As with any investment, your capital is at risk.
The starting point
In December 2022, when ChatGPT burst onto our browsers, we were in a strong position to make the most of AI in asset management. We had invested in companies such as Amazon since 2004, followed the journeys of Chinese tech giants from 2016, and bought our first stake in NVIDIA that same year. Our proximity to emerging AI leaders meant we had a rich store of knowledge in our research library to draw on, but the realities of adopting AI across the business, and beginning to transform ourselves into an AI-first organisation, were much harder than simply researching the technology. As a 100-year-old business in a regulated industry, keeping pace became the medium-term goal.
So how do we keep up in such a fast-moving race? We have approached it from five different angles.
Strategy
AI has been at the centre of our organisational strategy for three years. This top-down blessing has given momentum to AI initiatives and lent support to colleagues who want to adopt AI in their own areas. It has also helped signal that AI is not a side project, or the preserve of a small group of technologists, but a firmwide priority.
We purposely embraced a strategy that allows for individual experimentation because we have a decentralised, bottom-up culture. If we had chosen a different strategy, one that was highly hierarchical or purely data-driven, it may not have meshed with our investment approach or company ethos. Instead, we have tried to follow individual enthusiasm, trusting that colleagues know their own use cases best. Three real examples are:
- One investment analyst used AI to prepare for 15 company meetings on a trip to San Francisco, compressing the time required for background work while improving the quality of each conversation.
- The Platform Engineering Team created a new project analysis tool that transforms a project idea into a full list of tasks, timelines, estimated work and number of sprints for the developers to implement.
- Back office colleagues have designed an automation that generates batches of files for the fund accountants every day, which are processed with another automation, freeing up thousands of pounds a year in staff time.
However, decentralised experimentation still needs accountability. We ask colleagues to consider AI-related goals that are recorded in our HR system, and some teams have chosen to mandate AI objectives. Role-specific AI goals are one of the key tools we have to ensure colleagues take responsibility for staying up to date with AI applications. The aim is not to make everyone a developer, but to make everyone sufficiently AI literate to understand how these tools may change their own work.
Governance
In order to keep pace with AI development, we revisited our governance frameworks to ensure they could cope with emerging technologies. After review, we decided to apply established governance, risk and control processes to the use of AI and other emerging technologies. Relevant subject matter experts, control functions and governance forums provide oversight to ensure that technology use remains aligned with our risk appetite, regulatory obligations and security requirements. Our approach will keep evolving as the technology and associated risks develop.
Before deploying new AI tools, we also revisited our contracts. We use a lot of third-party data and software vendors, and we needed to understand how all our commercial agreements could cope with AI. In some cases, we needed to renegotiate commercial terms, and we spent a lot of time with model providers making sure their agreements were appropriate for a financial services firm operating in a regulated environment.
A key feature of our governance for now is that all our AI decisions remain human-in-the-loop. This allows us to stay safe and keep pace because it helps us understand the steps AI goes through, and learn more about the tools, before moving towards more agentic systems. In investment research, this principle is especially important: AI can help structure a reassessment of an investment case during a drawdown, but the judgement on whether the original thesis still holds must remain with the human investor.
Investment
Baillie Gifford’s independent partnership allows us to make long-term investments counter cyclically. We have given all staff access to large language models, and have bought enterprise licences in additional cutting-edge tools, which can be used on request.
Trialling different products and services allows us to monitor changes in this fast-moving area rather than becoming tied to a single provider too early.
We also built our own internal AI platform, Sidekick+, which includes tools for translation, meeting transcription and a confidential data mode. This matters because, at least initially, regulated firms could not simply adopt consumer tools wholesale. We needed to give colleagues access to AI in ways that were useful, secure and aligned with our data governance expectations.
We have invested in data infrastructure because, like many businesses with a long history, our data is not in perfect shape. It is stored in different programmes and databases, some of them offline, and metadata has not always been consistently ordered or tagged. An early attempt to build a CRM querying system taught us important lessons about data architecture and data cleanliness, and those lessons are now informing more ambitious infrastructure projects.
We have invested in people too. Our R&D Department has grown from 14 to 32 in two years. Their job is not just to follow the frontier of AI-led technology development, but to translate that knowledge into tools that our people can use every day in their work.
Training
Keeping pace requires constant education. We have tried to train our workforce in AI at all levels, but in different ways. At the partnership level, we are members of a network of AI leaders led by an academic contact. This helps us learn from businesses that are further on than ourselves and test whether our assumptions are keeping pace with practice elsewhere.
We harness internal expertise as much as we can because not all colleagues are at the same level with AI. We have established a network of 300 AI Champions that spans every department, so peers can learn from each other through formal sessions, show and tells, and peer-created resources. This is important because AI adoption often moves fastest when colleagues can see someone in a similar role using the tools in a practical, relevant way.
