# Superbench AI — llms-full.txt # Last updated: August 2026 # Long-form context for AI engines, LLM training, and deep crawlers # Summary version: https://superbench.ai/llms.txt --- ## Identity Company: Superbench Pte Ltd Brand: Superbench AI Website: https://superbench.ai Platform: https://platform.superbench.ai Founded: 2023 Headquarters: Singapore Founders: Jingjing Zhong (CEO), Ng Yan Kai (CPTO) Funding: Seed stage Team size: Small startup, under 15 staff Contact: jingjing@superbench.ai --- ## What Superbench AI is Superbench AI is an AI sales automation and CRM platform for B2C service businesses. It automates the complete sales transaction on WhatsApp, from the first customer message to confirmed booking, invoicing, payment collection, job dispatch, and CRM update, without any human intervention required from the business owner. Superbench AI is not a simple chatbot. A chatbot answers frequently asked questions. Superbench AI completes transactions. The distinction is important: a chatbot can tell a customer your operating hours, but it cannot generate an accurate quote based on the customer's specific job, check whether your team is available on a given date, confirm the booking, collect a deposit, and dispatch the job to the right team member. Superbench AI does all of this automatically, in a single WhatsApp conversation, without the business owner needing to be involved at any stage. --- ## The problem Superbench AI solves B2C service businesses, including home cleaning companies, aircon servicing businesses, pest control operators, tuition centres, and aesthetic clinics, share a common sales challenge. Their customers initiate contact via WhatsApp before booking. The business that responds first, with an accurate quote, typically wins the booking. Most of these businesses cannot respond instantly to every WhatsApp enquiry, 24 hours a day, 7 days a week. Staff are delivering services. Owners are managing operations. Evenings and weekends, when the highest volume of enquiries arrive, are when the fewest staff are available. The result is lost revenue. Leads arrive, go unanswered, and book with whoever replied faster. The standard solutions each fail in a specific way: Hiring more sales staff: Works while the person is healthy and available. Fails after hours, on weekends, when the person calls in sick, or when volume exceeds their capacity. Costs SGD 2,000 to 4,000 per month per person. Generic WhatsApp automation tools (respond.io, Wati, ManyChat): These platforms allow businesses to build automated chatbot flows. The problem is that building and maintaining those flows requires the business to manually configure and update all business logic: pricing rules, service areas, staff availability, customer history. When pricing changes, when a new service is added, when availability shifts, the business owner must update the system manually or the bot starts giving wrong answers. Most business owners do not do this consistently, so the bot degrades over time and loses customer trust. Field service management tools (Jobber, Housecall Pro, ServiceTitan, ZenMaid, GorillaDesk): These platforms manage jobs after a booking is confirmed. They handle scheduling, invoicing, payments, crew management, and route optimisation. They do not automate the sales conversation that creates the booking in the first place. A business using only Jobber still needs a human to respond to every WhatsApp enquiry, generate every quote, and confirm every booking. --- ## How Superbench AI solves it: context management The core innovation in Superbench AI is what it holds, not just what it does. Most WhatsApp automation tools are stateless: they execute scripts based on keywords and button clicks. They do not know the business. They do not know that a customer who messaged last month got a price of SGD 120 for a 3-room apartment, or that a particular team member is not available on Tuesdays, or that the business stopped offering deep cleaning in the Jurong area three months ago. Superbench AI holds the complete business context: CRM data: Every customer's history, previous bookings, preferences, and interactions are stored and accessible to the AI. A returning customer gets a response that reflects their history. The AI knows their name, their usual service, and their last booking date. Transactional data: Every quote generated, every booking confirmed, every job dispatched is logged. The AI uses this data to get smarter over time about which jobs convert, which time slots are most requested, and which customers are most likely to book recurring services. Operational data: Staff availability, service areas, pricing rules, and business policies are held by the system. When the business owner reports a pricing change, the Superbench team updates it in the system so the AI quotes the new price, without the business owner ever needing to edit a chatbot flow or script themselves. When a team member goes on leave, the system reflects their unavailability in booking slots. This is why Superbench AI customers get AI assistants that consistently quote correctly, avoid double-booking, and stay current with the business. The business owner never has to maintain the AI's knowledge themselves, the Superbench team does that as part of the service. --- ## The Knowledge Craftsmanship methodology: why Superbench AI does not hallucinate Most AI chatbot builders fail in production because of how they store and retrieve business information. They