Only 20-30% of corporate knowledge is documented and retrievable in IT and product organizations. At Steady, we call this hard data–information like activity from project management tools, documentation systems, or DevOps platforms. Hard data consists mostly of lagging indicators used to coordinate work among teams and individuals.
We call the other 70-80% soft data. Soft data consists of intentions, risks, challenges, discernment, judgments, and domain expertise. Soft data is stored in people’s heads. Soft data mostly represents leading indicators.
The manual assembly required of hard and soft data for hybrid, remote, and distributed teams to operate is why coordination is so difficult and damaging to productivity. Teams need both at all times to make progress.
Soft data is particularly important in coordination, not only because there is more of it, but also because it contains intention and therefore enables proactive management and leadership. The best time to course correct as a manager? When team members declare what they plan to do is before they do it, not after. For leaders, it’s also an opportunity to understand if goals and missions are understood and aligned, again, before the work is started.
Studies show that 60% of the average tech worker’s time is spent on this manual context treasure hunt: performing “work about work” instead of actually doing the work. The Wall Street Journal and others, including us, call this the coordination tax.
Not only does the coordination tax negatively affect productivity, but it also affects work quality and team engagement. In other words, people simply don’t have the time or the context to make the right decisions, and the beauraucratic overhead of assembling knowledge and coordinating with others is exhausting.
The coordination tax is a $100B+ problem right now. Meanwhile the market is teeming with AI toolkits that promise to address this issue as a side effect of conquering AI-powered enterprise search, business intelligence, and DIY workflow automation built on top of that. These automations are often labeled (and mislabeled) as “agents.”
Glean is perhaps the most interesting relative newcomer, but many platform incumbents are claiming to deliver AI-search automation as agents, including ServiceNow, Microsoft, and Salesforce.
The problem, as we’ve discovered as we talk to our customers, is these offerings only operate on the hard data within their ecosystems, largely ignore human input and behavior patterns, and, therefore, do not truly target the coordination tax. Ungenerously, they are solutions looking for problems, but broadly, they are addressing business intelligence writ large and then leaving the automation implementation up to the customer (the DIY part). They are toolkits for gathering hard data, not a tool for solving the coordination problem with soft data.
Our unique value proposition is that not only have we cracked the code for a process to collect the soft data–Continuous Coordination–but that our AI is operating on it to synthesize it for our customers in the backround on a schedule, so they can focus elsewhere. This approach directly and demonstrably eliminates the coordination tax and unlocks speed-to-impact. We know this because we create and validate coordination “scorecards” with our customers.
So, in 2025, this is where we are doubling down on our product, and we’ve just scratched the surface. We are building true autonomous agents as a tool to solve for coordination specifically: crossing multiple domains of relevant hard and soft data, using a proven process on behalf of our users while they are off doing real work.
