Key Take Aways
-
Many organisations are worried about falling behind on AI, but most are still at an early stage of practical adoption.
-
AI adoption is highly uneven within firms: some employees are experimenting actively, while others have not yet used available tools such as Copilot.
-
The biggest issue is not always access to AI tools, but knowing what to do with them and how to apply them safely.
-
AI adoption is following a familiar technology adoption curve, but the pace is much faster than previous technology waves.
-
Senior leaders are more directly involved in AI than in many previous technology shifts because AI affects the whole business, not just IT.
-
The most valuable starting point is often mundane operational work, not high-profile or “headline” use cases.
-
Internal workflow use cases can be high value, lower risk and easier to quantify because they often save time, improve capacity and avoid direct exposure to external stakeholders.
-
Domain experts are essential to AI adoption because they understand where work happens, where inefficiencies sit and where AI can create measurable value.
-
Agents and assistants should be introduced progressively, starting with lower-risk use cases before moving towards more autonomous systems.
-
AI adoption is as much a cultural shift as a technical shift; employees need to feel involved rather than having AI imposed on them.
-
Governance is critical, but it should enable safe experimentation rather than become a reason for inaction.
-
First-mover advantage is less about owning the technology and more about identifying the right problems to solve before competitors do.
Innovation
-
Use cross-functional AI governance councils made up of domain experts and technical specialists to evaluate AI opportunities.
-
Score AI use cases by risk, value and complexity to support prioritisation and governance decisions.
-
Start with internal operational workflows where AI can reduce repetitive work and improve productivity.
-
Build knowledge assistants to make organisational knowledge easier to access for employees, new starters, clients or suppliers.
-
Use AI in compliance processes to check work against internal standards or regulatory expectations.
-
Introduce assistants before fully autonomous agents, allowing employees to understand the capability in a lower-risk way.
-
Use semi-autonomous agents with human triggering, triage and oversight before moving to greater autonomy.
-
Create sandbox, proof-of-concept and MVP environments for safe experimentation.
-
Bring frontline and operational teams into use case design, rather than leaving AI deployment solely to technical teams.
-
Avoid “over-agentifying” processes where a simpler system replacement, API or process redesign would be more appropriate.
-
Use open-source or off-the-shelf agents and models where they already solve a large part of the target task.
-
Run models within secure internal environments where organisations need to avoid sending data to external models.
Key Statistics
-
Zeigend has designed nearly 400 agents across financial services.
-
The speakers refer to their first year of building agent roadmaps for financial services.
-
A single assistant has been seen to save eight to ten hours per week.
-
The transcript references the average SME having around 120 SaaS platforms.
-
The speakers refer to organisations having had 20 to 30 years of digital innovation, adoption and transformation cycles.
-
The conversation frames the next planning horizon for AI as six months rather than three years because of the pace of change.
-
The discussion references Copilot, ChatGPT, OpenAI, Nvidia and Microsoft as part of the current AI ecosystem.
Key Discussion Points
-
Organisations are anxious about whether they are moving fast enough with AI compared with peers.
-
Many firms have deployed tools such as Copilot, but employee capability and confidence vary significantly.
-
People often underestimate AI because they have not learned how to prompt or apply it effectively in work contexts.
-
The pace of AI change risks widening the gap between early adopters and those who have not yet engaged.
-
Leaders need to understand where AI can be applied safely and effectively, but often lack enough skilled people internally.
-
Use case identification is not the main challenge; the challenge is knowing where to start and how to prioritise.
-
The most practical opportunities are often embedded in routine operational workflows rather than strategic headline use cases.
-
Operational teams can usually quantify the time and capacity benefits of AI more quickly than central technical teams.
-
Agents and assistants need to be explained in simple terms so non-technical teams can contribute to innovation.
-
Employee fears about job security, hallucinations and risk are common and need to be addressed through education and involvement.
-
Governance, security, data protection and compliance teams need enough information to make informed decisions about AI adoption.
-
Firms should start experimenting with simple, low-risk use cases rather than waiting for perfect certainty.
Description
This podcast is a discussion between Chris, Zandra Moore and Charlie Bartle from Zeigend on how organisations can get ahead with AI adoption. The conversation focuses on the practical realities of AI implementation inside firms, especially in financial services, and moves beyond hype towards operational adoption.
The central message is that many organisations are interested in AI but unsure where to begin. Senior leaders are concerned about moving too slowly, while employees vary widely in their understanding and use of tools such as Copilot and ChatGPT. The speakers argue that the gap is primarily one of education, confidence and practical application.
The podcast is particularly relevant for senior leaders because it highlights the importance of starting with lower-risk, high-value internal use cases. Rather than chasing headline-grabbing AI projects, firms should look at everyday operational workflows where AI can save time, improve throughput and free employees to do higher-value work.
A major theme is the role of agents and assistants. The speakers describe a progression from chat-based tools to assistants, then to semi-autonomous and more autonomous agents. They stress that organisations should build confidence gradually, involve domain experts and ensure governance is in place from the outset.
The broader conclusion is that AI adoption should be treated as a cultural and organisational transformation, not just a technology project. Firms that succeed will be those that educate their people, identify the right problems to solve, experiment safely, involve frontline teams and balance innovation with governance.
RO-AR.com contact list
