Bringing AI in-house: Why investment firms need AI prompt fluency and not just tools

In recent months, generative artificial intelligence (GenAI) has moved from novelty to necessity in the investment industry, while the pace of adoption among industry members is accelerating. As competitors embed AI into research, client reporting and communications, the question for many investment firms is urgent: how can they harness these capabilities without losing the clarity, credibility and competitive edge that define their voice?

Global asset management firms are investing heavily in AI, yet many still outsource one of their most strategic assets: their voice.

Content is not just communication; it is part of a firm’s capital and credibility. With increasing pressure to publish more, faster and with greater precision, many companies are wondering if AI-enabled content creation can gain further momentum. The rise of GenAI tools, such as ChatGPT, Gemini, Claude, Perplexity and proprietary in-house models is encouraging investment teams to rethink how content is produced. Many now ask whether more can be done internally, provided the right structure and skills are in place.

Some firms are already leading the way. Goldman Sachs has explored AI for code generation and analysis. JPMorgan is piloting large language models (LLMs) to support equity research and client communication. BlackRock continues to expand its Aladdin platform, embedding AI across portfolio modelling, risk analysis and investment operations.

What these firms recognise is that tools alone are not enough. The real competitive advantage lies in how they are applied, and at the centre of that shift is the ability to prompt effectively.

Why AI prompting matters

Research from the World Economic Forum shows that over a third of financial services work could be automated, and another third enhanced, through AI. Global spending on AI in the sector is expected to nearly triple from $35 billion in 2023 to $97 billion by 2027.

Yet, despite this momentum, a Thomson Reuters survey of more than 2,000 professionals found that only 22% of organisations have a clearly defined AI strategy. Those with a defined strategy are nearly twice as likely to report revenue growth and more than three times as likely to realise critical benefits.

The technology is advancing quickly, but effective application remains uneven. AI prompt fluency, or the skill of guiding AI towards relevant, accurate and well-structured output, is becoming as essential as market knowledge itself.

What prompt fluency looks Like

In investment workflows, prompt fluency supports a range of outputs. These include macroeconomic updates, fund commentaries, ESG insights, sector briefings and educational content.

Useful output requires more than selecting a topic. A strong prompt must define the audience, the objective, the tone and the structure. There is a clear difference between asking AI to "write about inflation" and asking it to "draft a 300-word inflation outlook for high-net-worth clients, focusing on portfolio positioning, using a cautiously optimistic tone consistent with our quarterly strategy."

Prompting is not a mechanical step. It is a strategic input that determines the clarity and quality of what AI produces. Those who approach prompting with structure and intent are already seeing gains in both efficiency and relevance.

Where teams often struggle

As firms adopt AI tools, common problems quickly emerge. Vague prompts tend to produce generic or shallow content. Many outputs miss the intended tone or audience. Others lack structure or consistency. These are not necessarily failings of the technology. They reflect the absence of clear instruction.

There are also challenges related to data. According to MX’s 2025 Risks and Rewards of AI in Banking report, 77% of financial organisations face data quality issues, and 91% say these  affect performance. In the context of content creation, poor data inputs, such as outdated macroeconomic data or inconsistent earnings figures, can significantly undermine AI-generated output.

AI hallucinations, where models generate plausible-sounding but false information, remain a concern. Bias in training data also poses risks, especially in areas like product recommendations, credit analysis or financial guidance. These issues are particularly serious in regulated environments like financial services.

Prompting alone does not eliminate these risks, but it can help mitigate them. Clear prompts guide the model towards more relevant content, reduce overreach and create space for human oversight. This is especially valuable for client-facing material, research summaries or any content carrying reputational or compliance sensitivity.

Towards a more structured approach

Prompting is a skill that improves with practice. When treated as a structured task rather than a casual command, it enables better control over tone, message and accuracy.

In our experience, well-constructed prompts, used across everything from internal briefings to portfolio commentary, consistently deliver clearer, faster and more consistent output. They reduce editing cycles and help teams maintain a coherent voice. Building this fluency in-house allows firms to retain ownership of their message while improving both quality and speed.

For now, the takeaway is clear: prompt fluency is no longer optional. It is a core capability that blends domain insight with editorial structure, essential for producing timely, credible content in today’s fast-moving environment.

EquityEdge Studio is ready to work with investment teams to refine their prompting approach to content creation.