + PPT are uploaded.
2. ProcessDocument analyzes each document asynchronously.
3. Criteria, specifications, and insights are persisted.
4. GenerateTechnicalMemory groups judgment criteria into dynamic sections.
5. GenerateTechnicalMemorySection generates each section and updates status/metrics.
6. Users review the memory, regenerate sections if needed, and export to PDF/Markdown.

For developers

Tech stack

• PHP 8.4
• Laravel 12
• Livewire 4
• Pest 4
• Laravel AI SDK (laravel/ai)
• Tailwind CSS 4
• SQLite by default

Local setup

Quick option

1. Copy environment variables and create local SQLite file:

cp .env.example .env
php -r "file_exists('database/database.sqlite') || touch('database/database.sqlite');"

2. Install and bootstrap the project:

composer setup

3. Configure at least one AI provider key in .env (OpenAI is the default):

OPENAI_API_KEY=your_key

4. Run the full development environment (server, 3 queue workers, and Vite):

composer dev

Then open the app and start from the tenders list.

Manual option

composer install
npm install
php artisan key:generate
php artisan migrate
php artisan serve
php artisan queue:work --timeout=300 --tries=1
npm run dev

Domain map

• app/Livewire/Tenders/CreateTender.php: tender creation and document upload.
• app/Actions/Documents/ProcessDocumentAction.php: extraction and persistence for PCA/PPT data.
• app/Ai/Agents/DocumentAnalyzer.php: structured document analysis.
• app/Ai/Agents/PcaJudgmentCriteriaExtractorAgent.php: dedicated extraction of judgment criteria.
• app/Actions/Tenders/GenerateTechnicalMemoryAction.php: dynamic section orchestration.
• app/Jobs/GenerateTechnicalMemorySection.php: per-section generation, quality gate, and metrics.
• app/Ai/Agents/TechnicalMemoryDynamicSectionAgent.php: section drafting.
• app/Ai/Agents/TechnicalMemorySectionEditorAgent.php: post-generation style editing.
• app/Livewire/TechnicalMemories/ShowMemory.php: memory page and section regeneration.
• app/Livewire/TechnicalMemories/OperationalMetrics.php: internal metrics dashboard.

Useful routes

• /: redirects to tenders list.
• /tenders/create: create a new tender.
• /technical-memories/operational-metrics: internal monitoring dashboard.

Testing and quality

php artisan test --compact
vendor/bin/pint --dirty --format agent

To run one specific test:

php artisan test --compact tests/Feature/Jobs/GenerateTechnicalMemorySectionTest.php

Operations and maintenance

• deadline_date is stored as text to support natural language deadlines.
• The system depends on active queue workers for analysis and generation.
• Old metrics are purged daily with technical-memory:purge-metrics.
• If code changes are not reflected in running jobs:

php artisan queue:restart
php artisan optimize:clear

Troubleshooting

• Stuck jobs: verify active workers (composer dev or php artisan queue:work ...).
• Memory generation not progressing: check section states in the memory view and application logs.
• Frontend changes not visible: run npm run dev or rebuild with npm run build.
• Vite asset runtime errors: restart npm run dev.