Welcome to our revamped newsletter, which we are expanding into a learning companion for leaders and practitioners working to advance real-world applications of AI and emerging digital workplace technologies. We will continue to publish this free edition every other week, followed by a premium edition the following week that includes practical implementation support such as capability maps, recipes, playbooks and other learning content from our advisory and executive education work, plus occasional videos, interviews, podcasts and more. See the bottom of this post for more details.


Looking beyond basic LLM use in the enterprise

The current wave of GenAI development continues to elicit breathless coverage, despite some evidence of a plateau in both LLM gain of function and usage and a growing sense that its imitation learning approach might not rival real human (or animal) intelligence.

Adoption and proof of value in narrow domains such as medical imaging, law and co-piloting for programmers are looking strong and credible; similarly in very broad but shallow use cases such as knowledge synthesis, image generation, writing and research support, where LLMs look like becoming the new search engines.

But some of the most impactful use cases are in messy human/machine domains that are neither as narrow nor as broad, especially in the enterprise.

As with every emerging technology, the first wave of adoption efforts has been characterised by throwing new tech at existing ways of working in the hope that it can help us do the old things more efficiently. Just as the first wave of automation was led by Robot Process Automation (RPA) – basically training software bots to fill in badly-designed enterprise systems and forms faster and more reliably than people – so too the first enterprise AIs are being embedded in old systems such as basic office tools (e.g. the tool that replaced overhead projector acetates, and the other one that replaced paper memos), enabling people to perpetuate corporate spam, but faster and with less effort or mastery. Looking ahead, it is likely that LLMs could solve the problem of terrible enterprise search, make predictive analytics easier and more accessible, and enable us to move beyond RPA towards real service and process automation.

If we want to move beyond this basic first stage and uncover ‘deep value’ use cases, we need to zoom out to understand the bigger picture, and then zoom right into the detail of what we really need to change, rather than continuing in the quest for faster horses or adding another confusing ribbon to MS Word.


Organisations as software and a new literacy

Increasingly, organisations are like software systems for collective human endeavour. They have clear fitness functions such as profit, long-term value creation, customer (or societal) outcomes, employee satisfaction, and so on. They need to evolve, adapt, learn, and they need to become more collectively intelligent over time, not less.

Early software used to share the simple, top-down centralised structure of process and task assignment that continues to be the norm in large corporations; but over time it has evolved to become so much more sophisticated, layered, smart and connected. Processes and systems that might once have been written from scratch are now invoked by a few magic words in a command line interface, which is as impressive in its own way as the use of prompts in GenAI.

Just as human language evolved to convey a lot of information in small sequences, programming languages followed a similar path. It is entirely feasible that we will soon see this kind of powerful invocation used in organisations to operate complex structures: “calculate the feasibility of entering the Saudi market with our product, and if the 3 year projection looks positive, create an online sales and marketing operation and start feeding me internal candidates for a team to run it.”

The programmable organisation, in which small autonomous teams and groups using smart systems and structures can achieve more than entire divisions of a C20th corporation, might become a reality sooner than we think, especially in areas of high competition and dynamism.

But we need to know what to ask, and in the short-term we also need to know how to ask LLMs questions and engage in turn-based dialogue to get the most out of them. Plus, if we want to design specialised bots or AI helpers, we will need to get better at reverse prompting and the shaping of conversational interaction to gently nudge users towards the questions and topics the bot is there to help with, rather than just present a generic text input and then potentially disappoint people, as many early chatbots found to their cost.

This is a new literacy – and just as writing helps us structure our thoughts and ideas, harnessing AI starts with knowing what we want to achieve and bring able to express it clearly.


Real-time coordination and adaptation over manual meat-based micro-management

Even before the AI hype cycle got going, we were already seeing signs of a management shift away from the dark arts taught by C20th business schools, with layoffs, removal of excess management layers and more focus on autonomy and agility in teams.

