Tribus at Leeds Digital Festival

"Brilliant products go nowhere, not because they were badly built, but because nobody designed for the environment they were entering." Drawn from cancer screening programmes; reads as a description of tier-one contractors. Attendees leave with a roadmap built in the room.

Misaligned expectations, missed deadlines, scope creep, communication breakdowns, disagreements over delivery — treated as normal, with mediation as prevention rather than last resort. This is the vocabulary a client's legal team uses when a build sours.

"The majority of tech leaders and their teams are using AI, but few are probably showing an ROI." Where it drives efficiency in product delivery and where it *creates* inefficiency. Questions are collected in advance — worth submitting one.

"Lessons learned from the frontline of industrialised AI delivery" — why most organisations never get past isolated pilots into measurable outcomes. Case-study led, aimed at executives.

Establishing a practice from scratch inside a large, low-margin, operationally messy retail business, failures included. The closest structural analogue in the programme to a large construction client.

Resilience reframed as a growth enabler rather than a compliance chore — what lets a firm "launch new propositions with confidence." Useful framing for selling robustness to clients who see it as cost.

Home-improvement brands as the test case, because buyers research for months and involve multiple decision-makers — the closest consumer analogue to a considered B2B purchase. Real client-side marketing leaders on the panel, not agency people.

Backed by their own Checked In report. Travel as the stress-test for a journey now spread across search, social, AI tools, influencers, reviews and word of mouth. Ends on what brand leaders need to do in 2026.

Original consumer research on a genuine paradox: customers need to use technology to trust it, but need to trust it before they'll use it. Transparency as commercial differentiator. Christina Miller (Enjoy) and Will Tisdall (Titan Wealth).

Framed around the questions clients keep asking: in-house versus agency versus freelancer, more impact on a shrinking budget, specialists versus integrated, and "can AI help — if so where do I start?"

Diagnoses what they call "the graveyard of the middle" — businesses with no clarity on audience or differentiation. Attendees score their own go-to-market in the room. Useful on Tribus as much as on clients.

Four senior Leeds business voices giving "an evidenced read on the region's next five years." Thin on published detail, but the room is the point.

"Historically, becoming a great architect required years of lived engineering experience. Today, AI is commoditising much of that knowledge." Argues architectural judgement becomes *more* valuable as build cost falls, and coordination becomes the bottleneck. No panellists named yet — worth asking.

Real named engineering leaders, which most of the programme lacks: Macs Dickinson (Director of Engineering, LHV Bank), Nayza Fioritta Neves Ferro (Head of IT, Lhasa), Simon Shaw (Delivery Director, Codurance). Team structures, leadership dynamics, skills that now matter.

"The tools teams standardised on last year are already being replaced." Specifically covers where risk lies when the pace of change outstrips an organisation's capacity to absorb it, which describes a lot of client organisations.

The most technical AI session in the programme. Agent architecture, context engineering, memory management and deployment pitfalls; then AI red teaming with open-source tooling. Overlaps with the Development · AI selection.

Aimed at technical leaders whose teams are capable but who remain the central decision-maker. Attendees leave with a Team Ownership Audit. Virtual and light on ceremony.

Hands-on collaborative problem-solving on the premise that core engineering practice still matters. Works well attended in pairs rather than alone.

Products failing because nobody designed for the environment they entered, drawn from breast and lung screening programmes. Interactive, and attendees leave with a shared roadmap. The single strongest fit in this selection.

The five failure modes named verbatim, and mediation positioned across prevention, live management and resolution. Useful for anyone who has to raise a problem before it becomes a dispute.

Names the hidden costs of faster output: expanded review cycles, coordination overhead, reduced learning, low-value deliverables. A delivery problem dressed as an AI problem.

Operating models for citizen services, reconciling service design with a product mindset, how success gets measured. Buyer-side psychology from organisations that fund projects rather than products, as many client organisations do.

Nicola Pinkney (Herd) and Paul Haywood (Principal BA, Hiscox) on reading a room — tone shifts, body language, holding control of a conversation. Unfashionable, no AI angle, and directly applicable to a client steering group meeting.

