Jon Smart

"Every technical decision that we made along the line was basically an organisational one. It’s not about the technology." When Telenet started rolling out agentic coding, they had a timeline, a budget, take-up assumptions and a change plan. Within weeks, not months, it was wildly wrong.

18h

It wasn’t bad planning. It’s the reality of emergent work. So they stopped managing it like a programme and started managing it through principles and guardrails, asking: “If you can’t plan it, how do you make sure it happens safely?”

18h

Lesson: assume your plan will be wrong, then organise yourself to manage that. The full Telent case study is now on YouTube. #BVSSH

From PMO to Value Realisation Office (VRO). ▶️ Portfolio OKR controls (not too many Os or KRs, alignment over cascade, nested OKRs with parents, transparent, well-written outcome hypotheses) ▶️ Coaching & support for value-aligned multidisciplinary teams (cadences such as MBR and QBR,

coaching teams on outcomes and cadences, visualising, managing & breaking dependencies, taking an enabling leadership stance 'how can we help?') ▶️ Data feedback loops (ensuring that the KRs are measurable and measured, visualising data in tooling, reporting, inspect & adapt) #BVSSH

The solution to all of our Psychological Safety problems! #BVSSH

Increasing organisational outcomes, with the help of agentic AI, is first and foremost a cultural challenge. In this joint session, Barbara Arnst and Benjamin Geens share their case studies on introducing AI agents at work, and why the hardest challenges aren’t technical.

* Tacit knowledge limits AI, if it’s not made explicit, agents simply can’t use it * Change outpaces adoption, with organisations unable to absorb the speed of progress This session goes beyond the hype, sharing what worked, what broke, and what’s still unresolved.

youtu.beAgentic Delivery on Steroids: Why the Hardest Problems Weren’t Technical | TelenetWhat Happens When You Scale AI in a Large Organisation? AI doesn’t fix organisational problems. It amplifies them. Barbara Arnst, VP Transformation at Telenet, and Benjamin Geens, Director at Nova…

If you’re exploring AI in your organisation, this is a grounded, practical perspective on what really needs to change to make it work. Watch here: youtu.be/O0SgIxV2SiE #BVSSH

These are real-life case studies of change: a knowledge platform build-out (small-scale) and an enterprise-wide initiative (large-scale) at Telenet. The problems that emerge in small AI experiments are the same ones that arise with transformation at scale.

The recording from their talk at Sooner Safer Happier LIVE is now available. Expect an honest look at what happens when execution speeds up, but organisations don’t: * Decision-making becomes the bottleneck as teams struggle to keep up with accelerated delivery

The defining organisational challenge of the industrial age was how to maximise productivity. The defining challenge of the intelligence age is how to optimise for balanced value. #BVSSH

Improve via ▶️ Align: why, who, where (today and future North Star shared value goal) and how will we know? ▶️ Act: bias to action, learn by doing, experimentation, with psychological safety and thin slices of learning & value, within minimal viable guardrails.

▶️ Reflect: what did we expect to happen? What actually happened? ▶️ Adapt: what do we want to learn next? How soon can we learn it? What help is needed?

Over annual output planning with a "business case" (not a "customer case") that is rarely, if ever, revisited and individual objectives, in role silos, in the HR system, imposed top down. #BVSSH

From Objectives To Outcomes #BVSSH

How do you modernise a 926-year-old organisation without breaking anything? HMRC collects £875 billion a year and serves millions of citizens and businesses. It also runs one of the UK’s biggest, oldest and most safety-critical digital estates. So how do you change something that can’t stop?

* Start small to scale big: prove value quickly, then expand what works * Cut time to value: remove barriers and empower the teams closest to delivery * Build safety into speed: innovation that strengthens critical services rather than putting them at risk

youtu.beHow to Change Everything Without Breaking Anything | Tom Skalycz, HMRCHow do you transform a national system that can’t stop? HMRC collects £875 billion a year and serves more than 35 million people and 5 million businesses. It has to modernise a large legacy estate…

* Put people first: better experiences for users, teams and partners * Learn by doing: a bias to action, within guardrails If you’re leading transformation in a complex environment, this is the one to watch. Watch the recording: youtu.be/-nhsTBeVzp4 #BVSSH

In this Sooner Safer Happier LIVE session, Tom Skalycz, HMRC’s Chief Technology Officer, shares what the team has learned so far. The recording is now available.

