In the past couple of weeks, we have seen some interesting developments in AI with the potential to improve both personal and organisational productivity; but the former category seems to be moving a lot faster than the latter.
On the personal productivity side, Claude’s announcement of their new ‘computer use’ capability does for personal computing what Robot Process Automation (RPA) did for enterprise automation, but with more intelligence and adaptability. Essentially, you can give Claude access to remotely operate your computer, and it will perform tasks on your behalf. It seems to be a fascinating development that shows the vision Anthropic have for the positive impact of AI on our lives, set out in their CEO’s recent article Machines of Loving Grace.
Azeem Azhar wrote about his experiences using it and concluded that we are all potentially managers now, with an AI intern in our pocket:
Testing it out myself, I gave Claude a task to collect financial information on 10 companies. It managed to complete the task in 15 minutes… It used Firefox and LinkedIn to get the info and then failed to put it in a spreadsheet, so instead, it saved it to a text file.
The process was impressive. You can try it out here (if you know a bit of code and have an Anthropic API key). I would think it would take an intern 1-2 hours to complete this task – it takes a while for a person to orient themselves around a task and a good hire would double or triple-check the numbers (which I wish Claude did). It cost me $4 to do, the equivalent of paying an intern $2-4 an hour for their work.
He also referenced a useful categorisation of AIs into Gods (AGI etc), Interns (co-pilots, etc) and Cogs (narrow agents, automations, etc). So, leaving aside debates about the desirability or feasibility of creating AGI Gods, where are we with the many cogs that are needed for organisational productivity?
Enterprise use cases advancing less quickly than personal productivity
Microsoft hit back last week against Salesforce’s launch of their own agentic AI framework and criticism of co-pilots with the announcement of 10 new agents focused mostly on a similar area of customer service and sales automation. It will be interesting to watch both companies’ efforts in this space, but compared to personal usage, agentic AI in the enterprise has a much bigger readiness challenge ahead of it.
Constellation Research shared an overview of the challenge recently, arguing that until companies optimise their processes, there are limits to how we can apply AI to orchestrate and connect them:
“Agentic AI is a hot topic in enterprise technology, but without process automation and orchestration the vision is unlikely to be realized. CxOs sifting through the marketing hype of agentic AI should keep process optimization and orchestration in the forefront of planning.”
This might mean that existing vendors in the enterprise process optimisation space who are investing heavily in AI – ServiceNow, Celonis, UIpath, etc. – have a key role to play, as well as bigger players like Microsoft and Salesforce.
Secret cyborgs and new hybrid AI roles
Are we seeing a mismatch between personal and organisational adoption? Perhaps we will end up with organisations that look like fantasy sci-fi movies, where the built environment and rituals seem ancient, but people have amazing tools and technology to do incredible things nonetheless.
Scene from the movie Brazil (Terry Gilliam, 1985)
Wilko Wolters shared a useful roundup on AI adoption, which found more evidence for Ethan Mollick’s idea of “secret cyborgs” – people in knowledge-intensive roles who are using AI personally to get their work done faster, but are shy about it in case it changes management expectations or crosses a line in terms of AI usage rules in their organisations.
“The AI transformation in the workplace is in full swing, but its full potential remains largely untapped at the enterprise level. By recognizing the prevalence of “Secret Cyborgs,” bridging the gap between individual and organizational gains, and fostering a culture of open AI innovation, companies can position themselves at the forefront of this transformative wave.”
Also on the subject of AI adoption, Jabra published a research report that summarised its view of AI adoption, which suggested voice will be the principal gateway to AI (which is lucky, since they sell professional headsets!), but also found that AI decision makers in companies they talked to were not the usual CIO/CDO suspects, but came from more hybrid backgrounds:
“A significant proportion of AI decision makers are young (aged less than 39), do not come from IT backgrounds or technology industries. Almost 80% are not in the C-Suite. This is a job that requires broader alignment and buy-in, and yet it appears that those making the decisions on AI implementation in organizations do not come from the same traditional departments tasked with other technology decisions.”
Collaboration and Collective Intelligence
Another area that is both an AI readiness challenge and also one of its biggest use case opportunities for organisational productivity is what we used to call knowledge management. Much of the work that AI interns and cogs are expected to do will rely heavily on a company’s knowledge stocks and flows that contribute to its collective intelligence.
As we have written about previously, AI has huge potential to augment human collaboration and help solve the coordination of work in a much smarter way than manual management.
Prof. Christoph Riedl wrote about this opportunity area for AI in HBR recently, and identified three areas that leaders should focus on in looking for proactive AI use cases:
“Recent research suggests that collective intelligence emerges from three interdependent ingredients: collective memory, collective attention, and collective reasoning. Managers can apply this idea to target specific areas in which AI can elevate the organization’s collective cognitive abilities and drive more informed decision-making in ways that are human centered and amplify human creativity.”
For about 20 years, our team has worked primarily in wikis rather than document or slide decks, and between our old Confluence wikis and our current Notion wiki, we have a comprehensive corporate memory that AIs can mine to recreate decisions, analyse projects we have run and summarise our ideas. Everything can be cross-linked and cross-referenced to create meta-level knowledge from the individual pages. Everything is searchable. Everything can be updated, commented upon or given health warnings if out of date when accessed later. I find it puzzling why this had never caught on as the dominant form of knowledge creation. You start with content and only create the structure it needs as you go along, rather than shoe-horning everything into standard structures or formats.
Teams that have worked this way over the long-term have a natural advantage when it comes to training LLMs or AI agents on their own content. First, it is much easier than mining emails and document stores. Second, the content has a better signal to noise ratio because it is raw and real, rather than polished into corporate-speak for ‘distribution’.
Atlassian, the makers of Confluence, seem to be developing their AI strategy in the direction of augmenting workforce collaboration. Notion have an AI offering already in their platform. But Microsoft has always seemed allergic to wiki-based collaboration and even when they shamelessly copied Notion with their own version called Loop, they seem reluctant to embrace it as a more collaborative alternative to Sharepoint.
But as an area of rich AI use case potential, cultivating, connecting and using the collective intelligence of an organisation seems both more interesting and valuable than some of the basic process automation cogs that are the focus of much enterprise AI work today.
If you would like to discuss organisational use cases for AI, or run a learning session on digital orgs & leadership, please get in touch…