How leadership brings people along with a shifting technology strategy without losing direction, morale or delivery pace.

"In a world obsessed with dashboards and algorithms, are we at risk of creating work that's efficient but forgettable?" The argument that "best practice slowly becomes average practice." Two senior agency practitioners on the central risk to creative work.

They've dropped the panel: real briefs built live across virtual production, CGI, AI and traditional crew. Explicitly aimed at commissioners and creative directors. A rare chance to watch senior creative-technical judgement being exercised rather than described.

About software architecture, not buildings — but the argument transfers exactly to design: as execution gets cheap, judgement about direction becomes the scarce thing. Worth pairing with the Technical Leadership selection and comparing notes.

Designers who arrived at the craft by unconventional routes, on what those earlier careers gave them and how it shows up in the work. Practical on portfolios and positioning as well.

Sixth edition. Nate Langley (Growth Design, Monzo), Lara Dawson (Head of Product, Mastercard), Paul Jervis Heath (HSBC). Notable for being the one substantial design event with no AI framing at all.

Why well-built products die on contact with the organisation. The single most useful idea in the programme for anyone designing software that has to survive a client's internal politics.

"When every research project starts to look the same, the insights often do too." Hands-on techniques beyond interviews and usability testing, plus how to advocate for them internally. Two sittings, so it's easy to fit. Refreshments throughout.

Three short writing bursts led by prompts, for people dealing with digital fatigue. Spelling and grammar explicitly don't matter; "everything produced in this session is valuable." The page is clear it isn't therapy.

"Making light fit real-world surfaces" — aligning and shaping projected images onto physical objects and environments. Pure craft, no screens, no discourse.

Sixth edition, and the programme's main design gathering: talks, Q&A and relaxed networking with refreshments. Deliberately informal.

Leeds' electro-pop history and the city's part in the birth of techno-pop in 1981, Soft Cell included. Creativity through sound rather than software, and it's outdoors.

Digital clutter, scattered information, and systems that no longer work — and what thoughtful organisation does for confidence and wellbeing rather than just productivity.

Explicit big-supplier-meets-SME matchmaking, with a panel spanning CGI, Leeds Business Anchors, Leeds City Council and NHS England, then open networking. Subject matter includes health tech, AI, software development and digital enablement. Day one of the festival.

Four senior Leeds business voices on the region's next five years — pitched as "an evidenced read," citing a digital economy growing 125% faster than the national average. Useful ammunition and a useful room.

The argument that B2B teams working in channel silos are leaving pipeline behind, because AI Overviews and answer engines actively cite YouTube, Reddit and LinkedIn. Closest to the search discipline of anything here, and B2B-specific.

Attendees score their own go-to-market against their frameworks, then pool findings across the room. Aimed at founders and marketing leads at scaling businesses — read it as a Tribus exercise, not a client one.

Hazel Savage (ex-CEO Musiio, now VP Music Intelligence at SoundCloud), John O'Gara (Sentio Partners), Brannan Coady (ex-CEO YourParkingSpace). Founder and investor company, and a strong room for BD.

Small studio session — the challenge, a framework, live pushback, Q&A. Attendees leave with a content calendar. Practical and small-format.

What "optimising for AI" actually means, how to measure performance when clicks fall and journeys fragment, and how to connect AI visibility to traffic, conversions and revenue. The most substantive GEO session on the programme.

Specialist lender, expensive clicks, low conversion, volatile demand. Campaign restructuring plus richer conversion data: +200% leads, −70% cost per lead, +120% conversion rate, −7% media spend. Rare to see figures published like this.

Full-funnel paid media: search cannibalisation, brand spend constraints, strategy across Google, Meta and TikTok, with triple-digit generic search uplift while reducing brand spend. Good companion to the Shawbrook session — same venue.

"Search isn't what it was." Short, blunt, and framed around the steps to stay visible. Lightweight but it's the local conversation and worth being in the room for.

Moving from measuring historical activity to predicting outcomes, and the "cohort of one" idea. Argues most organisations still optimise for a "happy path" that reflects how businesses want customers to behave. Directly CRO-relevant.

