The current state of the office market is no Black swan event. 

Right from the very start of Covid, when suddenly 95% of ‘knowledge workers’ went from working in the office to working at home, it was screamingly obvious that this would forever change how we thought about ‘the office’. 

The first thing people learnt was that there were already digital tools available that would make interacting virtually much easier than many, even most, thought. And these of course got much better very quickly. The likes of Zoom suddenly had the money, inclination and incentives to iterate on an already good product very fast.

And today, with a plethora of new and improved tools, working apart really is no big deal. In very many circumstances.

We also all learnt of course two other very important things, that none of us can unlearn. First, that commuting every day of the week is absolutely unnecessary (for almost all ‘knowledge workers’) and secondly that having some agency over our time, some flexibility of when we put in the hours required of us, is a very good thing.

And the future is just going to bring more of the same as the tools available to us get better.

In fact getting ‘better’ understates the point. Because we are now in a world where Generative AI has arrived….. and that changes everything.

The amount of computing power available to AI researchers has doubled every three months for more than ten years. Massively out performing the exponentiality of Moore’s Law. And this has led to the Cambrian explosion of AI in the last year. OpenAI’s GPT-4 was not just a bit more powerful than GPT-3, it was an order of magnitude more powerful. Ten times the power. When GPT-5 arrives (perhaps within a year) that in turn is likely to be 10X GPT-4. In fact it may be much more. Mustafa Suleyman, the co-founder of leading AI research lab DeepMind (now owned by Google) reckons the compute power available to his current company, Inflection AI, will be 100X today in just three years time.

10X or 100X isn’t important. Both represent extraordinary increases in computational power, and both will have, cannot fail to have, serious consequences. For good or ill – though that is a topic for another article.

Very rapidly we are moving to living in a world that will be mediated, in one way or another by AI. Much, even most, of what we do will be filtered through the lens of AI. AI will help us be better humans. It will enable us to call on exceptional intelligence at every step of our way. Used wisely (and everything I say here is caveated by the need to act wisely) AI should be able to make our lives easier, more fulfilling and dramatically more productive. Each of us will have ‘infinite interns’ (to steal Benedict Evans’ phrase) at our disposal. Each of us will have ‘our people’ to assist us with whatever it is we wish to do.

Given this, it is obvious that how we work, and indeed the work we do, is set for considerable change. Misleading media headlines aside, the business world has already largely accepted that hybrid working, in various forms, is the new baseline for knowledge work. So the questions going forward are how to make our new ways of working optimal, and what real estate do we need, as individuals, teams and companies, to enable us to be as happy, healthy and productive as we are capable of being.

And this is where Generative AI ‘could’ be hugely important, and powerful.

Let’s break down the mechanics of how Generative AI could function in the context of remote work management.

Assuming that the office is going to be ‘a network of spaces’ how might Generative AI be leveraged to manage a remote first workforce. And by ‘remote first’ I do not mean ‘remote’ but rather a company where not being in a central HQ is the default setting. Just the reverse of ‘office first’ where being in the central HQ IS the default setting.

Here then are the process steps we’d implement with Generative AI:

1. Data Ingestion: Initially, the AI system would ingest a wide array of data points: project timelines, task requirements, team members’ skills, their working hours, time zones, historical productivity levels, and perhaps even qualitative inputs like self-reported team morale or job satisfaction scores.
2. Pattern Recognition and Prediction: Traditional AI might use this data to allocate existing resources most efficiently, based on historical trends. Generative AI goes a step further by simulating thousands of potential scenarios to anticipate future challenges. For instance, it might predict that a team is likely to face burnout in two weeks, given their current workload and historical morale data.
3. Optimal Workflow Generation: The “generative” aspect comes into play here. The AI generates a variety of workflow options tailored to these future scenarios, rather than merely adapting existing workflows. These aren’t just reactive changes but proactively designed workflows that take the predicted future state into account.
4. Dynamic Adaptation: As work progresses, the AI system continues to ingest new data, refine its predictions, and dynamically update its generated workflows accordingly. If it predicts burnout, it might suggest changing the sequence of high-effort and low-effort tasks, rearrange team roles, or even recommend breaks or team-building activities.
5. Iterative Learning and Refinement: As the team follows (or deviates from) the generated workflow, the AI learns from the outcomes to continuously refine its future predictions and generated workflows.

