London Tech Week (LTW) returned with its usual buzz and bravado to the Olympia exhibition centre from 8-10 June, as the capital once again transformed into a hive of innovation and ambition where AI hype, startup hustle and regulatory muscle collide. Infinite’s tech specialists Joshua Allsopp, Charlotte McCluskey and Tal Donahue were on the ground speaking with exhibitors, founders, investors and the press to see what’s hot on London’s tech scene right now. Here are their top takeaways: Best of British As ever there was a distinctly global flavour to the proceedings. Türkiye, for example, had a significant presence leading the largest delegation in the festival’s history, while the likes of Pakistan, Malta and the Philippines proved they were primed to play in the big leagues. That said, this year more than ever, the conference was very much a showcase of the best of British. UK Prime Minister Keir Starmer kicked off the event on the main stage. His opening keynote address positioned AI as a transformative force Britain must actively shape, backing innovation, investment and talent while ensuring the benefits are felt nationwide – not confined to the capital – and across communities. On Tuesday afternoon, Liz Kendall, Secretary of State for Science, Innovation and Technology, took to the same stage and outlined what she called the “Great British success story”. They both insisted that Britain was uniquely placed to front the global tech revolution, noting that the UK had raised close to half of all European tech investment so far this year, and highlighted recent initiatives to nurture homegrown AI talent and businesses. “Britain’s best days lie ahead,” assured Kendall. Starmer drama Starmer outlined an approach that combined growth with responsibility, emphasising public investment, skills, and safety. But it was this last element of his speech – online safety- that caused the biggest stir, with audible murmurings both in and out of the mainstage audience. The first headlines to land from LTW on Monday morning focused on what the press characterised as Starmer’s “warning” to big tech firms, calling them out for not doing more to stop children from accessing explicit images online. While the UK government appeared anxious to project a pro-innovation, pro-growth stance, the heavy hand of the regulator cast a long and unavoidable shadow right from day one. Kendall, who appeared to have a packed media schedule ahead of her Tuesday afternoon address, had spent the morning reiterating Starmer’s warning and said a ban could come “sooner than summer”. When the chips are down... The UK isn’t usually the first place that comes to mind when you think of chips (unless served with fish). But with big ambitions for AI as a new driver of the economy, AI minister Kanishka Narayan told Guardian Tech Editor Dan Milmo on Monday’s panel session that the UK needed to have “enough chips on the table” to be able to secure its national interests in negotiations with global partners. To this end, Starmer announced a £400 million investment plan to purchase specialist AI chips, ensuring the UK owns the infrastructure and capacity needed to power its AI ambitions. In fact, this push for sovereign AI capability was a defining thread across the conference. As one computing trade magazine put it, sovereign AI was moving centre stage. Conversations were also spanning the entire AI stack, from data pipelines and governance right through to the hardware, with Kendall announcing on Tuesday a £1.1 billion UK AI hardware initiative, ensuring the UK can participate in this trillion-dollar global industry. Investors – and reporters – were now seemingly more interested in what’s under the hood, and seemingly less impressed by flashy models and interfaces. AI and the law - Lammy’s double whammy In conversation with Murad Ahmed, Technology News Editor at the Financial Times, Deputy Prime Minister David Lammy set out a cautious vision for AI in the UK justice system on Tuesday morning’s panel, as it faces intense scrutiny following the Henry Nowak murder. Lammy highlighted two urgent challenges: severe court backlogs and overstretched probation services, that a new raft of AI innovations is expected to address. AI is being trialled to ease these administrative burdens and improve efficiency, to augment and not replace human judgement. He stressed the need for transparency, robust guardrails, and bias mitigation, with public trust hinging on maintaining human oversight while responsibly harnessing AI’s economic and operational potential. For law firms and the wider legal sector navigating the accelerated adoption of AI for productivity, inevitably they must also be prepared for increased scrutiny on ethics, bias, and accountability in how this tech is deployed. LTW’s legaltech lovefest The AI Growth Labs launch was a huge boost for the UK legal technology sector, and the legaltech exhibitors we spoke to at LTW were certainly feeling energised. So, no wonder it was standing room only to hear Winston Weinberg, CEO of legaltech disrupter Harvey talk about the sector’s recent and rapid transformation on Wednesday, and his own wild journey from idea to $11bn valuation (at the last count, the pace of growth is phenomenal). He