Product and material management
Identify, compare and enrich technical and price information for products and materials specified on a project, from photos, PDF spec sheets or supplier quotes.
Without AI
- Collect photos, PDFs and quotes20 min
- Search for the exact product reference45 min
- Read and compare technical sheets35 min
- Extract prices and conditions25 min
- Update the price and product database30 min
Environ 2 h 35
With AI
- Prepare and group source documents10 min
- Generate product candidates from the image or description15 min
- Extract and structure technical and price data20 min
- Review discrepancies and verify sources10 min
- Update the shared product database10 min
Environ 1 h 05
Why this use case
La recherche de références produits mobilise souvent des informations dispersées entre photographies de chantier, fiches techniques PDF, devis fournisseurs et tableaux de prix. Une méthode structurée permet de retrouver plus rapidement les caractéristiques, les variantes et les montants associés aux matériaux prescrits.
L’IA peut accélérer l’identification et l’extraction des données, mais elle doit rester un outil de présélection et de mise en forme. La référence finale, la conformité technique, le prix applicable et la disponibilité doivent être vérifiés par l’architecte ou le fournisseur avant toute prescription.
Process steps
Collect and standardize the source material
10 à 20 minGather the project photos, PDF specification sheets, supplier quotes and existing product records. Rename files consistently and note the project, lot, location and date associated with each source.
Identify candidate products from images and descriptions
AI-assisted10 à 15 minUse the image and contextual clues to generate a shortlist of possible manufacturers, collections and references. Treat the result as a candidate list and retain the visual or textual clues supporting each possibility.
Extract technical and commercial information
AI-assisted15 à 25 minExtract structured fields from specification sheets and supplier quotes: reference, dimensions, finish, performance, unit, price, currency, validity date, minimum order and delivery conditions. Preserve the page or document source for every important value.
Compare alternatives and flag inconsistencies
AI-assisted15 à 20 minCompare candidate products against the project requirements and identify differences in dimensions, finish, performance, certifications, price basis and availability. Separate factual differences from missing information and calculate price comparisons only on a consistent unit basis.
Validate and update the product register
AI-assisted10 à 20 minManually confirm the selected reference with the manufacturer or supplier, then record the validated data in the shared product and price register. Keep alternatives, assumptions, source links, quotation dates and review status visible.
Prompts
Tools used
ChatGPT
Steps 2 and 3 and 4 and 5Candidate identification and document comparison
Key takeaway : Ask for sources, uncertainty levels and missing fields rather than a single definitive answer.
Perplexity
Step 2Web search for manufacturer and product sources
Key takeaway : Use the candidate reference, manufacturer clues and technical attributes together to narrow the search.
Airtable
Steps 1 and 5Shared product and price register
Key takeaway : Add fields for source document, quotation date, unit, validity and verification status so old prices are not mistaken for current ones.
Limitations
A photo rarely proves an exact reference
Similar finishes, lighting, camera settings and installation details can lead to several plausible candidates. The exact product must be confirmed with a source or supplier.
Extracted values require source checking
PDF layouts, scanned documents, tables and quotation abbreviations can cause missing, shifted or misread values. Check important figures against the original page.
Prices are time- and context-dependent
A price may exclude transport, installation, taxes, discounts or escalation and may no longer be valid. The database is not a substitute for a current supplier quotation.
Technical compliance remains a professional decision
AI can organize requirements and expose differences, but it does not certify performance, regulatory compliance, compatibility with the detail or suitability for the project.
Related use cases
Tender search and response
Spot relevant tenders, quickly extract the key information needed to decide whether to bid, then prepare a structured response and the required administrative documents.
Business Offer Analysis - Quotation
Extraction of structured information to create a comparison table of offers.