In sum, Generative AI doesn’t just tell you what’s likely to happen based on current data (that’s predictive analytics). It suggests entirely new workflows, team structures, or task sequences that are most likely to succeed in the future, given a host of variables. The goal is to increase productivity, decrease burnout, and enhance job satisfaction, which makes it far easier to manage remote teams efficiently.

So far so good, but how do we ensure that the ‘jobs to be done’ are being done in the right environment, in the right real estate?

We know that the future of the office is evolving towards fulfilling human-centric needs like congregation, collaboration, and socialisation. How then do we use technology, Generative AI, to make sure ‘the office’ is used when and where it adds the greatest value?

The answer is that Generative AI could play a significant role in understanding and integrating these more nuanced, qualitative factors into remote work management in a few ways:

1. Data Collection on Human-centric Metrics: Generative AI can start by incorporating data on employee engagement, creativity, and the success of collaborative projects. This could include surveys, employee feedback, and even metrics like the frequency and quality of inter-team collaborations.
2. Simulating Virtual Collaboration Scenarios: Generative AI could simulate various team compositions, meeting frequencies, and virtual collaboration tools to predict which combinations yield the highest levels of engagement or the most successful collaborative outcomes.
3. Holistic Work-Life Models: Beyond work tasks and deadlines, Generative AI could incorporate data on employees’ personal lives, like preferred times for social activities or outside-of-work commitments. It can then generate work schedules that balance professional responsibilities and opportunities for socialisation and mentorship, both in virtual settings and in scheduling in-person office interactions.
4. Adaptive Learning Environments: Generative AI could tailor individual learning paths based on each team member’s interaction with mentorship or collaborative programs. For example, if it identifies a gap in soft skills that are often developed through in-person interactions (like negotiation or public speaking), it could recommend remote training programs or simulate virtual scenarios for practice.
5. Facilitating Hybrid Models: Understanding these human-centric needs can help companies decide when physical presence is truly beneficial. Generative AI could optimise a schedule that combines remote work with periodic in-person congregation aimed specifically at activities that benefit most from being face-to-face, such as brainstorming sessions, team-building activities, or mentorship programs.

In this way, Generative AI would help us understand that the office is not just a place where work happens, but where people grow, collaborate, and innovate. By optimising for these more qualitative aspects of work, the AI could influence real estate strategies to focus on creating spaces that enable these human-centric activities rather than just providing a seat and a desk for every employee. Thus, the “Space as a Service” model would evolve to offer not just flexible spaces, but spaces designed for specific human needs.

So what we’d have is an AI mediated way of working that took all factors, quantitative and qualitative, into account, that was optimised around human-centric criteria, and that also maximised productivity.

It is important to appreciate that each such system would yield different results because the particularities of any given company, in terms of inputs and outputs, vary so much. Each company has different ‘jobs to be done’ and different people to do them. Extreme nuance and personalisation is the order of the day. The algorithms themselves, that underpin the Generative AI, will be, and need to be, tweaked accordingly. Think of it like a music mixing desk with 64 sliders where we orchestrate the sound we are after.

Trust though is a massive issue here. Systems like this are very deliberately focussed on creating ‘win wins’, where everyone (within reason) is listened to and tuned for. So theoretically almost everyone should be better off. As happy, healthy and productive as they can be. But ….. there needs to be company wide trust that these are the aims, and that such ‘snooping’ is being done for everyones benefit. This’ll work much better in an ‘let them go surfing’ (Patagonia) type company than a ‘sorry but you need to sell your family dog and get back to the office’ (Clearlink) type one.

That said, over the next few years we are all going to become more used to having ‘our AIs’ help us on a day to day basis and so are likely to become more comfortable with such AI mediated work practices.

With our real estate hats on we need to focus on creating spaces that are human centric and deeply technological. That understand wants, needs and desires and flex to accommodate them. At a macro level, humans still need somewhere to work so will still need as much real estate as ever. They will still be spending 90% of their time IN real estate. So demand is and will always be there. We just need to supply what our customers want.

Generative AI will be a powerful tool in helping us work that out.

This article was supplied by Antony Slumbers, Founder antonyslumbers.com who is delivering the keynote address at the Workplace Experience Summit 2024