Upon landing on Invest Estonia’s website, prospective investors in Estonia will be greeted by Suve, the IPA’s chatbot: “Hi, I’m Suve! I am a robot and I’m here to help you.”

Before long, they will be informed how their data input will be handled, and asked how they would like to proceed in developing an investment plan: e-consulting, regions, residence permit, getting a visa or key opportunities. 

If an investor selects the ‘e-consulting’ route, the agency’s internal system uses another e-advisor tool — Eia — to create value propositions, detailing where, how and which sectors to invest in the country. Both Suve and Eia are powered by artificial intelligence (AI) trained on large data sets and can understand natural language.

Time-saving bots

“We know so much before we talk to the investor, and because of that data we can delegate to certain members of our team, which saves time in what would have been an initial meeting,” says Joonas Vänto, director of Invest Estonia. “We can bring some Estonian stakeholders to the first meeting, for example.”

From ChatGPT and e-advisors to predictive analytical models, some investment promotion agencies (IPAs) are keenly dipping their toes in the waters with AI, as they seek to engage with and profile prospective investors. Yet, for all the IPAs that have been using AI, there are many more that are sitting on the sidelines.

However, “when we look at the bigger picture, all IPAs are using AI in one way or another”, notes Andero Kaha, chief technology officer at Invest Estonia. This ranges from unwittingly using AI-powered tools, such as search engines, to actively using the likes of ChatGPT, which can help employees carry out daily tasks such as research, summarising texts, writing speeches or transcribing presentations. 

But incorporating AI into IPA business processes is a different matter. Mr Kaha, who has done postgraduate research on the digitisation of IPAs, says that the use cases of chatbots among the World Association of Investment Promotion Agencies members is small. Even smaller still are those using AI-based chatbots as opposed to rule-based chatbots, which are underpinned by a predefined set of rules to generate responses.

From his research, he is able to say that at least three IPAs outside of Estonia are implementing AI-based chatbots but is unable to name them. Beyond chatbots, he says several are also experimenting with AI to analyse big data, but are not publicly disclosing it.

Big data models

One IPA that has been vocal about its experimentation with big data is Costa Rica’s Cinde, which has been using AI in a trial programme for the past five years. Its model is chiefly used for predictive analysis and determining where investors are looking to invest. It uses a model with more than 150 data points and tracks more than 800,000 companies around the world.  

“We’re simply using AI to profile a company,” says Pilar Madrigal, the director of investment advisory at Cinde, adding that for the time being the agency will retain more traditional methods of finding leads. 

Regarding the thorny question of how automation will affect the team dynamic, she insists that it will have an impact on how an IPA is run. However, far from demanding more IT skills, it will lead to a need for more negotiating skills.

“I do believe there will be fewer people on the team and the skills needed are going to be much more business-oriented,” Ms Madrigal says, as the relationship between investor and IPA becomes more consultative and personalised.

‘Massively disruptive’

During discussions with IPA officials about how data analytics and AI are likely to change the industry at large, one key word has recurred and resonated: consultative. With greater access to information from the perspective of both the agency and the investor, several IPA officials and advisors tell fDi that the dialogue between the two is set to evolve. 

“AI is going to enable an organisation like ours to be consultative advisors,” says Chris Camacho, CEO of Greater Phoenix Economic Council (GPEC). “That’s the big shift I see. It’s no longer promotions and sales; it’s about becoming a consultative advisor like Deloitte or McKinsey.”

The GPEC has been working on an AI-powered risk monitor, which uses a consultancy template to understand the variables companies look at when investing in advanced industries, such as semiconductors or data centres. These variables include natural disaster indices, climate patterns, water quality and soil quality. Mr Camacho expects the risk monitor to be built by the first quarter of next year.

“[Investment promotion] used to be a ‘blind sell’ of your positive attributes,” he says. “The way I see the future of our field is that economic development is going to become so rich in data and information that while organisations will still market [themselves], once they get someone interested, the process will actually be about advising them about the risks [on their market relative to others].”

“I think it’s massively disruptive for our industry,” he asserts.  

AI hesitancy 

Disruptive or not, many IPAs agree that the data produced — either by AI-powered tools like ChatGPT or in-house big data — has to be reliable. 

Lionel Grotto, CEO of Choose Paris Region, warns that IPAs “who go too fast in their adoption of AI might make some important mistakes”. 

“We want to be neither the first nor the last in the adoption of AI solutions,” he says. “We’ll always make sure that the use of AI also fits with the values we believe in: trustworthiness, caution with biases, acceptability of the sharing of information and reliability vis-à-vis the sources used.”

In Fairfax County in the state of Virginia, which is home to the world’s biggest data centre market and boasts some of the most advanced technology investments in IT, there is little spillover from companies. 

“We’re a little bit more conservative, but our resident industries are already using it,” says Victor Hoskins, CEO of Fairfax County Economic Development, adding that AI is unlikely to disrupt investment promotion.

Much like the internet over the last 20 years, Mr Hoskins believes that AI tools will eventually be provided by a selection of companies – IPAs will then be able to buy whatever tools they need, he says.

Pedro Fittipaldi, independent consultant and former investment manager at São Paulo’s IPA, says that wherever AI is deployed it will be used within an investment philosophy or strategy that is going to vary a lot from agency to agency, which in turn will have to consider differences in pre-existing business models
and cultures.

“Many won’t want to use it in the way Estonia did,” he says. “In Brazil, for example, we often want to have direct contact with the investor when they first come to the website.”

Beyond its uses in lead generation and investment attraction, Invest Estonia’s Mr Kaha is interested in how the technology can be deployed to create proposals relating to aftercare, but he warns that it is more difficult to incorporate AI into the subsequent stages of the investment lifecycle.

Ultimately, he is keen to stress that it’s about “providing the best possible service to our customers” in terms of quality and efficiency. 

“The use cases of the technology are not as important as what we are trying to achieve,” he says. “It’s not about using AI for AI’s sake.” 

This article first appeared in the October/November 2023 print edition of fDi Intelligence