Client-side marketing leaders rather than agency voices, on considered purchases with long research cycles and multiple decision-makers. Includes where the panel sees consumer marketing in 2027.

The most concrete paid-search case study in the programme, with published figures. Campaign restructuring, conversion data quality, continuous optimisation — the actual mechanics of the job.

Search cannibalisation and cross-platform strategy across Google, Meta and TikTok. Same venue as Shawbrook, so pair them if the timings allow.

Thousands of SKUs, brutally competitive categories, a decade of iteration on how search contributes to growth. Retail search at scale.

Measurement when clicks fall away — the problem that will define performance reporting over the next two years.

"Think Race Across the World, but for marketing strategy" — pairs competing against regional marketers to build a strategy across four locations. Genuinely good fun and a fast way to meet people.

Negotiating pay, promotions and scope, on the premise that progression in tech isn't only about technical expertise. Practical, scenario-based, and more concrete than most panels.

Whether architecture becomes an expected capability of every engineer as AI absorbs implementation, and why architectural judgement gets more valuable as build cost falls. The most directly relevant session to technical decision-making.

A sharp critique of cloud's unmet promises — meant to simplify, delivered complexity, cost and rigidity — resolving into the case for building on Cloudflare's edge. It is a vendor-aligned argument, but architecture challenges can be submitted in advance and addressed on the night, which is worth the ticket alone. Optional developer lab after.

Agent architecture, context engineering, memory management, deployment pitfalls — then AI red teaming with open-source tooling. The most technical AI content available.

Ignore the title — it's really about triage: why severity scores rank findings badly, and how reachability and exposure cut thousands of findings to the dozens that matter. Two interactive exercises, including an attacker's-eye vote.

Collaborative problem-solving on the premise that core engineering practice still matters. Also the best place to meet the Leeds engineers who care about craft.

"Three tech challenges. Three solutions. One open discussion." Real problems from Microsoft Fabric implementations — architecture decisions, obstacles, outcomes. Different stack, but honest post-mortems are transferable.

The programme's best argument: as implementation gets cheap, judgement and coordination become the scarce goods. Directly about where technical judgement sits once build cost falls.

Starts from "raw LLMs aren't enough" and argues three things must move together: data infrastructure with semantic layers, engineered business context for agents, and team adaptation. The most architecturally serious take on agents in the programme.

"AI is fundamentally changing how software is built, but it isn't replacing engineering. It's redefining it." Where AI actually lands inside engineering teams today, plus the governance needed for safe adoption.

Three real customer problems from Fabric implementations: lakehouse architecture, fragmented data estates, data quality and governance. Honest engineering post-mortems rather than platform features.

"Live, Unscripted, Bring Your Laptop." Community builders shipping in front of an audience with no safety net. Short, cheap, and the format says more than a panel would.

Real coding challenges, collaboratively, with other Leeds developers. Open to all levels.

Three parts: build an app from a design brief with AI tools, swap projects with another team, then run a "Vibe Code Review" for security, quality and vulnerabilities against their framework. Participants take the framework away. Open to engineers at any level, mentored by senior engineers throughout.

Real problems solved collaboratively, open to any level, on the premise that core practices still matter. The single best-value thing on the programme for a developer.

Team build-along against a deliberately silly brief — a wedding seating planner with feuding relatives and top-table protocol — using a current AI engineering setup. Lunch, prizes, bring a laptop. Two sittings.

Explicitly pitched at both "seasoned engineer" and people earlier in their career: changing team structures, the skills that now matter, non-linear paths, and closing the gap between technical teams and the business.

"No slides, no sales pitches, just live demos and honest conversation about what's working." Tim Flagg (UKAI) and Nick Toft (River AI). Pizza and beers after.

A live demo where the room and their phones become the system under test, to see how things behave outside perfect conditions. Hands-on and unpredictable. Pairs well with the QA & Testing selection.

A full technical walkthrough of Moti-Me, an IoT therapeutic device with a paired mobile app: the ecosystem architecture, the .NET MAUI implementation, Bluetooth communication with the device controller, activity tracking, and the GitHub Actions pipeline to both app stores. Genuine depth, and a product worth caring about.

