When You Automate the Work, You Automate the Relationships
Automation Impact: Vendors promote efficiency, but ignore informal networks key to organizational health and function.
Human Element: Before automating tasks, understand the human interactions and mentorship that may be affected.
Replacement Strategy: Design new structures to replace lost coordination and mentorship before automation begins.
Monitor Effects: Track cross-functional communication, knowledge transfer, and trust to assess automation impact.
Identity Shift: Involve employees in defining new roles to manage identity disruption post-automation.
The efficiency math on automation is easy to do. Hours saved, headcount redirected, cycle time cut. Vendors will show you those numbers in a slide deck with a lot of green arrows and charts that go up and to the right.
What the slide deck doesn't capture is the network of informal relationships, coordination habits, mentorship channels, and professional identity anchors that routine work quietly sustains.
You can't measure what those are worth until they're gone, and most organizations don't know they're gone until the damage is already visible in rising attrition, eroding trust, and teams that are technically capable but struggling to function together.
Keep Reading—and Keep Leading Smarter
Create a free account to finish this piece and join a community of forward-thinking leaders unlocking tools, playbooks, and insights for thriving in the age of AI.
Have an account? Log In
Organizational tissue is a useful frame for this. It's the connective material between people, be it who they turn to when they're stuck, what shared processes give them regular touchpoints, how they build cross-functional awareness, where junior employees develop judgment in low-stakes settings before they need it in high-stakes ones.
Unlike formal org structures, tissue isn't documented anywhere. It lives inside the work itself.
When you automate the work, you automate the tissue too, unless you're intentional about what you're removing and deliberate about what you put in its place.
Leaders navigating this well aren't doing anything exotic. They're asking harder questions before they automate, mapping what they actually have before they remove it, and building replacement structures on purpose. Here's what that looks like in practice.
Ask What the Work Does
Most automation planning starts with a task list. The question is whether a given task can be done faster, cheaper, or more accurately by a machine. That's a reasonable filter, but it's the wrong starting point.
The better starting point is a work map. Before removing a workflow, understand what human judgment, coaching, escalation, and relationship-building is hiding inside it. The stated output of a task is rarely the whole picture.
Consider a recurring cross-functional report that three team members collaborate on weekly. The stated output is the report. The unstated outputs are the coordination conversation that happens while building it, the visibility each contributor gains into the other's priorities, and the regular rhythm that keeps relationships functional between those teams.
Automate the report and you've gained efficiency. You've also eliminated the only standing touchpoint between teams that now need to find other ways to stay aligned, and probably won't.
Before any significant automation decision, run a parallel analysis and answer the following questions:
- Who touches this work?
- What does touching it teach them?
- What coordination or relationship function does it serve?
This doesn't require an elaborate methodology. It requires asking the people doing the work what they tangibly get out of doing it, which is a conversation a lot of automation planning processes skip entirely.
The answer will sometimes be "this work is genuinely friction and everyone would be relieved to stop doing it." Useful to know.
More often, you'll find that even tedious tasks carry informal functions their owners have never articulated because nobody asked.
Map the Informal Architecture
Every organization has two org charts. The one that exists in your HRIS, and the one that exists in practice. The second one shows where decisions actually get made, who people go to with real problems, and where knowledge flows across team lines.
In most organizations, those two charts are substantially different, and the informal one is never documented anywhere.
Automation pressure tends to make the informal architecture visible in the worst possible way; by disrupting it.
One pattern that surfaces repeatedly is mentorship that lived inside coordination tasks. When a senior person stopped doing the thing, junior people lost the unstructured exposure that was preparing them for the next level. Nobody flagged it because the development was invisible, it was happening in the doing, not in any formal program.
That's the nature of informal architecture. It doesn't announce itself, which means it doesn't get protected.
Leaders who map the informal architecture before automating are less likely to be surprised by these disruptions. Organizational network analysis — the practice of examining patterns of communication and collaboration across teams — can surface where knowledge actually flows and which relationships are load-bearing for daily function.
It doesn't have to be a research initiative. Structured conversations with team leads about who they rely on, who relies on them, and what coordination mechanisms their teams depend on will surface the critical nodes.