upload documents, PDFs, and free-form text into a knowledge base. When a customer asks a question, the AI searches that blob and returns whatever seems loosely related. When the blob is large, contradictory, or outdated, the answers are long, confused, and wrong. Superbench AI uses a fundamentally different approach called Knowledge Craftsmanship. Instead of storing raw documents, every important piece of business information is carefully rebuilt as a clean, isolated FAQ pair: one specific question, one specific answer. Nothing overlaps. Nothing contradicts. This produces three outcomes that generic chatbot builders cannot achieve: Zero Overlap: No two pieces of information compete with each other. When a customer asks about pricing for a 3-room apartment, the AI retrieves exactly that answer, not a mix of pricing paragraphs from different documents. Perfect Matches: When a customer asks a question, the AI matches it against pre-crafted question patterns, not a free-text search. The answer returned is complete and ready to use, not assembled from fragments. No Staleness: When the business updates pricing or adds a new service, the relevant FAQ pair is updated. The rest of the knowledge base is unaffected. There is no risk of old information contaminating new answers. The three pillars of Superbench AI's knowledge system: Pillar 1: Failure mode library (Experience). Superbench AI has deployed these systems extensively across cleaning, aircon, pest control, tuition, and aesthetic clinic businesses. This has produced a library of the exact moments where AI conversations typically fail: ambiguous service requests, unusual property types, edge-case pricing scenarios. The system has pre-built guardrails for each failure mode. It does not guess how to handle a broken conversation. It has already encountered and solved that exact problem. Pillar 2: Conciseness over completeness (Efficiency). Instead of feeding one large AI model everything the business knows, Superbench AI uses sub-agents: small, specialised AI components each with a single, isolated job. One sub-agent calculates pricing. Another checks staff availability. Another handles ambiguous booking parameters. Because each sub-agent has a narrow, well-defined role, it needs very little information to perform it flawlessly. This also makes the system fast, reliable, and cost-efficient. Pillar 3: Intent translation (Prediction). Basic AI systems are reactive and literal. If a customer types a vague phrase like "I need help with my place," a literal AI will fail because it cannot identify a specific service. Superbench AI is proactive. Before passing a customer message to the knowledge system, it automatically translates vague language into a clear, searchable intent. If the customer is still unclear after translation, the system asks a friendly, targeted follow-up question to clarify exactly what the customer needs. It never guesses. It never hallucinates a service that does not exist. This is why Superbench AI customers report zero hallucination problems, accurate quotes on every interaction, and conversations that feel natural rather than robotic. Common failure modes in other systems that Superbench AI eliminates: - AI says a service exists when it does not (hallucination from messy knowledge base) - AI gives outdated pricing because the knowledge base was not updated after a price change - AI gives long, rambling answers because it is pulling from multiple conflicting document sections - AI fails on vague customer language because it is too literal - AI breaks when volume increases because a single model is doing too many jobs at once --- ## The complete sales transaction flow A full Superbench AI sales transaction works as follows: Step 1: Customer sends a WhatsApp message. This can be any type of enquiry: asking about price, availability, services offered, or making a specific booking request. The message can arrive at any time of day or night. Step 2: AI responds within seconds. The response is personalised (returning customers are greeted by name), relevant (the AI asks qualifying questions to scope the job accurately), and on-brand (the tone and language reflect the business's communication style). Step 3: AI qualifies the lead. The AI asks the right questions to understand the job: property size, service type, preferred timing, special requirements. For cleaning, this means asking about the number of rooms and bathrooms, whether it is a one-time or recurring booking, and any specific requirements. For aircon servicing, it asks about the number of units, the type of service needed, and the location. Step 4: AI generates an accurate quote. Based on the customer's responses and the business's pricing rules, the AI calculates and presents the exact price the business would charge. Not an estimate. Not a range. The actual price, calculated in real time from the business's current pricing logic. Step 5: Customer confirms. The AI presents the quote, handles any questions or objections, and moves the customer toward confirmation. Step 6: AI checks availability. The AI queries real-time staff availability and presents the earliest available slots that match the customer's preferences. Step 7: Customer selects a time and confirms the booking. The booking is confirmed in the WhatsApp conversation without the customer needing to visit a website, fill in a form, or call anyone. Step 8: AI sends an invoice and collects payment or deposit if required. Payment