Phil Whittaker, Staff Engineer (AI) at Umbraco — a .NET CMS — on agentic software engineering, plus Avuyile Harvey (Accenture) on AI governance challenges. The closest thing to a .NET peer conversation about AI.

Real Microsoft Fabric implementation problems — lakehouse architecture, fragmented data estates, governance. Microsoft stack, honestly discussed.

Agent architecture, memory management and deployment pitfalls, then AI red teaming. Stack-agnostic but the most technically substantial AI session available.

Language-agnostic collaborative problem-solving. Worth it for the room as much as the exercises.

Build with AI tools, swap with another team, review theirs for security and quality against a written framework participants keep.

Talk one: how agents are built and use dynamic context — context engineering, memory management, agent architecture, deployment pitfalls. Talk two: AI red teaming, "thinking like an attacker to uncover vulnerabilities before they reach production," with open-source tooling and mitigations. The most direct treatment of production agent work in the programme.

Opens with "raw LLMs aren't enough" and covers semantic layers in data infrastructure, engineered business context for agents, and why teams have to adapt in step. A systems-level blueprint for the move from prompting to engineering.

Open to anyone interested in AI-assisted development, mentored by senior engineers throughout. Build with AI tools, swap projects, then review another team's output for security and vulnerabilities against a real framework. More learning per hour than any panel in the programme.

Agentic software engineering from Umbraco's AI staff engineer, and the governance problems from Accenture. First Leeds edition of a meetup that already runs in Manchester, so a good one to join early.

"Live, Unscripted, Bring Your Laptop." Competent people building without a net, with the room following along. Low commitment, high signal.

Named senior engineering leaders on what changes as agentic AI embeds in workflows, and explicitly open to engineers at any career stage.

The argument that cloud promised simplification and delivered complexity, cost and rigidity — and the case for the edge instead: fewer vendors, a unified infrastructure and security layer, faster deploys. Vendor-aligned toward Cloudflare, but architecture challenges can be submitted in advance and there's an optional developer lab after.

Crowdsources an idea from the room, an AI assistant transcribes and structures it into a product plan, then a vibe-coding tool builds a working application live in the session. Successor to their Innovation Coffee Morning.

"Live, Unscripted, Bring Your Laptop." The community gets the stage instead of a sponsor. Format is the content.

Nate Langley (Growth Design, Monzo), Lara Dawson (Head of Product, Mastercard), Paul Jervis Heath (HSBC). Worth crossing the frontend-design line for, and the one design event with no AI framing.

Agent architecture and deployment pitfalls, then AI red teaming. The most technical AI session going.

Real briefs built live across virtual production, CGI, AI and traditional crew. Not frontend, but the best example in the programme of technology used for something genuinely novel. Overlaps with the Mobile Development selection.

Experienced Sky Sports designers running a practical UX/UI introduction, explicitly welcoming to the curious as well as the trained. A fast route into the half of frontend work that isn't code.

Build from a design brief with AI tools, swap with another team, then review theirs for security and quality against a framework participants keep. Open to engineers at any level, mentored throughout.

Language-agnostic collaborative problem-solving with other Leeds developers. Best-value hands-on session on the programme.

Product and design leaders from Monzo, Mastercard and HSBC. Informal and relaxed, and useful context on what product and design leads optimise for.

An idea crowdsourced from the room, structured into a plan, then built live as a working app during the session.

"It isn't replacing engineering. It's redefining it." Where AI actually gets used inside engineering teams now, and the skills that will matter next.

Moti-Me: an IoT therapeutic device paired with a mobile app that turns physiotherapy into child-friendly activities. Covers the .NET MAUI app, Bluetooth communication with the device controller, activity tracking, and CI/CD to both app stores. Different stack, same problems: a mobile app talking to real hardware for a genuinely good reason.

They've explicitly dropped the panel — real briefs built live across virtual production, CGI, AI and traditional crew, at every budget scale. Ends with an off-record conversation about what's actually breaking in the industry.

Strolll uses augmented reality to support people with neurological conditions — NIHR funding, a clinical trial, international reach. The technology can be tried first-hand. AR with actual clinical evidence behind it, which is rarer than it should be.

