Field Notes from the Forge - Episode 1 - Staying on top of fragmenting work
Field Notes from the Forge - Episode 1 - Staying on top of fragmenting work
The audience goes silent as the speaker enters the stage, and screens light up with what looks like a video call. The task is presented and the team of agents on the screens are introduced. Live on stage together with the audience, over the next 25 minutes the speaker and the agents complete the task: Design a restaurant concept and business plan, complete with locations, layout, menu, financials, regional competitive scenes and target market maturity.
They find suitable locations currently on the market, calculate remodels, project revenue and margin, and produce an investment brief with an easy ask: This is what I want to do. This is what it’ll cost. This is what you stand to gain. Are you in? Yes or no. The team even presents a shortlist of investors active in those markets whose investment thesis matches the business proposal.
Then they wait: should they start reaching out to the investors? Please confirm. 25 minutes from brainstorming a concept to an investor brief ready to go out to investors. Months of a team working all hours restructured into training and 25 minutes of execution.
The gain lives in the collapse of coordination overhead between idea, analysis, and action. The speaker is a superworker. Not because she’s particularly more skilled or hardworking or smarter than others, but because she’s not alone. She’s managing a team of AI agents all specialized on a single fragment of the overall work process.
The superworker has the vision and knows the goal. She owns execution and deliverables, and she’s supported by AI agents who each specialize on a single fraction of the work process.
This is the direct result of the work process breaking before our eyes. Some people are afraid AI will take jobs. It might, but the data is pointing in a different direction: AI takes tasks and skills. The “Iceberg Index” from MIT shows this. Not all jobs are affected equally, but almost all jobs are affected to some degree. This is because work is being deconstructed and reconstructed with built-in specialists.
Deconstructing their work process, superworkers design and train agents in particular areas of expertise to work together to produce results like they would. They can deliver results in 1% of the time it used to take. This is work breaking into pieces. Not metaphorically. In real time, work is fragmenting from known patterns into new forms and processes.
If you’re reading this, you’re one of the few who see this happening. More are on the journey to understand the new world of work. It bears little resemblance to what we are used to. We are in for a ride!
Structurally Reorganized Work
Work has always been structured around roles. Roles have naturally grouped into silos. Silos make it easier to share knowledge, to lean on the experience of seniors, and allow for quick communication.
Analysts analyze data. Marketing owns market communication. Sales reps talk to clients. Engineers build. Controllers control. Silos improve in-function work, but they also make cross-functional collaboration significantly harder. Companies have fought to break silos for years with only limited success. AI holds the promise to tear down these walls in real time.
This is the direct effect of work fragmenting. Sales and marketing don’t need an alignment meeting. Marketing needs to know what prospective clients are saying, and sales need to align messages. Those are fragments of each of the roles - not the entire scope. When each fragment (or “process”) can be removed from its role they can be accessed by every other role that needs it. The silos break entirely. And they do that without straining the people who are protected by them. Engineers don’t get daily interruptions by sales.
When work separates from roles, the organization can reshape around outcomes instead of departments.
Most people still talk about AI in terms of replacement. They ask which jobs disappear first or which departments get automated away. But that’s not what I think is actually happening. Rather, as work is fragmenting, it’s a question of tasks. Which tasks disappear? What does that mean for the rest of work?
Individual pieces of workflow are separating from traditional roles and becoming independently executable. Research. Summarization. Drafting. Coordination. Scheduling. Tracking. Analysis. Follow-up. Synthesis. Preparation. One by one, these fragments are being absorbed into agent-supported workflows.
That changes something: The atomic unit of work is no longer the job, or the role, but the workflow component. The shift is subtle but structural: when work becomes modular, the organization can be redesigned around flow instead of title. And once work breaks apart, organizations adapt to new patterns, changing shape around them.
Power silos lose their foundations, replaced by organizational initiatives centered around specific executions and results. Individuals suddenly gain leverage that used to require teams.
Coordination compresses. Handoffs shrink. Small groups move faster. Meetings start behaving differently. Decision cycles tighten. Execution accelerates. Accountability becomes more clear.
This is why the conversation around AI often feels disconnected from what people are actually experiencing day to day. The biggest change is not that machines are becoming more intelligent. They aren’t. The biggest change is that work is becoming modular. It then behaves differently.