Whilst the people coaching side of management remains useful, and real leadership is still vital in navigating change, there is less and less need for manual management to coordinate work outputs. This is precisely the kind of repeatable task that AIs find trivially easy, and we anyway have open data and dashboards that mean people and teams can self-manage against their targets and goals.

How can we help managers move up the value chain to become prompt whisperers, coordinators or even peer contributors in their teams? And how can we capitalise on their experience to become change agents in a time when change is badly needed?

Whether management layers continue to be removed or managers evolve into new roles, one thing is clear: we need to re-think executive education to decisively move away from the notion of generic, fungible management – an out-dated idea that has brought down previously great companies such as Boeing and GE – and focus more on real skills and real work.

But the problem is not just how the idea of management has been practiced. It is also structural. It is extremely hard to create connected intelligence in a vertically-divided hierarchy. After over a century of cultivating artificial stupidity in structures that militate against thinking, it is quite a challenge to design for intelligence, whether human or machine.

We believe that AI will be most effective in organisations that take a service-oriented or modular approach – ideally platform-based – with a connected architecture that allows independent action, autonomy and fluid composability of services. But the good news is that AI can also help us get there by strengthening local team capabilities and reducing dependencies.

Even startups that have become scale-ups are not immune from this tendency to recreate the worst managerial structures of the C20th corporation, as we have seen with multiple phases of over-hiring and over-firing in high-growth tech firms.

Alternatives exist and are proven, but are often overlooked by general managers in the race for growth, despite being eminently more scalable than just squeezing the old system harder.

The platform model, agile teams / lattice structures and the ecosystem model of Rendanheyi are all established design patterns at this point. What they all have in common is the centrality of small multi-disciplinary teams empowered to take ownership of their own service outcomes, and this is where a combination of digital-first working and helper AIs / automations could produce impressive productivity gains. The key to adaptive structures that can evolve to meet changing needs seems to be small teams operating in a flexible, networked or cellular structure surrounded by ambient knowledge and data to help them navigate.


AI + agile structures + new ways of working

So whilst AI needs adaptive, connected structures to fulfil its potential in the enterprise, it is also true that AI combined with business agility and new ways of working could be the key to accelerating this transition. Organisations rarely have the luxury of pausing operations to undertake a big organisational redesign. Better to think of change as a series of small steps that each bring improvements, whilst laying the foundation for a new approach to coordinating work at scale that gradually removes the need for top-down process management. This would remove significant cost and inertia from our operating models over the medium term.

Most organisations agree that they want to see people using AIs to be more productive, rather than people being replaced by AIs – in other words automating tasks, not necessarily removing roles. With fewer time-wasting mundane admin and coordination tasks or ‘busy work’ to focus on, we can make roles more hybrid and interesting, and we can operate more efficiently, using automations, helpers, concierge bots, and so on.

But to really embrace AI with smart structures and start to see the benefits of agile, adaptive organisations, we need to work on three important challenges:

  • Readiness, both technical, governance, security, data and the upgrade of internal platforms & services to create a more connected ecosystem;
  • A reinvention of learning in the workplace that is agile, available when you need it, and not tied to mass classroom-based programmes; and,
  • Imagination – as we are seeing with GenAI, if you can describe how a function should work, then we should be able to deploy AI and smart tech to make it happen – but we need to cultivate the imagination to achieve our goals in new and better ways.

In our work on emerging technology adoption in large organisations, we have watched three previous waves of internet innovation wash over corporate structures – Web 1, Web 2 and two competing versions of Web 3 – but the silos and the vertical reporting lines are still standing.

For this new wave to enable the smarter, simpler, social organisational structures we have been advocating since 2002, we need to do a better job of bringing together different stakeholders – primarily tech functions, people functions, and ops/process functions – to find the optimum combination of human intelligence and machine pseudo-intelligence. C20th organisations were like dumb machines made of smart humans, and idealists often used to talk about the need for organic, natural, living organisations. But what we really need right now is for machines to be machines and free up people to be human.

