Conversation is how humans actually think and work
In Dublin, a publicly funded AI research centre and one of the world's largest asset managers have joined forces to teach machines the language of finance. The Adapt centre, backed by Science Foundation Ireland, is working with Fidelity Investments to build conversational AI that can speak directly to portfolio managers and analysts in real time — not as a novelty, but as a practical tool for navigating the relentless flow of market data. This partnership, unfolding across multiple years and phases, reflects a broader pattern: global financial institutions increasingly looking to academic research ecosystems for the kind of deep, specialised intelligence that commercial development alone cannot easily produce.
- Asset managers and analysts are drowning in real-time data, and the pressure to act faster and smarter has made the gap between human attention and machine output a genuine business problem.
- Fidelity is not waiting for off-the-shelf solutions — it has embedded itself into Ireland's publicly funded AI research infrastructure, treating academic centres as innovation partners rather than consultants.
- A proof-of-concept system is already in development, adding speech capabilities to Fidelity's existing technology stack — a deliberate first step designed to surface what works before committing to something larger.
- The roadmap moves from controlled experiment to real-world demonstrator, stress-tested against actual trading floor conversations and analyst workflows.
- The long-term target is architectural: a scalable, sustainable conversational AI system capable of growing with Fidelity's equity investment operations rather than becoming obsolete.
- A parallel partnership with SFI's Insight centre on knowledge graphs and computer vision signals that Fidelity is assembling a mosaic of Irish AI expertise, not placing a single bet.
Researchers at Adapt, a Science Foundation Ireland centre specialising in AI and digital content, have launched a multi-year collaboration with Fidelity Investments to build conversational AI systems tailored to the rhythms of asset management. The goal is practical: give portfolio managers and quantitative analysts a system that can listen, respond, and deliver real-time portfolio updates in natural language — helping them move faster without losing clarity.
Adapt brings expertise in dialogue systems, human-computer interaction, and personalisation. Fidelity brings the problem — the constant, data-heavy back-and-forth between traders, analysts, and their information streams. The two organisations have already moved past introductions, mapping out requirements and sketching a development roadmap together.
The project runs in three stages. A proof-of-concept is already underway, grafting speech capabilities onto Fidelity's existing infrastructure to test what the technology can do in a controlled setting. From there, a more rigorous demonstrator will be built around real business scenarios — conversations that actually happen in practice, not just in theory. The final phase looks further ahead: designing the architecture for large-scale, sustainable deployment across Fidelity's equity investment operations.
Fidelity's director of research, development and innovation highlighted Adapt's particular strength in intent recognition — the ability to understand what a user actually means, not just what they say. Adapt's director framed the partnership as a chance to solve genuine fintech problems while advancing the research frontier simultaneously.
This is not Fidelity's first engagement with Irish AI research. The Insight centre recently announced its own Fidelity-backed projects spanning knowledge graphs, natural language inference, and computer vision, with researchers at NUI Galway and Dublin City University involved. A clear pattern is emerging: Fidelity is deliberately weaving Ireland's publicly funded research ecosystem into its own innovation pipeline.
In Dublin, researchers at Adapt—a Science Foundation Ireland centre focused on AI and digital content—have begun a multi-year partnership with Fidelity Investments to build conversational AI systems that speak the language of portfolio management. The work is straightforward in its ambition: create technology that can talk to asset managers and quantitative analysts in real time, feeding them updates about their portfolios and helping them work faster and smarter.
Adapt brings together academic leaders and postdoctoral researchers who specialize in dialogue systems, human-computer interaction, conversational AI, and personalization. Fidelity brings the real-world problem: how to automate and streamline the constant back-and-forth between traders, analysts, and their data. The collaboration has already moved beyond the handshake stage. The two organizations have already mapped out what the technology needs to do and sketched a roadmap for how to build it.
The project unfolds in three phases. First comes a proof-of-concept system, already under development, that will use Fidelity's existing technology infrastructure to add speech capabilities—the ability to listen and respond. This is the test kitchen, the place to learn what works and what doesn't before investing in something bigger. Vinny Wade, the director of Adapt, framed it as an opportunity for his team's data and technology experts to tackle real fintech problems and advance the state of the art at the same time.
In the medium term, the researchers will build a more substantial demonstrator, grounded in actual business scenarios and use cases from Fidelity's operations. This version will be tested more rigorously, pushed harder, asked to handle the kinds of conversations that actually happen on a trading floor or in an analyst's office. The goal is to move from "does it work in theory" to "does it work in practice."
Looking further ahead, Adapt will help Fidelity design the architecture needed to deploy conversational AI at scale across its equity investment operations. This is the long game—not a one-off tool, but a sustainable system that can grow and adapt as the business changes. Fintan O'Malley, Fidelity's director of research, development and innovation, emphasized that the partnership taps into Adapt's particular expertise in conversational assistants and intent recognition, the ability to understand what a user actually wants beneath the words they use.
This collaboration is not Fidelity's first dance with Irish AI research. Last month, Insight, another SFI research centre focused on data analytics, announced its own partnership with Fidelity on knowledge graphs, natural language inference, and computer vision. Insight researchers at NUI Galway and Dublin City University are running three Fidelity-funded projects, and Fidelity is also backing two SFI Industry Fellowships involving Insight researchers at Galway. The pattern is clear: Fidelity is building relationships with Ireland's publicly funded AI research infrastructure, treating it as a source of innovation and expertise that can be woven into its own product development.
Notable Quotes
Our collaborations with Fidelity's asset management division offer an ideal opportunity for our data and technology experts to push the frontiers of knowledge for Fidelity and bring new solutions to concrete fintech challenges.— Vinny Wade, director of Adapt
It is through tapping into Adapt's expertise in conversational AI and related themes including conversational assistants and intent recognition that we can push the boundaries of the technology options available for the automation of our equity investment process.— Fintan O'Malley, Fidelity director of research, development and innovation
The Hearth Conversation Another angle on the story
Why does a financial services company need to talk to its own software? Why not just build dashboards and reports?
Because conversation is how humans actually think and work. A portfolio manager doesn't want to hunt through screens for information—they want to ask a question and get an answer, the way they'd ask a colleague. It's faster, it's more natural, and it frees them to focus on the decision-making part, not the information-gathering part.
So this is really about speed and efficiency in the trading operation?
Partly, yes. But it's also about reducing errors. When you're talking to a system that understands what you're asking, you're less likely to misinterpret data or miss something important. The AI can also surface relevant information you didn't even know to ask for.
Why partner with Irish researchers specifically? Couldn't Fidelity build this in-house?
They could, but they're choosing not to. There's real expertise in dialogue systems and conversational AI at Adapt and Insight. By partnering, Fidelity gets access to cutting-edge research and academic talent without having to build and maintain that capability themselves. It's also faster—these researchers have been thinking about these problems for years.
What's the risk here? What could go wrong?
Conversational AI is still imperfect. It can misunderstand intent, give confident-sounding wrong answers, or struggle with edge cases. In finance, those mistakes can be costly. That's why they're moving slowly—proof-of-concept first, then a more thorough demonstrator, then scaling. They're being careful.
And if this works, what happens next?
If it works at Fidelity, you'll see other financial firms wanting the same thing. Conversational AI in finance becomes less of a research project and more of an industry standard. The researchers move on to the next hard problem.