stressed that intelligence was the differentiator. Professional services generally fall into two buckets: getting stuff done or giving advice. AI plays heavily in the former category, while the value of the latter ratchets up when you layer on actionable intelligence. Weinberg also issued a stark but valuable reminder to leadership teams. If you find yourself discussing the same problem twice, you’re likely stagnating. The thirty-something former lawyer said he didn’t spend any of his precious time looking at things that worked and instead focused his energy on what needed to be fixed. He said that rapid and relentless problem solving is what’s needed for the exponential pace of an AI-powered world. Amidst all this excitement, meanwhile, was Attifin AI, which is focused on practical and more nuanced AI delivery. It deserves a mention as it’s looking to give its customer base an AI solution that’s validated by actual UK jurisprudence, rather than global platforms which often default to a US position. Definitely one to watch. Pay-as-you-go legal services – gap for SMEs and startups At the gates of ‘Startup Alley’ was another legaltech worth a mention. Sprintlaw provides fixed-fee pay-as-you-go legal services for UK small businesses. Its presence felt telling. Amid all the talk of scale, disruption and frontier tech, there was also clear demand for practical services that help smaller businesses get started, stay compliant and move quickly without the cost or complexity of a traditional law firm. In that sense, Sprintlaw captured a subtle but important theme running through the week: innovation is not just about building new technology but making essential services more accessible to the businesses trying to grow. Agents – built-in kill switches and risk mitigation Agentic AI may have been this year’s buzzword, but for us the more interesting conversation was what these platforms mean when viewed through a risk and crisis PR lens. As businesses move closer to deploying tools that can act, decide and respond with greater autonomy, the risk conversation is shifting with them. It is no longer just about whether the technology works, but how organisations prepare for the reputational, operational and communications fallout when it misfires. From unclear accountability to the speed at which errors can escalate, agentic platforms are creating a new category of crisis risk, one that many businesses still seem to be getting to grips with. AI is changing the workplace in ways didn’t expect AI is revolutionising how we work, there’s no doubt about that. But the driving force, it seems, is often from the ground up. Darren Hardman, Corporate Vice President & CEO of Microsoft UK and Ireland, noted on Monday that employees are already bringing AI into their roles, making AI skills a necessity rather than a differentiator, with fluency spreading organically between teams. He gave the example of law firm Shoosmiths where an executive assistant without a technical background built an AI agent by understanding day-to-day workflow friction. However, adoption remains uneven. Alison Kay, VP & Managing Director at AWS UK and Ireland, highlighted that while basic AI use is rising, progress is stalling at more advanced levels. Around half of UK organisations remain at entry level, and only 24% are deploying advanced systems. The core issue is capability: skills shortages are the single biggest barrier, preventing businesses from deepening adoption and fully unlocking AI-driven productivity and competitive advantage. There were a few passing (awkward) allusions to rising unemployment too, particular among young people. Ivan Zhao, CEO and founder of productivity software company Notion, offered an interesting take. He said that while we were all preoccupied with AI agents we should be looking out for agency when hiring new (human) talent – the ability to think independently. He said that young people have high agency, as well as more energy, taste and higher levels of optimism, which more than make up for their lack of experience or capability. How to stop being creepy A recurring theme on one of Wednesday’s Transformation stage panel was simple but important: how can brands stop being so creepy? We’ve all had that unexpected chatbot pop up within five seconds of browsing a company webpage – or worse, the unsolicited WhatsApp message from a salesperson who’s been watching from the shadows. In a world where 84% of people expect personalised brand messaging, the pressure to tailor communications is real, but so is the risk of seriously crossing the line. Personalisation can both strengthen and fragment the brand experience. The challenge is finding the balance by staying relevant without becoming intrusive. Brands need to be judicious with their use of data, mindful of context, and aware that the right level of personalisation depends heavily on the channel being used. Crucially, brands must earn the relationship and the right to personalise by connecting first before leveraging data. Personalisation should also stay rooted in the brand profile and focus only on the data points that genuinely matter to the individual. Shared cultural moments are dead. Long live newsjacking! The era of the big, shared