Mar 2026 · Lab Notes
A workflow should do more than move tasks. The better ones help a team notice, prioritize, remember, and decide with less friction.
Idea
Most workflow design still assumes work is mainly about steps: receive, assign, complete, report. But in real teams, a lot of the difficulty lives between those steps. Context gets lost, urgency gets misread, and decisions get delayed because no one has a clean view of what matters now.
Research on cognitive load in organizations shows that mental fatigue from managing information — not the work itself — is the primary bottleneck for most knowledge workers. The term "brain fry" has entered organizational research to describe the cumulative effect of too many decisions, too much context switching, and too little clarity about priorities.
A smarter workflow does not replace judgment. It supports it. It surfaces what changed, what needs attention, what is still unclear, and what can move next. The goal is not to automate decisions but to make the information needed for good decisions available at the right time with the right context.
The gap
A common complaint about project management and workflow tools is that they create overhead instead of reducing it. Team members spend time updating status fields, moving cards between columns, and filling out forms — work that tracks progress but does not advance it.
This happens because most workflow tools are designed around visibility (making status clear to managers) rather than flow (helping individual contributors move work forward). The result is a system that serves reporting needs but adds friction to the people doing the actual work.
AI-native workflows can address this by automating the overhead layer. Status can be inferred from actions — if a pull request was merged, the task is done. If a client sent feedback, the review stage is active. If no response has been received for three days, the status is "waiting." The system tracks progress without requiring manual updates.
This frees the workflow to serve its more valuable function: helping the team think about what matters next. Instead of spending time updating a tool, team members interact with a system that tells them what changed, what is at risk, and what decision is needed — and provides enough context to act.
Approach
The most effective workflow designs follow a principle borrowed from cognitive science: reduce extraneous load, manage intrinsic load, and optimize germane load. In practical terms, this means removing unnecessary steps (extraneous), presenting complex information in manageable chunks (intrinsic), and ensuring the information supports meaningful thinking (germane).
Applied to workflow design, this looks like: summaries instead of full logs, exceptions instead of comprehensive reports, action prompts instead of status overviews. The system curates what the team sees, not to hide information, but to ensure the most important signals are not buried in noise.
Organizations that adopt the "thinnest viable platform" approach — creating systems with just enough capability to reduce cognitive burden without adding unnecessary complexity — tend to see better adoption and better outcomes. The temptation to build feature-rich workflows is strong, but features that nobody uses are worse than missing features — they add noise.
The practical test: after implementing a workflow improvement, does the team spend less time on coordination? Do fewer things fall through cracks? Can someone new understand the flow in five minutes? If the answers are no, the workflow is adding complexity rather than reducing it.
In practice
The most successful AI-enhanced workflows treat AI as an augmentation of human capacity, not a replacement for it. The AI handles the work that humans find draining — scanning long threads for key points, tracking deadlines across multiple projects, identifying patterns in incoming requests — while humans handle the work that requires judgment, empathy, and strategic thinking.
This division of labor is not just philosophical — it is practical. Research consistently shows that cognitive load is reduced when teams share the responsibility of managing information with AI tools. The human does not need to remember everything; the system surfaces what is relevant. But the human makes the call. This "think first, act together" pattern produces better decisions with less fatigue than either fully manual or fully automated approaches.
Design note
Summaries that compress status into actionable insights. Context-aware reminders that include why something matters. Exception surfacing instead of comprehensive reporting. Action prompts that reduce the gap between knowing and doing. Automatic status inference from actual work activities.
More steps, more forms, more notifications, and more places to check without clearer thinking. Systems that serve reporting needs but add overhead for contributors. Feature-rich workflows that nobody fully uses. Manual status updates that duplicate effort. Tools that track progress without helping advance it.