Aligning and shaping projected images onto physical surfaces, objects and environments — "making light fit real-world surfaces." Practical technique, no discourse.

Immersive AI and VR used to rehearse difficult workplace conversations — negotiations, leadership, customer service — with volunteers going into the headset rather than watching a presentation.

Teams build and evaluate an AI-powered Ohbot robot and an LLM-driven agent in Python via the Ohbot API. Student-facing, so eligibility may be limited — worth checking, and worth knowing it exists.

A live, demo-led session where "you (and your phone) become part of the system" — the room turned into a live environment to see how systems behave, and misbehave, outside perfect conditions. Billed as hands-on and a little unpredictable.

The third act is the reason to go: teams swap AI-built projects and run a "Vibe Code Review" assessing security, quality and vulnerabilities against a framework Yorkshire Software Solutions developed — and participants take it home. That's a QA artefact for AI-assisted delivery, which is exactly what Tribus will need.

Really about triage: teams have more findings than they can fix and need "a defensible answer for what gets fixed first." Argues severity scores rank findings badly while reachability and exposure reshape priorities. Two interactive exercises, including re-ranking once context is added.

The second talk is straight AI red teaming — attack techniques, open-source security tooling, practical mitigations, finding vulnerabilities before production. Closest thing to a QA-for-AI session.

Argues faster output expands review cycles and coordination overhead, because "every AI-generated output still needs to be understood, evaluated and connected to a meaningful outcome." That review burden lands on QA — worth hearing it named publicly.

Scope creep, missed deadlines and disagreements over delivery, and how disputes get prevented rather than litigated. Useful context for whoever tends to find the problem first.

Phil Whittaker, Staff Engineer (AI) at Umbraco, on agentic software engineering, with Avuyile Harvey (Accenture) on AI governance challenges. A platform-engineering view of where AI genuinely fits inside a content-managed product rather than around it.

Directly relevant to how client sites get hosted and secured: the case that cloud added complexity, cost and rigidity, and the argument for the edge — fewer vendors, unified infrastructure and security, faster deployment. Architecture challenges can be submitted in advance.

"Search isn't what it was." This is the question clients will be asking about their websites within the year, and it's worth having heard the local answer first.

AI Overviews and answer engines now cite YouTube, Reddit and LinkedIn — so a website stops being the single destination. Structural implications for what a client's site is actually for.

Moving from linear, organisation-led journeys to adaptive, customer-defined ones — and the argument that most sites still optimise for a "happy path" reflecting how the business wishes people behaved. Short, 45 minutes, drinks after.

Where AI is genuinely being applied inside engineering teams, the governance needed for safe adoption, and the skills that will define the next few years of the job.

Umbraco's AI staff engineer on agentic software engineering, plus Accenture on AI governance. A platform-engineering perspective rather than a vendor pitch. First Leeds edition of the meetup, so a good one to be in at the start of.

Fourth year at the festival, run under NumFOCUS Code of Conduct standards. Different language, but the most established community-run event on the programme — instructive on how another community organises itself.

"BUILD TECH. CHANGE LIVES. RIGHT HERE IN LEEDS." The launch of a volunteer engineering community working with a homelessness charity. The one place on the programme organising volunteer engineering work locally.

The cloud-versus-edge argument, with real relevance to how client sites get hosted, secured and deployed. Optional hands-on developer lab afterwards.

Collaborative problem-solving on the premise that core engineering practice still matters. Community-run in spirit, and open to any level.

"Live, Unscripted, Bring Your Laptop." The session explicitly "hands the stage to the community" instead of a sponsor. Short and low-commitment.

Names the five failure modes verbatim — misaligned expectations, missed deadlines, scope creep, communication breakdowns, disagreements over delivery — and treats them as normal. Mediation as prevention, live management and resolution. Core project-delivery material.

Built around the dilemmas marketing clients bring: in-house versus agency versus freelancer, more impact on a shrinking budget, specialists versus integrated, "can AI help — if so where do I start?" Useful context for managing those conversations.