At the heart of it is Path (Pioneering Advanced Technologies for HMRC), a protected space inside a live national organisation to experiment safely, learn fast, and adopt modern technologies without lowering the bar on security, resilience or public trust. What you’ll take away:

Why we work the way we do today: ▶️ We have come from 4 Industrial Revolutions. Cotton, Steam, Steel and Oil. The work for the majority of people has been repetitive and knowable, with people as cogs in a machine, increasingly specialised in narrow roles, with planning (order givers)

runs many times. There are robots on the factory floor. Applying for household insurance online involves no humans. We have automated many manual repetitive tasks. More people are spending more time on new value, on product development, making the most of the new means of production.

GenAI, as a stochastic probability engine, is emergence squared. We need to optimise for unknowability, for outcome hypotheses, experimentation and learning, if we want to optimise for outcomes. #BVSSH

physical properties, it does not apply to social systems to people. This is why we have role-based silos, with work passing, a focus on output and view unique work as predictable, with a focus on a plan put together at the point of having learnt the least.

▶️ We are now in the 1st Digital Revolution. This is the biggest change in how we think and work since the 1600s. Increasingly, the work that people do in organisations is unique and unknowable. It is emergent. We don't write the same code a million times. We write it once, refactor it and it then

separated from execution (order takers) and a focus on output. Previous ways of working have a direct link to the Scientific Revolution in the 1600s, with a view of the Clockwork Universe, where, once observed, there are universal laws that forever hold true. This may apply to Physics and

Product ≠ Value It has been a significant improvement in ways of working, going from deterministic, temporal projects (build it and leave it, treating the unknowable as if it were knowable) to emergent, long-lived products (customer-centricity, stable, not static,

The products and services combine to enable that value. This is the Shape of Value. Understanding how products, services and capabilities come together to create value for the consumer and the producer. Making visible the essential dependencies, while minimising the non-essential

and accidental dependencies. Otherwise, we risk locally optimising products, rather than optimising for value end to end. A product isn't necessarily the unit of value. What is your Shape of Value? What are your essential and non-essential dependencies? #BVSSH

high-performing multidisciplinary teams, etc.). And the product is not always the thing of value. Take a trading platform at a bank. There might be a Trading Product, a Pricing Product and a Risk Management Product, each with its own product team, backlog and priorities.

The customer doesn't value three separate products. They don't consume one of the products on its own. They value being able to trade effectively, safely and at an acceptable cost, which requires pricing, position management, execution and financial risk management.

As a leader, if you want better outcomes, what is needed from you is really quite simple: ▶️ Incentivise, recognise, reward everyone Continuously Improving (as measured by a balanced set of outcome metrics for the How such as Better Value Sooner Safer & Happier) ▶️ Provide support.

People and process to systematically identify the biggest impediments and to alleviate them ('Enablers') with a data feedback loop. It is as simple as that. 99% of organisations are too busy being busy. There is no priority to improve 'how we do, what we do' because we're too busy

(read incentivised, prioritised) to just do the doing. It's simple: it's not about <insert shiny object name here>. It's about creating incentive and priority to go slower to go faster. #BVSSH

Lasting change requires many people to unlearn and relearn. People need time to undergo that process. The pace of change is determined by the speed of unlearning The speed of (un)learning is determined by the amount of psychological safety

Role-model and reward behaviours that reduce fear, encourage experimentation and enable intelligent failure: quick, unique, cheap and safe. To go faster, create the conditions for people to learn faster. #BVSSH

AI will not transform your organisation, it will amplify it. It will amplify good and bad.

✅ if there are fast feedback loops and adaptability, ✅ if culturally there is psychological safety and support to challenge, to suggest crazy ideas, to not 'stay in your lane', to try something new, to learn by doing, to optimise for emergence, to be a continuously learning organisation

then AI can become a positive multiplier for Better Value, Sooner, Safer and Happier. This makes getting the foundations in place even more important. AI does not transform your organisation. You do. #BVSSH

AI will not transform your organisation, it will amplify it. It will amplify good and bad.

❌ If ways of working are fragmented, siloed and output-driven, AI will help you produce more fragmented outputs, faster. Partially completed work will stack up ever higher on the brick walls between functional silos. ❌ If incentives reward local optimisation, AI will optimise locally at scale.