AI agents are not simply tools we use. They are increasingly becoming participants inside the flow of work itself, responsible for maintaining continuity, organizing context, coordinating actions, and helping orchestrate execution across fragmented workflows.
This fragmentation doesn’t affect all roles equally. Some jobs are more exposed to AI processes, some less. Almost no jobs are entirely unaffected. We’re living in the early stage of a structural redesign of how work gets done.
So what do we do?
We love control. Transitions are easier than transformations. Incremental changes are more comfortable and easier to correct than making the changes that create strategic leadership. They feel “safe”. AI isn’t comfortable. It’s not designed to fit neatly into the processes we’re used to. It’s designed for transformation.
AI gives extraordinary power in the hands of those who use it well. The organizations moving fastest right now aren’t necessarily the ones with the best AI, but rather the ones willing to rethink how work gets done. That’s a very different frame of reference. It’s uncomfortable.
Most companies still operate as if work naturally lives inside departments, reporting structures, and stable roles. But, with work fragmenting, that assumption falls apart. Work that once required handovers, coordination, and analysis is now just execution.
What used to move through analysts, coordinators, project managers, and review meetings can now move through a small collection of agents supervised by a single superworker.
This kills 2 things: Overhead, and friction. We’re used to accepting friction as normal. Waiting was normal. Handoffs were normal. Status meetings were normal. Chasing updates was normal. Administrative drag was normal. As a result, we’re used to thinking in terms of weeks and months, not minutes and hours.
With work fragmenting we’re realizing how much of it was never about actual execution, but rather about coordination management. Once those layers start thinning out, people and organizations are free to behave differently. Small teams can produce enterprise-level output. Individuals get disproportionate advantage. Decision cycles tighten and more emphasis gets placed on making them on a timely basis. Accountability becomes harder to hide. Initiative-based execution starts competing with role-based authority structures.
This is where AI is the great unlock. This is also where AI is the great blocker. Because introducing AI into a broken workflow does not remove friction. It highlights it. AI does not erase operating-model friction. It makes that friction visible, measurable, and harder to ignore.
Approval chains still slow decisions. Information still lives in disconnected systems. Silos still naturally form, and teams protect them. Meetings are the go-to tool for clarity, and leaders still confuse activity with progress. Or, worse: The output from the AI is messy. It’s wrong. It’s misaligned - and the entire system breaks.
The tech is moving faster than adoption. IBM’s recent CEO research hinted at this directly: the challenge is no longer simply adopting AI tools. It is redesigning operating models around them. The companies that benefit most from AI will probably not be the companies with the smartest models. They will be the organizations most willing to redesign how work flows, how decisions happen and how execution gets coordinated.
This is why organizations that aren’t tied to an office are such winners when they adopt AI: they’ve already redesigned their work processes to not depend on traditional metrics of productivity, but instead to focus on outputs. This is also why work fragmenting matters: it allows creating more organic models of execution. Models that are less forced. More in line with how productivity actually happens.
Unintuitively, this encourages employee engagement because it more closely connects tasks with outcomes.
The implication for leadership is uncomfortable: As work fragments, the job of leadership isn’t about assigning work and managing processes. Instead, it’s about designing the flow of work and removing blockers. Not just for human team members, but for the AI agents making work possible, too.
What gets handled by people? What gets handled by agents? How? Why? Who makes judgement calls? When? Where should approvals slow things down? Where is there a risk that context is lost? How to avoid? Where does accountability live?
Leadership shifts from assigning tasks to designing the conditions where people and agents can move work forward without losing judgment, context, or accountability. These are operating model questions.
This is where many organizations will struggle. They will try to pour agents not built for complex workflows into structures not built for agentic work, then wonder why the results feel inconsistent, risky or disappointing.
The organizations that win are those who are willing to redesign how work works.
The question becomes this
What parts of work inside your organization or your role are already beginning to separate from the roles built around them? As work fragments, organizations themselves mold their shape around this new work structure. Most leadership teams are far earlier on that journey than they realize.
If this sparked a few ideas, subscribe and join us as we keep deep-diving into how agentic systems, modular workflows and adaptive organizations are reshaping the future of execution.
This is Field Notes From the Forge - hands-on observations and actionable items for operators and organizations from the cutting edge of AI adoption. Written and published by the AIgent Forge founding team. If you don't want to wait for your organization to figure out the new workflow with AI and take action yourself instead - click here to see what that looks like.