GAEA Talks

Graeme Scott sits down with Rowan Curran, Principal Analyst at Forrester, who leads the firm's research on generative AI technologies, tools and strategy, and covers AI platforms, cognitive search and synthetic data. He pioneered Forrester's predictive analytics research in 2015, left for a spell in the public sector because he wanted to sit with problems rather than move on to the next client, and returned in 2022 covering both cognitive search and AI platforms. ChatGPT launched weeks later, putting him at the exact intersection of the technologies that mattered. He calls it the smartest accidental thing his brain has ever done.The first half is the most useful account of agentic AI testing you will hear, and his assessment is unsparing. He describes testing as one of the most horrifyingly limited areas of the entire agentic space. Model evaluations and leaderboard scores are directionally useful and make good marketing, but they do not translate into enterprise outcomes. Most organisations have no evaluation set for the task they are automating, so they cannot say whether the system works. Manual testing does not scale to an enterprise full of agents. Using a model to judge another model means first building trust in the judge, which he describes as turtles all the way down.He has been asking agentic platform vendors about their testing suites for two and a half years. Through the end of 2024 and most of 2025 the answer was that it was a next quarter problem. Only in the first half of 2026 has he seen anything he would call robust.The second half widens considerably. He argues the existential risk framing obscures what these mostly are, which is software and cybersecurity risks. Humans will not be destroyed by a tool. Humans will do it with one, negligently or deliberately. He is equally clear that any regulatory framework has to be designed so it does not hand control to the model companies, and draws the parallel with letting social media platforms set their own age limits.He is also sharp on what the field has forgotten. The algorithmic bias work of the late 2010s, facial recognition failing on darker skin tones, the accessibility opportunities in language and vision models, has largely dropped out of the conversation, and that loss shows up in what gets built.Then the line that gives the episode its sting. The marketing goal of creating an audience of one is an extremely lonely and sad place for every human to be in. If your content and mine share no source, what are we going to talk about? That leads into an exchange on music, memory and what a shared cultural soundtrack was for, including what happened when Graeme generated northern Iraqi folk music and played it to people who grew up with the real thing.

Episode four of our GAEA Talks Forrester Season, recorded in Austin, Texas.Graeme Scott sits down with Faram Medhora, Principal Analyst at Forrester, who covers enterprise business applications, ERP modernisation, SaaS governance and the mission critical functions around finance and operations. He describes his job simply: making sure technology leaders do not make multimillion dollar mistakes, and these days billion dollar ones.Faram has spent nearly two decades leading business technology initiatives, including eight years with Forrester Consulting before moving into research. He is unusually well placed to say what AI is genuinely doing inside large organisations, because he is in the room when the bill arrives.His diagnosis of the last two years is direct. The number one mistake technology leaders made was investing in the technology first and working out what to do with it afterwards. Everything is now labelled AI, which makes it hard to tell what any given capability actually is until it meets the rest of the ecosystem and the problems surface. Risk gets treated as a governance conversation for later. As he puts it, organisations understand the risks, but they do not understand the implications of the risks.The argument at the centre of the episode is one every board should hear. Even if every individual agent in an enterprise is successful, the company can still be failing. He gives a worked example. A sales agent is told to drive sales, so it discounts. An operations agent is told to close deals, so it adds services to handle fulfilment. Both hit their objectives. Margin collapses. Nobody defined the company's actual goal between them.Then there is the off switch, which is the part that should genuinely concern people. Imagine seventy percent of a company's work has moved to agents, and something goes wrong that means you can no longer trust them. You cannot simply switch them off, because the human capacity to absorb that work is gone. The operating model has already shifted.He is also precise about where value actually is. Most of what enterprises are getting today is efficiency, which is cost avoidance, and cost avoidance does not go back on the books. The move from efficiency to effectiveness is where the real argument sits, and most organisations are stuck in proof of concept because they can see the risk of scaling.