Nicola Pinkney (Herd) and Paul Haywood (Principal BA, Hiscox) on staying "hyper aware of shifting dynamics in the room" — reading tone, body language, steering a conversation. Unfashionable, entirely AI-free, and applicable weekly.

Each person identifies a pattern of workplace friction they've "normalised," works out where it actually sits in the system and what it costs, and leaves with one small testable next step. Small-group and facilitated, "using direct language without the consultancy jargon."

How organisations absorb constant technology churn without losing direction, morale or delivery pace. Pitched at senior leadership, but it's PMs who feel it first.

Why well-built products die on contact with the organisation they were built for. Interactive, roadmap-building, and aimed explicitly at people in delivery roles.

A squirrel-and-acorn metaphor for gathering soft-skill tools and knowing which to reach for: reading tone shifts, the science of body language, holding control of a boardroom conversation. Nicola Pinkney (Herd) and Paul Haywood (Principal BA, Hiscox). No AI angle at all, and virtual so it's easy to fit in.

The five named failure modes, treated as ordinary rather than exceptional, and mediation across prevention, management and resolution. One of the most practically useful sessions in the programme for delivery roles.

The framing maps closely onto project delivery. Identify a friction pattern that has stopped being noticed, locate it in the system, work out what it costs, leave with one testable next step. Facilitated small groups.

Negotiating scope, pay, promotions and role, on the premise that progression isn't only about technical expertise. Scenario-based and practical.

"Brilliant products go nowhere… because nobody designed for the environment they were entering." Explicitly for people in service transformation and delivery roles, and attendees leave with a shared roadmap.

Names the real cost of faster output: expanded review cycles, coordination overhead, low-value deliverables. That coordination overhead lands on delivery, so the language is worth having.

"The biggest opportunity isn't replacing people — it's helping them become better communicators." Live demos of practising difficult conversations in VR with AI-powered characters, and customer service is one of the named scenarios. Volunteers go in and do it rather than watch. Directly relevant to front-line support.

Subtitled "when AI validates assumptions instead of facts." As businesses rely on AI to approve requests and verify information, "trust is rapidly becoming the newest attack surface." Sharp, and directly relevant wherever AI touches a support flow.

How their Customer Safety Centre of Excellence combined technology, immersive experiences and culture to change behaviour, using human-centred design so colleagues make better decisions. A support-and-safety operation described properly.

User-centred design capability and maturity frameworks, run by a regulator rather than an agency. Front-line support roles have more to contribute here than most attendees.

The argument that organisations optimise for a "happy path" reflecting how they wish customers behaved, rather than how people actually move through a product. Support roles see the unhappy path daily.

For the delivery side: the five failure modes named plainly, and how disagreements get defused before they become disputes.

On firms "brilliant at designing software" that leave "their greatest assets feeling invisible and unmotivated." Introduces the Conversation Canvas, a free one-page tool that gives managers confidence to have meaningful wellbeing conversations — with applications in onboarding, performance reviews and exit interviews. Also covers psychological safety, trust and turnover. Virtual, so it costs an hour.

"AI handles tasks. Humans drive innovation… Real productivity isn't about working faster like a machine; it's about enabling your team to think strategically and creatively." Engagement's effect on productivity and how to foster it. Explicitly aimed at heads of people and wellness leads.

Five generations at work at once, framed as an opportunity rather than a problem and deliberately anti-stereotype. Victoria Tomlinson (CEO, Next-Up), Helen Kewell (Partner, Q5), Emmanuel Togoh (Purposed NGO), facilitated by Joe King (Q5). Presentation, panel and live Q&A.

A free, lived-experience-led taster on the barriers neurodivergent and learning-disabled employees face, with practical fixes and an actionable improvement plan. Note the detail: attendees can then enrol up to 20 colleagues on the full CPD-accredited one-day training. That covers most of Tribus.

"Common watchouts and pitfalls when growing your workforce." The three-discipline panel is the draw: legal, people and founder perspectives on the same problem in one room.

Three short prompt-led writing bursts for people dealing with digital fatigue, with techniques transferable to work and home. A strong candidate for something to run internally afterwards.