Manage LLM prompt instructions
INSTRUCTIONS FOR CALCULATING SCANNED PRODUCT MATCH SCORE Calculate a personalized match score (0-100%) between the user and scanned product using three components: **COMPONENT 1: PERSONALIZATION ANSWERS MATCH (40 points maximum)** - Compare user's answers to personalization questions with product's answers - Weight: 40 points total - Calculation: Divide 40 by number of questions to get points per question - Award full points per question when user reaction matches product answer - Award zero points when answers don't match **COMPONENT 2: USER REACTION TO PRODUCT (20 points maximum)** - Check if user has previously rated this specific product - Scoring: * User rated 'like': 20 points * User rated 'dislike': 0 points * No rating (neutral/new user): 10 points (50% weight) **COMPONENT 3: SENSATION PREFERENCES MATCH (40 points maximum)** - Compare user's top sensation preferences with product's sensations - Weight: 40 points total - For each user sensation in their top 10: * Find matching sensation in product (same name and type) * Calculate similarity: 1 - (|product_value - user_value| / 100) * Accumulate similarity scores - Final calculation: (total_similarity / sensations_checked) × 40 - If no sensations to compare: award 20 points (50% weight) **FINAL SCORE CALCULATION:** 1. Sum all three components: total_points 2. Calculate percentage: (total_points / max_points) × 100 3. Round to whole number 4. Ensure result is between 0-100 **FALLBACK BEHAVIOR:** - If no personalization or user profile exists: use product's verification_percentage - Always prefer calculated score over default verification_percentage when data is available
You are a taste profile analyst. Analyze this user's preferences and write DIRECTLY TO THEM using second-person language. Liked products: {liked_products} Disliked products: {disliked_products} IMPORTANT RULES: 1. NEVER use third-person ("the user", "they", "their", "the individual") 2. ALWAYS use second-person ("you", "your", "you seem", "you prefer", "you enjoy") 3. Start summary with phrases like: "Your taste preferences...", "You seem to enjoy...", "You prefer...", "Your palate gravitates toward..." 4. Write as if speaking directly to the user 5. Be conversational and personal Generate: 1. Summary (2-3 sentences): Describe their taste preferences using "you/your" 2. Recommendations (1-2 sentences): Give advice using "you" language EXAMPLE CORRECT FORMAT: "Your taste preferences lean toward bold, complex flavors. You seem to enjoy products with rich, savory notes..." EXAMPLE WRONG FORMAT (NEVER DO THIS): "The user's taste preferences...", "They enjoy...", "The individual prefers..."
ANALYZE THIS PRODUCT Based on: Image: {image_url} Product Name: {product_name} Brand: {product_brand} Type: {product_type} Sub Type: {product_type_sub} 1. SENSATIONS (exactly 5): Identify the 5 most prominent sensations: SENSATION TYPE RULES: - Taste: Sweet, Salty, Sour, Bitter, Umami, Kokumi, Oleo - Odor: Any smell - Texture: Physical feel (Smooth, Crunchy, Creamy, etc.) - Aroma: Complex scent combinations - Flavor: Complex taste + odor combinations - Taste [F]: Trigeminal sensations (Spicy, Cooling, Metallic, Astringent) CRITICAL: For Taste type, use ONLY the 7 basic tastes listed above. Do NOT create new taste sensations like 'Tangy', 'Tart', 'Sweet Peel', etc. SCORING SCALE (0-120 - Absolute Intensity): 0-5 Barely perceptible; trace; background level; detected only with focus 6-10 Weak but noticeable; light presence; clearly felt; gentle impact 11-15 Low intensity; light but consistent; noticeable without effort 16-20 Low-mild; clearly noticeable; light but meaningful 21-25 Mild-moderate; soft but clearly present; balanced 26-30 Moderate-low; noticeable and stable; light impact but not weak 31-35 Moderate-low; clear and well-defined; includes Coca-Cola Sweetness 34; Fanta Zero Sweetness 34 36-40 Moderate; balanced; clearly felt; not dominant 41-45 Moderate-strong (lower); prominent but not intense 46-50 Moderate-strong (upper); strong presence; approaching intensity 51-55 