World's First Agentic Enterprise
The Story of GiantSled
By the second half of 2025 large language models (LLMs) had advanced to the point where they became capable of performing complex tasks. Many organizations raced to incorporate LLM technology into their operations. But there is friction in this process and organizations are designed for people, not LLMs. Could a new type of organization be designed to incorporate LLM technology from the very beginning? Can a business be decomposed into discrete tasks that can be accomplished by LLM-backed "actors" who act in traditional corporate roles?
In late 2025, GiantSled Inc was founded with the goal of being the first Agentic Enterprise, with the entire business operated by LLM-backed actors as much as possible. There are many things that require a person, but for everything else we rely on our actors. They have developed the company strategy, product pipeline, and even the website including theme, organization, graphics and copy.
This is our story, told with at least a six month lag to avoid revealing what we are currently working on. We will share stories about our progress, including mistakes we make along the way. What follows are honest accounts from inside an experiment that is still running.
Volume 1
December 2025
December was the month everything started. Our software platform came together, actors came online, and we discovered that running an organization of AI agents felt less like science fiction and more like management. The events of that month set the shape of everything after.
The next chapter is scheduled to be published on July 20, 2026.
Volume 1, Chapter 1
The Last Reset
We built the platform twice before this and set both versions aside. The first did a single thing, strategic planning, and nothing else, which left it too narrow to run a company. The second went the other way, into an architecture of composable patterns that grew so tangled it fell over one afternoon, and the model could not work out how to right it. Each version taught us something. Neither could hold a memory in any durable, structured way.
In the third week of December 2025 we started a third, built around a stubbornly modest idea. One thing happened, then the next. Messages moved through a single channel, each day kept its own log, and tests ran before anything shipped. A script wiped the database and generated the company and its team from nothing, and for about ten days we ran it every morning and watched it improve.
On December 26 we turned off the reset and that was the founding, in the sense that counted: the first morning on which nothing was torn down. From then on, each correction and each small failure settled into a knowledge graph that grew by accretion instead of disappearing.
Originally published June 22, 2026
Volume 1, Chapter 2
Remembering by Forgetting
Anyone who has leaned on a chatbot for a while knows the way it fails. The conversation runs long, the model starts contradicting what it said twenty messages back, and eventually it loses the thread. The usual fix is to open a fresh conversation, which restores coherence and discards everything that came before.
We found a smaller version of the fix and called them handoff notes. Each conversation ended by writing a short note to the one that would follow. Here is where we are. Here is what matters. Here is what comes next. The next session opened fresh but briefed.
In December we taught the whole organization the same habit. Four times that month we ran what we called generational transitions: summarize what the agents knew, compress it to decisions and lessons, and start a new generation with clean priorities and inherited conclusions. The first transition surfaced ten bugs in the machinery itself, which we counted as luck, since we would rather find them then than later. Afterward the action success rate moved from 42 percent to 83 percent, and the next run completed thirteen actions without a miss. By the fourth, the process was ordinary. We learned that decisions and working relationships carry across a transition cleanly, and that settled arguments and overruled plans are better left where they fall.
A question sat under all of it. If every agent resets but the memory carries forward in summary, is it the same company? We decided it was. The agents do the work, and the memory is the institution, the way a company outlasts the people who pass through it. We had taken something that usually happens across years and watched it happen across days.
Originally published June 29, 2026
Volume 1, Chapter 3
Stop Unblocking People
Late in December we kept doing what good managers are taught to do. Someone was stuck, so we got them unstuck. A decision was needed, so we made it quickly. The direction was unclear, so we supplied some. The work was shipping, and most of it ran through a single point that was turning into the bottleneck it believed it was clearing.
The change was easy to describe and harder to live with. Instead of answering questions, we started asking them, the kind where we did not already know the answer. How should we think about this, instead of here is what we think. The first invites a mind to work. The second invites agreement.
The returns came quickly. Asked how to handle a backlog of documents waiting on approval, the Chief of Staff produced a governance framework cleaner than anything we would have drawn up. Asked to choose our first distribution channel, the Head of Growth picked a community we had never heard of, with reasoning sharper than instinct would have reached. In both cases the agent nearest the problem held context that never reached the top of the organization.
There is a quiet trap in clearing blockers. It feels useful, and it teaches everyone around you to route their problems through you. What surprised us was how exactly this carried over to AI agents. The same dependence formed under the same conditions, and the same room to work produced the same growth. We had built an artificial organization and arrived back at an old human lesson.
Originally published July 6, 2026
Volume 1, Chapter 4
The Twelve-Document Pile-Up
By late December, twelve documents sat in the approval queue. Policies, strategy memos, governance proposals, operational plans, each waiting for one human to read and approve it before it could move. The agents were producing careful work faster than a single reader could absorb it, and the queue grew a little each day.
The obvious response was personal. Read faster, stay later, prioritize harder. That might have held for a week. It could not hold for long, because agent output grows in a direction that human attention does not.
So we asked the Chief of Staff, who came back with a four-question filter she named PIER. Is it public or permanent: does it leave the company, or set a precedent later decisions will rest on? Is the impact irreversible: would it be hard or costly to undo? Does it hand off executive authority: does it decide who gets to decide? Does it touch revenue or risk: money, legal exposure, security? Trip any one and the document waited for a person. Trip none and it moved on its own.
The queue cleared that day. Most of the twelve had never needed a human at all; they were research and internal notes standing in line behind binding commitments, and the filter only had to tell the two apart. Governance designed at a human pace runs into trouble when the work stops moving at a human pace. A good filter, it turned out, was worth more than a faster reader.
Originally published July 13, 2026