《投资促进机构观察家》：投资促进与便利化见解

Turning Potential into Practice Investment Promotion Agencies (IPAs) are under growing pressure to deliver high-quality investments,with faster, smarter, and more targeted services to investors. Artificial intelligence (AI) offerstransformational opportunities, from automating routine tasks to identifying high-potential investorsthrough machine learning and predictive analytics. Yet AI adoption raises important questions aboutaccessibility and implementation frameworks. This IPA observer presents key findings on how Understanding AI AI is “the capability of a machine to engage in cognitive activities typically performed by the humanbrain”.2 For the purposes of this report, three types of AI are particularly relevant to IPAs and their Discriminative AI:The most established type, which focuses on classifying data and identifyingpatterns. In IPA practice, this includes chatbots, document validation, permit navigation, and Predictive AI:This system uses historical data to predict outcomes and inform decision-making.When applied to IPA practice, it can be used for lead scoring, modelling investment likelihood, and Generative AI:The current frontier, capable of producing entirely new content rather than simplyclassifying or forecasting. Within IPA applications, this includes proposal generation, marketing AI adoption remains concentrated inIPAs from high-income economies 82 per cent of IPAs reporting on the use of AI are from high-income or upper-middle-income countriesand 16 per cent from Least Developed Countries (LDCs) and Small Island Developing States (SIDS)(see Annex I). Additionally, UNCTAD’s research, examining the websites and digital platforms of 44LDCs and 32 SIDS, identified only 10 IPAs (13 per cent) that have implemented chatbots or visible AI Meaningful AI integration requires foundational digital capabilities, such as data infrastructure,technical expertise, digitalised processes, and innovation resources, which are not widespreadin developing countries.3On the contrary, developing countries tend to face significant barriers,including limited digital infrastructure, skills gaps, and data governance frameworks, which certainlyaffect IPAs’ capacity to deploy AI.4However, experience from early-adopter IPAs from developingcountries suggests that AI adoption does not necessarily require full readiness. The targeted use AI expansion for IPAs could significantly decrease repetitive work tasks, improve response times toinvestor inquiries, and enhance overall investor engagement. Rather than discouraging late adopters,the findings clarify that AI adoption requires careful preparation, with tailored pathways for IPAs at Four core functions of AI in IPAs Research analysis of early adopter IPAs (see Annex I) revealed that AI deployment clusters around fourdistinct functions (Table 1), each with different value propositions and technical requirements.Thefirst three, operational automation, information synthesis, and predictive decision-making, classify,synthesise, or forecast from existing data, progressing in sophistication from pattern recognition toforecasting. The fourth, generative production, goes further, producing entirely new and customised The IPA ObserverInvestment promotion and facilitation insights Operational Automation:Among narrow AI applications, the most accessible use involvesautomatingrepetitive tasks,particularly investor inquiries and information provision.Invest inEstonia has deployed multiple AI tools, including an e-advisor for automated inquiries, named Eia,with measurable efficiency gains in staff time and response speed. In the case of Investment Fiji, Invest in Sharjah (UAE), Invest Guyana, Jamaica’s JAMPRO, and Belize’s Beltraide, the chatbotshandle routine inquiries and direct investors to relevant services and incentives. More specifically,Invest in Sharjah applies AI to automate trade licensing, recommend optimal license types, andvalidate documents in real-time.6More sophisticated operational applications include case routing Information Synthesis:More sophisticated AI applications enable knowledge managementthrough AI’s ability to process and structure unstructured information. For example, Invest inEstonia’s toolsEmmaandCompareEST, respectively monitor media to assess investor sentiment inreal time, and compare systematic competitiveness using machine learning.7Flanders Investmentand Trade8uses AI-powered enterprise search and question-answering systems that help advisorsextract insights from institutional knowledge.9Invest India uses AI tools to scan large volumes ofdata, including news articles and earnings call transcripts, to extract high-value investment insights Predictive Decisioning:AI targeting applications use machine learning to predict investmentlikelihood, enabling resource concentration on promising opportunities. CINDE from Costa Ricauses predictive analysis tracking over 800,000 companies globally with 150+ data points to identifyinvestment p