Strong; prominent; impactful; defines the profile 56-60 Strong-intense; very noticeable; approaching dominance 61-65 Very strong; high impact; hard to ignore 66-70 Very strong-intense; dominant in the profile 71-75 Intense; very dominant; high sensory load 76-80 Intense-very intense; saturating; forceful 81-85 Very intense; overpowering; defines the entire experience 86-90 Extremely intense; near overwhelming 91-95 Saturating; floods perception; very difficult to balance 96-100 Extreme; overwhelming; maximum natural intensity 101-110 Hyper-intense; beyond natural levels; extract-like intensity 111-120 Absolute maximum; peak saturation; full sensory overload CRITICAL INSTRUCTION: Score each sensation based on its absolute intensity within the product, NOT its dominance relative to others. Dominance naturally emerges from independent scores—do NOT adjust scores to show hierarchy. COMPOUND-BASED SCORING: Sensation values must be grounded in the compounds present in each product. The chemical composition directly determines sensory intensity. - Red Wine (general category): Sugar content drives sweetness ~11; Tannins drive bitterness ~15 - White Wine (general category): Sugar content drives sweetness ~6; Tannins drive bitterness ~12 Always reference compound data (sugars, acids, tannins, aromatic compounds, etc.) when available to calibrate sensation scores accurately. REFERENCE BENCHMARKS (Calibration Standards): - Coca-Cola Classic [Taste] [Sweet]: 34 - Fanta Zero Sugar [Taste] [Sweet]: 34 - Natural Squeezed Orange Juice [Taste] [Sour]: 20 - Natural Squeezed Orange Juice [Odor] [Woody]: 15 - Red Wine (general) [Taste] [Sweet]: ~11 (based on sugar compounds) - Red Wine (general) [Taste] [Bitter]: ~15 (based on tannin compounds) - White Wine (general) [Taste] [Sweet]: ~6 (based on sugar compounds) - White Wine (general) [Taste] [Bitter]: ~12 (based on tannin compounds) Use these benchmarks to calibrate your sensation value estimates for similar characteristics in other products. For each sensation provide: * name: One of the sensation types listed above * type: "Taste", "Taste [F]", "Odor", "Texture", "Aroma", or "Flavor" * value: 0-120 (absolute intensity) * color: Hex code like #FF5733 2. PRODUCT DESCRIPTION: Write 130 chars max describing taste/smell/texture 3. BRAND COLORS (exactly 9): Extract 9 colors from packaging: - name: Color name - bg_color: Hex code with # - text_color: "#ffffff" or "#000000"
Analyze this product image and extract ONLY: 1. Product Name (be flexible with spelling) 2. Brand Name (look for logos, brand marks) Return just these two fields.
ANALYZE THIS PRODUCT IMAGE - YOU MUST FILL ALL FIELDS! **RULE #1: NEVER leave product_name or product_brand empty. ALWAYS provide something.** Look at the image and identify: 1. PRODUCT NAME (REQUIRED): - Read the main product text on package - If can't read: describe what you see ("Orange Juice", "Chocolate Bar", "Energy Drink") - NEVER return empty - make an educated guess 2. PRODUCT BRAND (REQUIRED): - Find the brand/company name or logo - If unclear: describe the brand style ("Premium Brand", "Local Brand", "Craft Brand") - NEVER return empty - always guess something 3. PRODUCT CATEGORIZATION HIERARCHY (CRITICAL - 3 LEVELS REQUIRED): **Step 1 - TYPE (Primary Category):** - Food, Beverage, Cosmetics, Household, Pet Care, etc. - Examples: "Food", "Beverage", "Snack" **Step 2 - SUB-TYPE (Secondary Category):** - If TYPE = "Food": Dairy, Bakery, Snacks, Confectionery, etc. - If TYPE = "Beverage": Alcoholic, Non-alcoholic - Examples: "Dairy", "Non-alcoholic", "Chocolate" **Step 3 - SEGMENT TYPE (Specific Product Category - REQUIRED):** - If TYPE = "Food" AND SUB-TYPE = "Dairy": Yogurt, Milk, Cheese, Ice Cream, etc. - If TYPE = "Beverage" AND SUB-TYPE = "Non-alcoholic": Carbonated Soft Drink, Juice, Energy Drink, etc. - If TYPE = "Beverage" AND SUB-TYPE = "Alcoholic": Wine, Beer, Spirits, etc. - Examples: "Yogurt", "Carbonated Soft Drink", "Wine", "Potato Chips" **CATEGORIZATION EXAMPLES:** - Yogurt: TYPE="Food", SUB-TYPE="Dairy", SEGMENT_TYPE="Yogurt" - Coca-Cola: TYPE="Beverage", SUB-TYPE="Non-alcoholic", SEGMENT_TYPE="Carbonated Soft Drink" - Wine: TYPE="Beverage", SUB-TYPE="Alcoholic", SEGMENT_TYPE="Wine" - Chocolate Bar: TYPE="Food", SUB-TYPE="Confectionery", SEGMENT_TYPE="Chocolate Bar" **CRITICAL: ALWAYS provide all 3 levels - TYPE, SUB-TYPE, and SEGMENT_TYPE. NEVER leave segment_type empty.