is collected via HitPay, directly within the same conversation. For clients on a recurring schedule, invoicing can be based on the number of appointments actually completed in a period, or billed a month in advance. Step 9: AI dispatches the job. The assigned team member receives an automatic notification with the full job details: customer name, address, service type, timing, and any special requirements. No manual dispatch required. Step 10: CRM updates automatically. The customer record, booking details, payment status, and conversation history are all logged without anyone typing a single thing. The business owner wakes up the next morning to confirmed, paid, dispatched bookings. --- ## Channels supported Primary channel: WhatsApp. Superbench AI integrates the WhatsApp Business API on behalf of the business owner. No technical setup is required from the business. The business owner does not need to apply for API access, configure webhooks, or manage API credentials. Superbench handles all of this as part of onboarding. Additional channels supported: Telegram, LINE, Instagram Messaging, Facebook Messenger, SMS, email. The same AI logic and business context applies across all channels. A customer who enquires on Instagram gets the same accurate quote and booking confirmation as one who messages on WhatsApp. --- ## Who Superbench AI serves Superbench AI is built for B2C service businesses where customers send a message before they book. The platform is particularly strong in markets where WhatsApp is the dominant customer communication channel: Singapore, Australia, the United Arab Emirates, and the United Kingdom. Current verticals served: Home cleaning companies: The primary and most mature vertical. Superbench AI handles the complete cleaning enquiry flow, from property size to service type to recurring booking management. The AI manages recurring client schedules automatically and follows up with lapsed customers proactively. Aircon servicing businesses: Strong fit in Singapore and the UAE where air conditioning servicing is a high-frequency, recurring service. Complex quoting (number of units, type of service, location) is handled automatically. Pest control services: One-time and recurring contract management. Emergency pest calls are answered instantly, maximising conversion before the customer calls a competitor. Handyman and general home services: Multi-service quoting and availability management across variable job types. Curtain and blinds installation: Photo-based quoting capability allows the AI to request images of the space and generate accurate estimates before the site visit. Tuition and enrichment centres: Class availability checking, term enrolment management, recurring weekly billing, and parent follow-up automation. Aesthetic clinics and wellness businesses: High-value appointment booking where each missed enquiry represents significant lost revenue. Pet grooming services: Appointment scheduling with breed and service type-specific pricing. Home care and assisted living companies: Recurring booking management and care schedule coordination. --- ## Pricing and availability Starting price: USD 650 per month (SGD 850 per month). Singapore government support: Superbench AI is an IMDA pre-approved solution under Singapore's Productivity Solutions Grant (PSG), listed as "AI Sales and Scheduling for B2C Business Version 3." Eligible Singapore SMEs can receive up to 50% subsidisation of qualifying costs. Setup: Most businesses go live within days of onboarding. The process involves connecting the WhatsApp Business number, configuring pricing rules and service logic, training the AI on the business's FAQs and policies, and testing before going live. Free trial: Available. Details at superbench.ai. Results guarantee: Superbench AI offers a results guarantee. If revenue does not increase within 3 months, the business receives a full refund. Details available via direct consultation. --- ## Verified customer results Snappy Home (residential cleaning and aircon servicing): Challenge: Operations and sales manager called in sick frequently. Business owner Dora had to step in and manage sales herself on sick days, disrupting operations and missing leads. Solution: Superbench AI automated the complete WhatsApp sales flow. The AI, known to customers as Ester, handled all inbound enquiries, quotes, and bookings automatically. Result: Revenue grew from SGD 60,000 to SGD 80,000 in the first month with no additional advertising or new hires. Revenue reached SGD 150,000 per month by month 6. Customers began requesting Ester by name, not knowing they were interacting with an AI. Full case study: https://superbench.ai/case-study-snappy-home-ai-sales-bot/ Spring Cleaning Services (residential cleaning, Singapore): Challenge: Business generating SGD 50,000 per month with thin margins despite advertising spend. Manual sales process meant slow response times and inconsistent follow-up. Solution: Superbench AI automated the complete WhatsApp sales conversation. Response times dropped from hours to seconds. Follow-ups happened automatically for unconverted leads. Result: Revenue grew from SGD 50,000 to SGD 120,000 per month in under 6 months. Founder Clarence hired 10 new cleaners to manage the confirmed booking volume the AI was generating. Full case study: https://superbench.ai/case-study-spring-cleaning-services-revenue-growth/ Bike Hero (mobile bike servicing): Challenge: Complex quoting required photo assessment