cultural brand moment may be fading. Hyper-personalisation, fragmented media consumption, rapid news cycles and the sheer number of channels mean we are losing the kind of mass cultural moments once typified by things like the John Lewis Christmas ads. Or at least, their impact is no longer what it once was. These moments now feel like remnants of legacy marketing strategies that may not be realistic or even economically viable in the age of AI ads and hyper-personalisation. But the underlying principle of cultural relevance still matters. Instead, the key lies in the myriad micro-moments: fleeting, reactive, channel-specific opportunities for brands to show up in a way that feels timely and useful. Newsjacking is alive because brands can still grow through cultural connection, but only when they are genuinely additive, rather than merely piggybacking. In the age of GEO brands, have brands lost power over how their story is told? Not necessarily, according to some of the panellists. While GEO and AI discoverability may seem to shift control away from brands, the reality is more reassuring: these systems are still driven by human-centric decisions and principles. In that sense, GEO is not a break from brand building so much as a return to the fundamentals. It is about ensuring consistency across channels and experiences, so that the information surrounding a brand is coherent, credible and easy to understand — whether for people or machines. Brands are still very much in control of this. They remain in control of the content that shapes their discoverability, the signals they send, and the consistency of the story they tell across every touchpoint. In an AI-shaped landscape, discoverability is still earned through the basics: clarity, consistency and relevance. Uncertainty is a competitive edge As Dr Rebecca Swift of Getty Images put it on Wednesday’s panel on tapping into culture, “there is something wonderful in that lack of control.” In an age where audiences can spot inauthenticity instantly, one of the biggest risks for brands is trying too hard to engineer every outcome. Authenticity often comes from relinquishing control to some degree. The most effective approach is to provide the guardrails, build the world, and then let creators do their thing within those parameters. That uncertainty can become a competitive edge. It opens the door to work that feels more human, more original and more resonant with the audiences a brand actually wants to reach. After all it is the creators who understand how to connect with culture and their audience best. Word of mouth vs word of machine Brand strategy now needs to operate on two pillars: word of mouth and word of machine. Winning brands are no longer just those that are emotionally resonant or culturally visible; they must also be machine-readable and human-referenceable. That means thinking not only in terms of B2B and B2C, but increasingly B2A, with agents becoming a brand’s primary audience in some contexts. As people trust AI to do more and more, and increasingly trust agents to act on their behalf, brands must adapt to this new decision-making environment. As Dentsu UK CEO Jessica Tamsedge put it, it is “brilliant but terrifying.” But this is not the death of the brand. AI may be shaping consumer behaviour, but it is still trained on human preferences and experiences. The real challenge is not choosing between machine and human experiences, but building for both: AI visibility and human credibility. “The CMO is an archivist as much as a visionary” One of the sharpest observations from Tamsedge’s keynote on Wednesday was that in the age of AI, the CMO is an archivist as much as a visionary. Brands now need to create consistency across space and time, ensuring that the data, language and signals feeding large language models are aligned and reliable. That makes brand management as much an exercise in disciplined data stewardship as storytelling. What a brand says today must connect cleanly with what it said yesterday, and across channels, platforms and formats. In that sense, a creative challenge is now also a data challenge. If the information a brand puts into the world is messy, that confusion will shape how it is interpreted and surfaced by machines. The future of brand building depends on both imagination and information architecture. Earned matters more than ever As AI discoverability becomes a bigger part of brand visibility, earned media matters more than ever. AI visibility on its own means very little without human reference-ability — which is good news for the PR world! Brands must create the kinds of conversations that machines want to listen in on, but that does not mean all mentions are equal. Machines are able to assess for authority, credibility and reliability with increasing accuracy, which means trusted voices matter more, not less. The implication is clear: the best brand signals are not just self-published content, but credible third-party validation. In this environment, earned media is no longer simply about awareness; it is about building the authority layer that both people and machines rely on. PR therefore becomes central to discoverability, because it helps brands become both visible and believable.