** 4. SENSATIONS (exactly 5): SENSATION TYPE RULES: - Taste: Sweet, Salty, Sour, Bitter, Umami, Kokumi, Oleo - Odor: Any smell - Texture: Physical feel (Smooth, Crunchy, Creamy, etc.) - Aroma: Complex scent combinations - Flavor: Complex taste + odor combinations - Taste [F]: Trigeminal sensations (Spicy, Cooling, Metallic, Astringent) CRITICAL: For Taste type, use ONLY the 7 basic tastes listed above. Do NOT create new taste sensations like 'Tangy', 'Tart', 'Sweet Peel', etc. SCORING SCALE (0-120 - Absolute Intensity): 0-5 Barely perceptible; trace; background level; detected only with focus 6-10 Weak but noticeable; light presence; clearly felt; gentle impact 11-15 Low intensity; light but consistent; noticeable without effort 16-20 Low-mild; clearly noticeable; light but meaningful 21-25 Mild-moderate; soft but clearly present; balanced 26-30 Moderate-low; noticeable and stable; light impact but not weak 31-35 Moderate-low; clear and well-defined; includes Coca-Cola Sweetness 34; Fanta Zero Sweetness 34 36-40 Moderate; balanced; clearly felt; not dominant 41-45 Moderate-strong (lower); prominent but not intense 46-50 Moderate-strong (upper); strong presence; approaching intensity 51-55 Strong; prominent; impactful; defines the profile 56-60 Strong-intense; very noticeable; approaching dominance 61-65 Very strong; high impact; hard to ignore 66-70 Very strong-intense; dominant in the profile 71-75 Intense; very dominant; high sensory load 76-80 Intense-very intense; saturating; forceful 81-85 Very intense; overpowering; defines the entire experience 86-90 Extremely intense; near overwhelming 91-95 Saturating; floods perception; very difficult to balance 96-100 Extreme; overwhelming; maximum natural intensity 101-110 Hyper-intense; beyond natural levels; extract-like intensity 111-120 Absolute maximum; peak saturation; full sensory overload CRITICAL INSTRUCTION: Score each sensation based on its absolute intensity within the product, NOT its dominance relative to others. Dominance naturally emerges from independent scores—do NOT adjust scores to show hierarchy. COMPOUND-BASED SCORING: Sensation values must be grounded in the compounds present in each product. The chemical composition directly determines sensory intensity. - Red Wine (general category): Sugar content drives sweetness ~11; Tannins drive bitterness ~15 - White Wine (general category): Sugar content drives sweetness ~6; Tannins drive bitterness ~12 Always reference compound data (sugars, acids, tannins, aromatic compounds, etc.) when available to calibrate sensation scores accurately. REFERENCE BENCHMARKS (Calibration Standards): - Coca-Cola Classic [Taste] [Sweet]: 34 - Fanta Zero Sugar [Taste] [Sweet]: 34 - Natural Squeezed Orange Juice [Taste] [Sour]: 20 - Natural Squeezed Orange Juice [Odor] [Woody]: 15 - Red Wine (general) [Taste] [Sweet]: ~11 (based on sugar compounds) - Red Wine (general) [Taste] [Bitter]: ~15 (based on tannin compounds) - White Wine (general) [Taste] [Sweet]: ~6 (based on sugar compounds) - White Wine (general) [Taste] [Bitter]: ~12 (based on tannin compounds) Use these benchmarks to calibrate your sensation value estimates for similar characteristics in other products. For each sensation provide: * name: One of the sensation types listed above * type: "Taste", "Taste [F]", "Odor", "Texture", "Aroma", or "Flavor" * value: 0-120 (absolute intensity) * color: Hex code like #FF5733 5. PRODUCT DESCRIPTION: Write 130 chars max describing taste/smell/texture 6. BRAND COLORS (exactly 9): Extract 9 colors from packaging: - name: Color name - bg_color: Hex code with # - text_color: "#ffffff" or "#000000" CRITICAL: Product name and brand CANNOT be empty. If image is unclear, make logical guesses based on what you see!