Episode three of our GAEA Talks Forrester Season.Graeme Scott sits down with Mark Moccia, VP, Research Director at Forrester, whose team produces the research and daily client guidance for CIOs and their leadership teams across technology strategy, enterprise architecture and IT financial management.Mark came to Forrester two years ago from the other side of the table, after many years as a technology executive inside a Fortune 100 company. In his last role he owned around three hundred applications, a portfolio forty years deep, running everything from mainframe jobs to desktop clients to modern cloud services.His opening position sets the tone. A prominent researcher has just put the odds of AI wiping out humanity above ten percent, and Mark still cannot get speech to text to work reliably on his phone. The truth is neither end of that spectrum, and on the evidence Forrester is seeing from clients, closer to the early stage than the narrative suggests. Standing up one agent is straightforward. Standing up a thousand that share context, talk to each other and solve problems autonomously and accurately is a different proposition, and he is not seeing much evidence of it yet.He is precise about why. Not legacy systems exactly, but data scattered across hundreds of stores where different humans know where it lives and what it means. He gives the example of the word premium, which means one thing on an individual auto policy and something else entirely on a global cyber risk policy. That kind of context is human by default, and the effort to teach it at scale is where reality meets the hype.Then there is the part he calls the boring truth about AI. The organisations Forrester expects to win are doing the unglamorous work: data cleanliness and governance, talent and upskilling, culture, change management, and financial discipline. He calls these no regrets investments, because they pay off whether or not AI goes the way of the Segway.The last section is the one to stay for. Mark's strongest advice to technology leaders is not about technology at all.

Episode two of our Forrester Technology and Innovation Forum series, recorded live in Austin, Texas.Graeme Scott sits down with Brian Hopkins, VP of the Emerging Tech Portfolio at Forrester, for the most direct conversation we have had on what AGI actually means, what the compute buildout has to earn back, and why the answer to both is less dramatic and far more useful than the headlines suggest.Brian came into analysis from industry, working in financial services and defence across IT and architecture. About fifteen years ago he became the first analyst at Forrester to use the term big data in research, at a point when nobody knew what it meant. He has spent the last four years applying that data and analytics grounding to AI.The question that opens the episode is the one his colleague Mike Gualtieri put at the centre of their research. If you had shown today's AI capabilities to an AI researcher ten years ago and asked whether this was AGI, what would they have said? Yes. Every time we get close to something we would once have called AGI, we move the goalpost further out. So Forrester stopped moving it. Their report, The Quiet Roar of Artificial General Intelligence, argues AGI arrives in stages rather than as a bolt of lightning, and that the first of those stages, competent AGI, is what we are already using. It needs a lot of supervision, it makes mistakes, it works in a limited domain over days rather than months, and it learns what it needs to know, asks when it is unsure, and writes its own tools to solve problems. The open question is not superintelligence. It is when the next stage arrives.The second half is about money, and it is the part enterprise leaders should watch twice. Forrester has been modelling the AI bubble question by comparing the compute buildout against non-farm worker productivity, the one measure that reliably shows economic value. Even on generous productivity assumptions, the numbers do not close. What is left over Brian calls dark matter, or latent value, and it has to come from engagement change and transformational change rather than efficiency. His conclusion is a challenge to his own industry. Stop selling enterprises frontier models, and start giving them a reason to trust the basics.

This is the first episode in our Forrester Technology and Innovation Forum series, recorded live in Austin, Texas. Over the coming months we will be sitting down with Forrester analysts and their clients at the Forums in Austin, London and New York.Graeme Scott opens the series with Alla Valente, Principal Analyst on Forrester's security and risk team, who covers governance, risk and compliance, third party risk, contract lifecycle management and, increasingly, all of the above as they collide with AI.Alla is asked the question every board is asking, which is whether AI is simply the next emerging technology. Her answer is no, and the reason is specific. With SaaS and cloud, organisations chose whether to adopt, when, where in the business and who got access. That choice no longer exists. AI is already inside your organisation whether you have a strategy for it or not. As she puts it, not having an AI strategy is a strategy. It is just not a very good one.From there she draws a distinction most organisations have not yet made. AI governance is a function. It is your policy, your charter, your process for deciding which use cases are acceptable. Governing AI is the execution of that, and it happens through compliance, risk management, security and responsible AI. Organisations reached for governance first because enterprise risk management was not mature enough to move at the speed AI demanded, and the gap between the document and the delivery is where the exposure sits.She is equally direct on third party risk. Almost nobody is building their own models. You are buying foundation models, buying data, using open source, which is also a third party. AI arrives through the ecosystem, and third party risk management in most organisations is deprioritised, federated and underfunded compared with enterprise risk.The section enterprise leaders should sit up for is contracts and concentration. Organisations are using AI to contract faster, but they have not contracted for AI. Most contracts still say nothing about model training, data access, who is responsible when there is an incident, who fixes it and what the recourse is. Her line on this is the sharpest in the episode: contracts are your AI guardrails that have teeth, and they are the only ones that do. Alongside it sits concentration risk. If your business runs on one provider's model and something makes that model unusable, how long is the disruption and how much can you absorb? Map it before the crisis, not after.