The Role of AI in Investment Research and Responsible Investing 

What investment firms need to know about adopting AI safely, effectively in investment research. 

Investment research teams are facing a growing challenge: data overload. ESG reports, regulatory filings, earnings transcripts, and sustainability disclosures are just a handful of the information sources analysts must navigate daily. Yet in many firms, manual document review and basic keyword searches remain the norm. 

AI is starting to shift this. While adoption is still in early stages, a growing number of firms are exploring how AI tools can help them analyse vast volumes of complex information and uncover insights that would be hard to spot manually. But the path to practical implementation isn’t always straightforward. 

Below, we explore some of the strategic options firms are considering — and what to bear in mind if you’re looking to adopt AI in a responsible, value-adding way. 

AI Adoption Strategies: From Light-Touch to Fully Bespoke 

There are several principal AI strategies that firms can choose to adopt – either singularly, or through a combination. These are:  

1. Using Open-Source AI Tools

Allowing employees to use tools like ChatGPT, FinGPT or Gemini is often the default for firms who do not proactively engage with, nor shape their AI strategy, will adopt. While these solutions are highly flexible, and there are no upfront costs, the potential losses which could arise due to a leak of proprietary information can be huge – particularly for firms such as Asset Managers and Pension Funds who are accessing sensitive / personal information.  

2. Adopting Enterprise AI Platforms

Adopting a cost-effective, but integrated AI tool (e.g. Microsoft CoPilot), and subsequently educating and upskilling employees from across the organisation to develop specific use cases. While this option may not yield as significant or rapid benefits as more bespoke solutions, it does provide a secure and cost-effective way of boosting productivity across an organisation by empowering all end-users to develop their own use cases / solutions. The critical determinant of this AI model’s success is the organisation of the underlying data – i.e. “input determines output”.  

3. Building Proprietary Large Language Models

Developing an in-house large language model (e.g. BloombergGPT) which is tailored to the organisation’s specific investment research / responsible investing needs, while also ensuring enterprise-grade security and privacy. Bloomberg have used this internally to great effect and have also commoditised this to sell to existing Bloomberg customers (N.B. A Bloomberg terminal is required). The tool is so successful that it is now outperforming generalist models like GPT-3 in finance specific tasks. The downside of this approach however is the significant upfront investment and configuration, as well as ongoing maintenance – meaning it is only really a viable strategy for the largest market players.  

4. Procuring Specialised AI Research Platforms

Procuring a specialised knowledge management tool which delivers comprehensive investment analysis and deep research (e.g. Hebbia AI, Sentieo or AlphaSense). Hebbia AI, for example, leverages several OpenAI models, Anthropic’s Claude as well as its own proprietary engine, to retrieve internal information, summarise, compare and even draft investment outputs (e.g. investment memos or contracts). AlphaSense and Senteio on the other hand use external financial content to synthesise investment research and responsible investing insights, as well as allowing users to upload internal files too.  

5. Acquiring Asset Class-Specific AI Tools

Some firms are choosing to implement AI tools tailored to specific asset classes, especially where unique data types or workflows are involved. 

For example: 

• Equities: Hebbia AI, AlphaSense, Sentifi, Bridgewise, Kavout 

• Fixed Income: Overbond, BondIT 

• Macro Analysis: Predata 

• Real Estate: Cherre, Reonomy 

• Private Equity: Affinity, Grata 

• ESG: Clarity AI, SESAMm 

These tools often support specialised tasks like ESG scoring, credit analysis, alternative data modelling or private market research. 

Asset & Wealth Management Platforms Starting to Embed AI 

In parallel to standalone AI tools, major system providers are embedding AI into their platforms. For example, BlackRock has invested heavily in to making Aladdin a smart, connected AI platform which combines predictive analytics, natural language processing (NLP), generative agentic interfaces, ESG insights, and scalable infrastructure. Similarly, Bloomberg have devoted time and resources to embedding AI into their AIM platform, to deliver automated earnings call summaries, document searches to allow users to retrieve insights, themes and analyst Q&A summaries. 

For many firms, making the most of existing platforms, rather than adopting entirely new tools, may offer the most pragmatic route to value. 

The Trade-Offs of AI Adoption (and How to Manage Them) 

While AI can yield significant benefits to organisations (e.g. efficiency, enhanced decision making, scalability, personalisation and increased employee/customer satisfaction), there are numerous risks that need to be considered carefully too. Some of the key risks are noted below: 

Cost 

Implementation and ongoing costs for AI tools can be significant, however so too can the cost of doing nothing. For example, inertia may allow other market participants to gain competitive advantage, and lack of strategy may lead to the leakage of proprietary information via web-based AI tools.  

Security 

AI tools, especially open-source or third-party platforms, can expose firms to serious data security risks, including leaks of proprietary or sensitive client information. Without strong governance, even well-meaning experimentation by employees can result in breaches. A centralised approach, a clearly defined AI policy, and ongoing employee training are essential to protect internal data and maintain trust.  

People 

There is a risk that people within the organisation are scared and/or reluctant to use AI tools, for fear of losing their job. Staff need to be brought along on the organisation’s AI journey to ensure the efficacy of the tools is maximised. This should be sponsored from the top of the organisation.  

Data 

Data is the bedrock of AI tools, and without good, clean, structured data AI tools which search internal databases can’t learn, adapt or perform. It is therefore critical for organisations using these types of tools to first ensure their data is appropriately formatted, structured and stored.  

Importantly, successful AI adoption isn’t just about the technology; it’s about people. The most effective implementations are those that empower humans to interrogate outputs, validate insights, and make better decisions, faster. 

Where to Start 

For small to mid-sized firms, the optimal strategy is often a hybrid approach: 

• Use secure enterprise tools like Copilot to increase productivity across the business 

• Explore specialised platforms for high-value research tasks 

• Leverage existing investments in platforms like Aladdin or BloombergGPT 

• Prioritise data readiness, cultural engagement, and internal policies to support sustainable AI adoption 

Unlike larger firms with the capital to build and maintain proprietary models, most firms don’t need to create their own LLMs to see value. This approach allows them to move forward without building complex infrastructure from scratch while still gaining real, measurable benefits from AI.  

Looking Ahead 

Most organisations focus on the technology and overlook the human investment – the cultural shift, education and training. The most successful AI implementations are those that are augmented with humans to deliver efficiencies, validate outputs and facilitate informed decision making.  

Get in touch to learn how Liqueo can help you to integrate AI across your organisation in a safe, scalable and effective way.