We have the opportunity to combine platforms, service automation, connected data and a social web of people and teams coordinating directly without expensive layers of hierarchy, using a variety of mini-AIs, agents and helpers to coordinate and provide visibility and insights.


Shift*Base: a mission to re-invent executive education and help design better organisational operating systems

To achieve this kind of change, and for the huge investments in raw AI tech to make sense in an enterprise context, we need to get practical and we need to reinvent the way we learn, plan and guide organisational development. If organisations are like software, where are the kind of development tools that coders have, like IDEs (Integrated Development Environments)?

How can we map our existing digital capabilities and dependencies, and define and design the new business capabilities that emerging technology makes possible? How can we build a common picture of the service landscape that we can all work from, rather than continue with a free-for-all of point solutions and bought-in software all over the place?

Defining a business capability, its service components & dependencies


Our mission with Shift*Base is to create a development environment for digital business capabilities that can do this (we call the platform a digital capability accelerator), as well as provide in-situ contextual learning and operational playbooks to put these new capabilities into practice. Today we are launching ShiftBase as a new venture, previewing the platform we have been building, and sharing some of the learning content that Shift*Base will provide.

As a first step on that journey, we want to help get all executives – technical and non-technical, senior and emerging leaders – up to speed on what is possible and where it can add value in the real world, and stimulate the imagination to think differently about how work is coordinated.

We need much faster and more practical learning, and a cross-functional conversation about use cases that starts from the same map – not one for tech, one for people functions and one for operations. If we can use our collective imagination to design and map meaningful use cases, build support for implementing them, and engage the workforce with learning about the changing world of work, then we can make a lot more progress, much faster.

We want you to join us on this journey, so we are creating the Shift*Academy – a personal learning companion for executives at every level, from senior leaders grappling with what to do with AI to emerging team and function leads who want their own leadership journey to be faster and more fulfilling.

We will be curating and sharing what we know from our own executive education work on digital leadership, digital organisations and transformation, and we want to learn from you (and with you) by sharing use cases, recipes, examples and ideas of how to move towards smart programmable organisations, where leaders can be both prompt whisperers and people whisperers to create smart, sustainable ways of working that cost orders of magnitude less than bloated management-heavy bureaucracies.


What you can expect from Shift*Academy

  • Shift*Academy is a personal learning companion for leaders and practitioners who want to apply AI and digital workplace tools to improve the way their organisations function. There are dozens of newsletters and courses about technical developments in AI, but few that focus on the challenges of real world adoption in organisations. This is where we will direct our focus.
  • The bi-weekly free edition of this newsletter will continue to help make sense of developments in AI and emerging technologies, especially those relevant to the enterprise. Please share it with anyone you think might be interested.
  • The bi-weekly premium version will follow a week later, deep diving into use cases, implementation recipes and playbooks and sharing detailed capability maps. We have gifted a 60-day free trial to existing subscribers as a thank you for your contributions, and we will be offering referral bonuses and further offers to our practitioner colleagues.
  • Premium subscribers will also receive bonus content such as video learning sessions, interviews, podcasts, online discussions and more, plus they can access a live feed of the curated links and articles shared by our research team, both as an RSS feed and Substack notes.
  • For executives, we will bring actionable insights, implementation ideas and real-time learning, freed from the constraints of classroom programmes.
  • For practitioners and consultants, we will bring playbooks and case stories that help your own practice, and we invite you to share your methods and ideas with us and our audience, and we will offer a generous referral programme for you to share this with your own clients and colleagues.

We hope you will continue to join us in this journey and share the learning as widely as possible. If you have any questions or suggestions about how we can better pursue our learning mission, please let us know – we would love to talk to you.

Our first premium edition next week will deep-dive into a big capability that is often neglected by people functions in fast-changing organisations, but could be a great use case for simple AI tools and services: digital talent discovery.