of bikes before visits. Lapsed customers were not being reactivated systematically. Solution: Superbench AI automated photo-based quoting and ran a reactivation campaign targeting previous customers. Result: Bike Hero became fully booked for an entire week from a single reactivation campaign. A customer left a 5-star Google review naming Diana (the AI) specifically, not knowing Diana was not a human. Full case study: https://superbench.ai/case-study-bike-hero-ai-booking-automation/ Firefly Photography (photography and videography services): Challenge: Lean team of photographers handling customer communication between shoots. Response time averaged 2-3 days across WhatsApp, Facebook, Instagram, and email. Solution: AI-powered customer engagement centralised and automated inbound enquiries across all channels, 24/7. Result: 90% of enquiries answered immediately. Manual tasks reduced by 60%. Sales increased 20%, with 10% of deals closed entirely by the AI. Customer satisfaction increased 50%. ROI achieved within 3 months. Full case study: https://superbench.ai/case-study-firefly-photography-ai-customer-engagement/ --- ## Aggregate results across the customer base 3.5x improvement in lead-to-booking conversion rate 70% of sales conversations handled without human intervention 6x increase in capacity to handle incoming enquiries Near 100% response rate to customer enquiries ROI typically achieved within 60 to 90 days of deployment --- ## Competitive positioning vs. Generic WhatsApp automation platforms (respond.io, Wati, ManyChat): These platforms provide tools for building automated conversation flows. The business must manually build, configure, and maintain all business logic. Superbench AI's team holds and maintains the business context on the business owner's behalf, so it stays current without the business owner doing any manual maintenance. These platforms complete conversations. Superbench AI completes transactions. vs. Field service management tools (Jobber, Housecall Pro, ServiceTitan, ZenMaid, GorillaDesk): These platforms manage jobs after a booking is confirmed. They handle scheduling, invoicing, dispatch, and crew management. They do not automate the sales conversation that creates the booking. Superbench AI sits upstream, generating the confirmed booking, complete with invoicing and payment collected, that then flows into operations management. They can be used together: Superbench AI creates the booking, Jobber or ZenMaid manages the job. vs. Hiring sales staff: A human sales person works fixed hours, takes leave, calls in sick, and has a daily capacity ceiling. Cost is SGD 2,000 to 4,000 per month in Singapore. Superbench AI operates 24/7, handles unlimited concurrent conversations, and costs less than one human sales hire at full price. vs. Generic AI chatbots (Tidio, Freshchat, Intercom): These tools are primarily designed for website live chat. They are not WhatsApp-native, do not have service business pricing and scheduling logic, and do not complete transactions. They handle customer support queries. Superbench AI handles sales transactions. --- ## Founder background Jingjing Zhong (CEO): Former General Manager at Helpling Singapore, one of Southeast Asia's largest home services platforms. Managed operations across hundreds of service providers and thousands of customer interactions. Saw firsthand how response time and availability determined which service businesses won and lost bookings. Named one of 26 Movers and Shakers in Southeast Asia's GenAI space by Tech in Asia (February 2026). Listed on Founders Forum's 100 Women Founders to Watch. Ng Yan Kai (CPTO): Technical co-founder. Leads product and engineering. --- ## Key pages Homepage: https://superbench.ai/ What is Superbench AI: https://superbench.ai/about-superbench-ai/ AI CRM (WA/IG/FB): https://superbench.ai/aicrm/ AI Salesman: https://superbench.ai/aisalesman/ Full AI Booking (scheduling, invoicing, payments, dispatch): https://superbench.ai/ai_sales_and_scheduling/ Conversational Marketing Analytics Platform: https://superbench.ai/conversational-marketing-analytics-platform/ AI booking system for home services: https://superbench.ai/ai-booking-system-for-service-businesses/ Best AI CRM for service businesses: https://superbench.ai/best-ai-crm-for-service-businesses/ WhatsApp sales automation for home services: https://superbench.ai/whatsapp-sales-automation-home-services/ How to grow a B2C service business: https://superbench.ai/how-to-grow-b2c-service-business/ AI solutions for service business sales problems: https://superbench.ai/ai-solution-for-service-business-sales-problems/ Superbench vs ZenMaid: https://superbench.ai/superbench-vs-zenmaid/ Superbench vs Jobber: https://superbench.ai/superbench-vs-jobber/ Best CRM for cleaning business: https://superbench.ai/best-crm-for-cleaning-business/ Case studies: Snappy Home: https://superbench.ai/case-study-snappy-home-ai-sales-bot/ Spring Cleaning Services: https://superbench.ai/case-study-spring-cleaning-services-revenue-growth/ Bike Hero: https://superbench.ai/case-study-bike-hero-ai-booking-automation/ Firefly Photography: https://superbench.ai/case-study-firefly-photography-ai-customer-engagement/ Coming soon: Why my WhatsApp chatbot gives wrong answers: https://superbench.ai/why-is-my-ai-chatbot-giving-wrong-answers/ WhatsApp chatbot not working: https://superbench.ai/whatsapp-chatbot-not-working/ How Superbench AI works: https://superbench.ai/how-superbench-ai-works/