Based on user's taste profile, recommend 3 products from this list that best match their preferences. PRIORITIZATION RULES: - Products matching BOTH sub_type AND segment_type should be ranked highest - Products matching only sub_type should be ranked second - Products matching only segment_type should be ranked third USER'S LIKED PRODUCTS: {liked_products} USER'S DISLIKED PRODUCTS: {disliked_products} AVAILABLE PRODUCTS (filtered by sub_type or segment_type match): {available_products} Return top 3 product IDs that best match the user's taste profile, ordered by match score.
SENSATION TYPE RULES: - Taste: Sweet, Salty, Sour, Bitter, Umami, Kokumi, Oleo - Odor: Any smell - Texture: Physical feel (Smooth, Crunchy, Creamy, etc.) - Aroma: Complex scent combinations - Flavor: Complex taste + odor combinations - Taste [F]: Trigeminal sensations (Spicy, Cooling, Metallic, Astringent) CRITICAL: For Taste type, use ONLY the 7 basic tastes listed above. Do NOT create new taste sensations like 'Tangy', 'Tart', 'Sweet Peel', etc. SCORING SCALE (0-120 - Absolute Intensity): 0-5 Barely perceptible; trace; background level; detected only with focus 6-10 Weak but noticeable; light presence; clearly felt; gentle impact 11-15 Low intensity; light but consistent; noticeable without effort 16-20 Low-mild; clearly noticeable; light but meaningful 21-25 Mild-moderate; soft but clearly present; balanced 26-30 Moderate-low; noticeable and stable; light impact but not weak 31-35 Moderate-low; clear and well-defined; includes Coca-Cola Sweetness 34; Fanta Zero Sweetness 34 36-40 Moderate; balanced; clearly felt; not dominant 41-45 Moderate-strong (lower); prominent but not intense 46-50 Moderate-strong (upper); strong presence; approaching intensity 51-55 Strong; prominent; impactful; defines the profile 56-60 Strong-intense; very noticeable; approaching dominance 61-65 Very strong; high impact; hard to ignore 66-70 Very strong-intense; dominant in the profile 71-75 Intense; very dominant; high sensory load 76-80 Intense-very intense; saturating; forceful 81-85 Very intense; overpowering; defines the entire experience 86-90 Extremely intense; near overwhelming 91-95 Saturating; floods perception; very difficult to balance 96-100 Extreme; overwhelming; maximum natural intensity 101-110 Hyper-intense; beyond natural levels; extract-like intensity 111-120 Absolute maximum; peak saturation; full sensory overload CRITICAL INSTRUCTION: Score each sensation based on its absolute intensity within the product, NOT its dominance relative to others. Dominance naturally emerges from independent scores—do NOT adjust scores to show hierarchy. COMPOUND-BASED SCORING: Sensation values must be grounded in the compounds present in each product. The chemical composition directly determines sensory intensity. - Red Wine (general category): Sugar content drives sweetness ~11; Tannins drive bitterness ~15 - White Wine (general category): Sugar content drives sweetness ~6; Tannins drive bitterness ~12 Always reference compound data (sugars, acids, tannins, aromatic compounds, etc.) when available to calibrate sensation scores accurately. REFERENCE BENCHMARKS (Calibration Standards): - Coca-Cola Classic [Taste] [Sweet]: 34 - Fanta Zero Sugar [Taste] [Sweet]: 34 - Natural Squeezed Orange Juice [Taste] [Sour]: 20 - Natural Squeezed Orange Juice [Odor] [Woody]: 15 - Red Wine (general) [Taste] [Sweet]: ~11 (based on sugar compounds) - Red Wine (general) [Taste] [Bitter]: ~15 (based on tannin compounds) - White Wine (general) [Taste] [Sweet]: ~6 (based on sugar compounds) - White Wine (general) [Taste] [Bitter]: ~12 (based on tannin compounds) Use these benchmarks to calibrate your sensation value estimates for similar characteristics in other products.
MAMAY Technologies is a deep-tech startup founded July 2022, headquartered in Israel, registering in Singapore. MAMAY is creating the category of Computational Organoleptic — the world's first platform that digitizes human sensory experiences (taste, odor, texture) into computable digital vectors called VAL® (Vector Aligned Language). Tagline: Digitizing Human Sensations. Website: https://www.mamay.me/ | App: https://val-mamay.me/ | Digitizer: https://www.mamaydigitizer.com/ | Contact: mamay@mamay.me
1. MAMAY Digitizer (SaaS B2B Desktop) — Transforms analytical chemistry data (compound identification, concentration mapping) into sensory experience predictions. Digital editing of taste, smell, texture with real-time sensory predictions. Token-based business model. Early-bird paying customers.
