Chatbots for Customer Engagement
Automate customer interactions through AI-powered digital assistants. The system handles customer enquiries, provides personalised recommendations, collects feedback, and manages lead generation, enabling 24/7 customer engagement while streamlining sales conversions and service improvement.
Please review all requirements. Your solution must meet all 🔴 mandatory requirements to qualify.
17.01🔴
Gen AI Does your solution utilise GenAI functionality to power the Chatbot? GenAI chatbots use a technology called Generative Pre-Trained Transformer (GPT) based on a Large Language Model. Compared to previous generation chatbots, GenAI chatbots have better natural language processing (NLP), providing more accurate responses and engage in context-aware conversations with the customers. Additionally, these chatbots can learn from user interactions and improve their responses over time, resulting in a more personalised and effective user experience.
17.02🔴
Risk mitigation briefing Does the vendor commit to conducting a briefing (online or in-person) for each potential client before client purchases the solution, to ensure that each client is aware of the risks associated with using the Gen AI solution for its business (e.g. accuracy, bias, reliability, toxicity, IP infringement, privacy etc), and the best practices to mitigate these risks (e.g. setting escalation criteria, out-of-bounds markers etc)? Note: 1) Vendors must adequately brief SME participants on these associated risks to ensure responsible Gen AI solution usage. Vendors will also be required to secure SMEs’ acknowledgement of the associated risks before provisioning the solution (e.g. via the customer quotation / contract) 2) Vendors may also consider offering documentation or further information about their models’ training data to their clients. 3) Provide the briefing materials (e.g. slides, pre-recorded video) that will be used to brief each client on the risks of using the Gen AI solution, and risk mitigation measures. 4) Provide a template of the customer quotation / contract where each client's acknowledgement of this briefing will be secured.
17.03🔴
Set-up and data pipeline a) Does your solution enable the company to perform the following functions during set-up? • Upload company-specific knowledge base in varied formats (e.g. FAQ flowcharts, customer service manuals, brand information/guide etc) to contextualise generated responses; • Implement criteria to automatically escalate certain type(s) of customer enquiries to human agents; • Ability for business owners to define the conversation boundaries to restrict the chatbot from providing information outside the scope of the business website.
17.04🔴
AI Chatbot Knowledge Integration b) Can your solution ingest at least TWO of the following data sources to build the chatbot's knowledge base? • Direct website crawl (with website URL as input) • Direct database integration (via SQL, API etc) Examples of database integration may include: - CRM (Customer data such as transaction history, communication history, loyalty programme etc) - Sales or E-Commerce platform (Sales data such as SKUs, available stock, prices, campaign, promotion etc) - Any other integrations to optimise and enhance customer engagement • Structured files, such as FAQ with Question-and-Answer pairing, Product and Services offerings etc • Unstructured files, such as PDF or Word Document
17.05🔴
Integration with customer communication platform(s) Can your solution be deployed on at least ONE of the following digital platforms/channels for customer engagement? • Company website • Messaging apps (e.g. WhatsApp, Facebook Messenger, Telegram, etc) • Email • Social media (e.g. Instagram, TikTok) • Customer Relationship Management (CRM) system • Audio calls / call center Note: Vendor must provide implementation services to ensure that the solution is successfully integrated with the company's selected digital platform(s)
17.06🔴
Either/Or Requirement Can your solution fulfill EITHER 17.06A OR 17.06B? Please indicate "Yes" for both 17.06A and 17.06B if your solution can fulfill BOTH requirements.
17.06A
Front-end customer engagement capabilities Can your solution provide a customer-facing chatbot that demonstrate at least TWO of the following capabilities? (1) Leads Management: to automatically collect information on and follow up on leads, facilitating tracking and conversion. (2) Sales Assistant: to guide customers through the sales process, answering informational enquires and providing recommendations. For chatbots taking the role of sales assistant in an e-commerce environment, the two following features must be demonstrated: - Able to assist the customer to complete an end-to-end sales via the chatbot conversation - Able to provide product upselling and cross-selling recommendation during the sales checkout process (3) Post-Sales Customer Support: to perform post-sales support, providing customers with information such as order status, product/service support, escalation. (4) Scheduling/Reservation: to assist customers in making bookings based on resource (e.g. personnel, facility) availability, customer's past preferences, etc.