The goal is to know what you have before you remove it. Once you've identified where informal tissue is concentrated, you can make deliberate decisions about what needs to be preserved, what needs to be redesigned, and what can be eliminated without consequence.
Design Replacement Structures Before You Automate
Where most organizations get the sequence wrong is they automate first, then discover that something is broken in team dynamics, try to figure out what happened and then address it. By that point, relationships have frayed, trust has eroded further, and the thing they're trying to rebuild is harder to build than it would have been to protect.
Replacement structures need to be designed alongside the automation, not after the damage is visible.
A common misconception is that dashboards and visibility tools fill the gap. They don't. Visibility isn't coordination. A dashboard can show you what's happening, but it it cannot recreate the informal check-ins and judgment calls that used to happen during manual work. Knowing the status of something is not the same as talking about it.
What a replacement structure looks like depends on what function you're replacing. If you're automating a task that served as a coordination mechanism, you need a different coordination mechanism such as an explicit touchpoint, a shared channel, a standing ritual that gives teams the regular contact the automated task previously provided.
If you're automating something that served as a mentorship channel, you need an explicit mentorship structure. That junior analyst who built analytical voice by contributing to the weekly report needs somewhere else to continue building it.
Informal mentorship depends on low-stakes repetition — regular, frequent opportunities to try things and get feedback without high visibility. If you can't replicate the conditions, you need to create new ones deliberately.
The test for any replacement structure is answering "does it restore the actual function the automated task was serving, or does it just look like a replacement on paper?" That requires follow-through and measurement.
Monitor Tissue Health
Organizations are reasonably good at measuring the outputs of automation. Time saved, error reduction, throughput, etc.
They are almost universally poor at measuring what automation costs organizationally. That imbalance in measurement produces an imbalance in decisions.
If the efficiency metrics look good but the tissue is fraying, you'll only see it when attrition spikes, collaboration breaks down, or a team that used to function well starts struggling. By then, months or years of degradation have already happened.
A few things worth building into regular reporting include:
- Cross-functional communication patterns, particularly whether quality and frequency hold up following major automation changes
- Knowledge transfer rates to junior employees
- Trust levels, measured through regular surveys that ask in-depth questions about the work environment rather than just satisfaction scores
- Informal network density through periodic organizational network analysis
None of this is technically difficult. Most organizations aren't measuring it not because they can't, but because they've never connected automation decisions to organizational health outcomes as a matter of practice.
The measurement cadence matters as much as the metrics themselves. An annual engagement survey after a major automation initiative will not tell you what happened or when. Quarterly pulse checks aligned to significant automation milestones can, if done well.
Handle the Identity Dimension Separately
The hardest part of this isn't processes. It's people.
Work is not just what people do. For most people, work is substantially how they understand who they are professionally, what they're good at, and where they stand relative to their peers. When you automate the tasks that housed those competencies, the loss is not only practical.
The Edelman Trust Barometer has tracked rising anxiety about AI's impact on work for several years. The 2026 edition offers the starkest read yet: more than half of low-income employees and nearly half of middle-income employees believe generative AI will leave them behind, not benefit them.
That's the backdrop against which every automation announcement lands.
The timing makes this harder to manage than most leaders realize. Identity disruption rarely shows up at the announcement. It tends to surface six to twelve months later, and when it does, it looks like disengagement.
Leaders read it as a culture problem and treat it accordingly, missing what it actually is, a delayed response to a shift in professional identity that was never addressed directly.
Leaders managing this dimension well are involving people in defining what their new work is, not just announcing what it will be. The difference matters more than most efficiency-focused planning processes account for.
When employees participate in designing their new responsibilities, they develop ownership over the transition. When the transition is handed to them as a finished fact, even when the new work is genuinely better, the message received is about who controls the terms of their professional life.
This is a design challenge rather than a communications issue. Determining what higher-value work looks like for a given role or team should include the people in those roles, at a stage early enough that their input shapes the outcome. That requires a longer planning horizon than most automation deployments currently allow, and building that time in is a leadership decision.
Handling the identity dimension is not efficient nor is it often simple work. The reason to do it is because the alternative — a workforce that goes through the motions of a transformation it doesn't believe in — costs more over time than any efficiency gain justifies.