To ensure everyone is capable of using AI tools to some degree, we rolled out mandatory AI training across our entire staff base. The expectation is clear: everyone at Baillie Gifford should become AI literate. Training is supported by ongoing information and data security education, as well as role-specific help for teams whose work may be changed more quickly by AI.
We have also found that training works best when it is linked to real examples. One analyst, for instance, used AI to review transcripts from expert calls, identifying gaps in questioning, flagging one-sided assumptions, and synthesising large volumes of qualitative data by theme and sentiment. Examples like this help colleagues understand that AI is not just a generic productivity tool, but something that can improve the quality of thinking when used carefully.
Experimentation
Finally, we think keeping pace requires a combination of decentralised experimentation and overarching projects that address unlocking issues, such as data architecture. Anyone in our firm can now initiate an AI change project, which will be appraised by our in-house project managers for feasibility and impact. By allowing subject matter experts to suggest AI solutions, we can understand AI’s potential at the local level.
Our goal is not cost reduction. It is to enable our people to do more high-value work, including activities that would previously have been prohibitively resource intensive, or would have imposed too great a coordination cost on the firm. We are investigating AI assistance in data platform migrations and legacy code reviews, both of which are time-consuming but important tasks within mature organisations. This kind of work rarely attracts attention from the outside, but it is essential if firms are to make their data and systems ready for more ambitious AI adoption.
Some of the most promising projects are emerging where local knowledge meets firmwide infrastructure. Our R&D team is working directly with investors on a hypothesis-tracking system, which structures forward-looking investment theses as explicit, monitorable assumptions that can be tested against incoming data in real time. This is the kind of task that would absorb prohibitive analyst time without AI, but could be valuable if it helps protect long-term insight from noise and behavioural bias.
The Clients Department is replacing its existing note-taking software with a structured, AI-native knowledge workspace. The aim is to turn years of accumulated investment and client specialist insight into a searchable, collaborative resource that any client-facing colleague can query quickly. This is not about replacing expertise; it is about making institutional knowledge easier to find, share and apply in service of clients.
We are also investing in data infrastructure to support more ambitious cross-firm programmes, with the aim that we adopt a new core data platform. While there is a great deal of focus and anxiety around power capacity in relation to AI, adoption capability within mature organisations is also constrained by heterogeneous legacy data systems. It is no coincidence that one of the tool releases associated with the so-called “SaaSpocalypse”, when the share prices of major SaaS and consultancy companies fell sharply, was an Anthropic tool aimed at accelerating the rewrite of legacy COBOL codebases. For firms like ours, modernising the data estate is not a peripheral task; it is part of the foundation for using AI well.
Our firmwide AI rollout is something we are proud of. By the end of March 2026, more than 90 percent of staff had used an AI tool in the quarter, and 70 percent were regular users. We also saw early evidence of sophisticated usage: when we had just 250 licences, the model provider ranked Baillie Gifford highly for deep research and data analysis, suggesting colleagues were using AI for more than basic prompting.
What comes next
Staying current in the AI era as a regulated business that predates the PC is less about a single breakthrough and more about sustained, coordinated evolution. Our experience shows that success lies in aligning strategy with culture, embedding accountability, and building governance frameworks that are robust yet adaptable. It requires meaningful investment not only in tools but in data foundations and, critically, in people who can translate technological potential into practical outcomes.
Equally, keeping pace depends on creating an environment where learning is continuous and experimentation is encouraged at every level of the organisation. By combining structured oversight with decentralised innovation, we can explore AI’s possibilities while maintaining the standards our clients expect. Human judgement remains central for now, but each step forward builds the confidence and understanding needed for a more automated future.
Ultimately, remaining at the cutting edge is not a destination but an ongoing process. In a field evolving as rapidly as AI, the firms that thrive will be those that stay curious, invest with conviction, and adapt without losing sight of their core principles.
Risk Factors
The views expressed should not be considered as advice or a recommendation to buy, sell or hold a particular investment. They reflect opinion and should not be taken as statements of fact nor should any reliance be placed on them when making investment decisions.
This communication was produced and approved in June 2026 and has not been updated subsequently. It represents views held at the time of writing and may not reflect current thinking.
Potential for Profit and Loss
All investment strategies have the potential for profit and loss, your or your clients’ capital may be at risk. Past performance is not a guide to future returns.
This communication contains information on investments which does not constitute independent research. Accordingly, it is not subject to the protections afforded to independent research, but is classified as advertising under Art 68 of the Financial Services Act (‘FinSA’) and Baillie Gifford and its staff may have dealt in the investments concerned.
All information is sourced from Baillie Gifford & Co and is current unless otherwise stated.
The images used in this communication are for illustrative purposes only.
198612 10063831