17.06B
Back-end customer engagement capabilities Can your solution provide an internal-facing chatbot that demonstrate at least THREE of the following capabilities? (1) Suggestions/Improvements for Customer Engagement Agents: Generate real-time suggestions to improve responses or automatically translate messages. (2) Automated Reminders for Follow-Ups: Provide timely reminders to agents to provide follow-up actions for all instances of customer engagement. (3) Centralised Platform for Multi-Channel Customer Engagement: Provides a centralised platform for agents to manage client conversations over multiple channels (e.g. Website, Social Media, Messaging Apps etc.) (4) Real-Time Interventions: Enable managers to intervene conversations and take actions where necessary. (5) Creation of Customer Profiles: Identify customer characteristics and build comprehensive customer profiles, based on conversations with customers.
17.07
Post-deployment self-learning and training Can your solution self-learn to improve the accuracy and/or relevance of chatbot responses generated over time? For example: • Reinforcement learning / feedback loop based on positive and negative customer interactions • Learning whenever a human agent takes over to answer customer enquiries • Manual "supervised" training / customisation of chatbot responses and conversation flows • Re-training of the chatbot based on business updates to their business website or information.
17.08
Citations Can your solution automatically cite sources or references as part of its factual responses? (e.g. to reduce hallucination risk when answering factual enquiries)
17.09
Languages Can your solution support conversations with customers in more than ONE language, other than English? (e.g. Simplified/Traditional Mandarin, Bahasa Melayu, Tamil etc)
17.10
Audio-based conversations Can your solution support conversations with customers over voice chat (instead of text)?
17.11
Mandatory to provide a response
Personal Data Protection Can your solution demonstrate compliance with the following Personal Data Protection requirements? Digital solutions that collect, use, disclose, process or dispose personal data should incorporate features that support the obligations under the Personal Data Protection Act (2020). Please complete the Compliance with the Personal Data Protection Requirements form at https://go.gov.sg/pdp
17.12
Vulnerability Assessment/Penetration Testing (VA/PT) You will need to comply to the following requirement IF your solution handles or stores Personal Identifiable Information (PII) and/or Sensitive Personal Information (SPI): Has your solution undergone a comprehensive security vulnerability assessment/penetration testing (VA/PT) conducted by a qualified third-party within the last 12 months? The scope of the VA/PT must cover network security; application security; data protection measures and access control (if applicable); API security testing (if applicable); Cloud security configuration review (if applicable). Specifically, for web application security, the scope must cover minimally all OWASP Top 10 vulnerabilities. Please submit the VA/PT report (dated maximum 1 year from the checklist submission date). The VA/PT Report must include Executive summary; Detailed findings and risk ratings; Remediation recommendations; Evidence of vulnerability fixes or mitigation plans; Testing methodology used; Scope of assessment; Assessor's qualifications and certifications. If you are the reseller of the solution, please obtain the VA/PT report from your product principal. SOC 2 Type II report can be accepted if the detailed technical vulnerability assessment results are part of the SOC2 Type II scope. Note: [1] Qualified third-party refers to: CREST-certified companies [ https://www.crest-approved.org/members/] or companies with security professional with relevant CREST certifications; Security professionals with recognised certifications such as: Offensive Security Certified Professional (OSCP); EC-Council Certified Penetration Testing Professional (CPENT); GIAC Penetration Tester (GPEN); or other equivalent industry-recognised certifications.
17.13
Project Moonshot Has your solution undergone evaluation by Project Moonshot? Project Moonshot is a LLM Evaluation Toolkit, designed to integrate benchmarking, red teaming and testing baselines. It helps developers, compliance teams, and AI system owners manage LLM deployment risks by providing a seamless way to evaluate their applications’ performance, both pre- and post-deployment. More information on Project Moonshot is provided at https://aiverifyfoundation.sg/project-moonshot/ 1) Please provide the report generated by Project Moonshot upon completion of the evaluation, where vendors are required to test their Gen AI system with the benchmarks stated below, which are available on Project Moonshot: • MLCommons AI Safety Benchmark - Hate • Bias Benchmark for QA (BBQ) • UCI Adult Dataset 2) Complete the Moonshot testing form. https://form.gov.sg/6699c08bfc8b1217e4d6662e\
Last updated