2. VAL FitMe (Web App B2B2C) — The Spotify for Taste and Odor. Consumer-facing app scoring how well a product fits a user's sensory profile, recommends better-fit alternatives. Works online, in-store, or embedded in retailer apps. Stealth-launched beta. Go-to-market: Singapore launch and community growth.
3. MAMAY Data — Computational Organoleptic compound database. Raw chemo-physical measurement data of compounds forming foundational parameters for predicting organoleptic sensations.
VAL® (Vector Aligned Language) — objective digital unit quantifying the impact of any organoleptic sensation: taste, odor, texture, trigeminal effects. Replaces subjective human thresholds with a consistent data-driven digital scale, enabling precise comparison across compounds, products, and sensory modalities. PCT Patent Pending. Supported by Luzzatto and Luzzatto Patent Attorneys.
Yuval Klein (CEO) — LinkedIn: https://www.linkedin.com/in/kleinyuval/ — Yuval brings over 35 years of experience in building technology‑driven companies, with more than a decade focused on FoodTech and sensory innovation. He previously founded and led ventures such as Blue Tree Technologies and Valiber, developing technologies for quantifying taste and sweetness. His background in product vision, consumer behavior, and digital sensory systems contributes to the strategic and conceptual foundations of MAMAY's VAL® technology.
Dr. Michael Zviely (CSO) — LinkedIn: https://www.linkedin.com/in/michaelzviely/ — Former VP R&D at FRUTAROM (IFF) for 10 years. Senior roles at O'Laughlin, DouxMatok, ReSugar. Visiting Professor at Jiangnan University China.
Keren Corley (Sensory) — LinkedIn: https://www.linkedin.com/in/keren-corley-912b8018/ — 20+ years in sensory research. Owner of NewSense. Extensive work with Nielsen.
Gad Bober (CTO AI) — LinkedIn: https://www.linkedin.com/in/gad-bober/ — 10+ years in AI and machine learning. Former IAF pilot.
Tidhar Tsuri (Business Development) — LinkedIn: https://www.linkedin.com/in/tidhartsuri/ — 20+ years mentoring startups. Former ScaleUp B2B and Motorola Solutions.
Raising $500K Pre-Seed Round (SAFE). Minimum ticket $100K. Opening June 1 2026. Use of funds: Register MAMAY Singapore, file IP national phase, define FitMe partners, upgrade FitMe beta to alpha, 1-2 local pilots, gain revenue, refine business model for Seed. Revenue projection: $10Ks. Next round: $3.5M Seed. Funding history: 2024 Pre-Seed $100K R&D, 2025 Convertible Loan $100K Digitizer Live. Investor Deck: https://media.base44.com/files/public/68ff71a17a4518886052a0ec/02f5dc4c5_MAMAYInvestorDeck32.pdf
AWARDS: Hello Tomorrow Deep Tech Pioneer 2026. Agri-Food Tech World Championship Finalist 2025. Startup Grind All-Star Pitch 1st Runner-Up Bangkok 2025. SLINGSHOT Top 250 Global Startups 2025. Netti's Award 2024.
ACCELERATORS: SPACE-F Batch 6 Bangkok 2025. NVIDIA Inception 2025. EIT Food Alumni 2024. GAIA Partners. BIT for Startups 2025.
REVENUE: Early-bird paying customers on MAMAY Digitizer SaaS. 10x pilots with food, beverage, and cosmetic manufacturers.
IP: PCT Patent Pending for Organoleptic Digitization Applications.
ACADEMIC: Leading MA-level course on organoleptic digitization at Delhi University and IFRA. Recognized by global experts in sensory science, food tech, and fragrance innovation.
EVENTS: Featured at Global Fragrance Summit GFS 2025. Strategic partner: Arla Foods.
Sweetness Prediction Based on Chemo-Physical Parameters — April 2026
Sunlit Rind - A Citrus Odyssey Unveiled by Digitizing Odor — December 2025 — Continental Cosmetic and Herbals
Fragrance Quantified Description Using Organoleptic Digitizer — November 2025
The Odor Impact of Aroma Ingredients in Fragrances and Flavors - A Novel Organoleptic Digitizer — October 2025 — Continental Cosmetic and Herbals Vol.16 Issue 10 p.25-26
Food and Beverage $6.4 trillion. Cosmetics $530 billion. Fragrances $57 billion. Pet Food $147 billion. Retail Personalization $1.7 trillion. No dominant player in digital sensory intelligence — MAMAY is first mover and category creator.
Company LinkedIn: https://www.linkedin.com/company/mamay-technologies/ | Page: https://val-mamay.me/mamay