The question ThreadWeave forces every organization to answer.
In 1995, a man named William Redmond Jr. resigned from PepsiCo to take a job at Quaker Oats, which owned Gatorade, Pepsi's most bitter competitor in the sports-drink market. Redmond had been PepsiCo's general manager for the entire California business region. He knew everything: pricing strategies, distribution plans, the upcoming marketing push for a new product called "Pepsi Max." He hadn't signed a non-compete. He hadn't stolen any documents. He just knew.
PepsiCo sued. They didn't claim Redmond had taken anything. They claimed he would, inevitably, unavoidably, simply by doing his new job. A federal court agreed. Redmond was enjoined from working for Quaker for six months. The court ruled that his knowledge was so specific, so deeply intertwined with PepsiCo's trade secrets, that he couldn't possibly do his new job without using it.
The inevitable disclosure doctrine was born. And with it, a question that had been simmering in employment law for decades finally boiled to the surface: where does the company's knowledge end and the employee's knowledge begin?
The law, for all its complexity, draws a surprisingly simple line:
Explicit knowledge, meaning documents, code, designs, formulas, emails, and reports, belongs to the employer. The work-for-hire doctrine is unambiguous. If you wrote it on company time with company resources, the company owns it.
General knowledge, skill, and experience belongs to the employee: the judgment, the instincts, the pattern recognition, the thing that makes a senior engineer worth five juniors. You can't alienate someone from their own knowing. Courts have affirmed this principle for over a century: an employee has "the right to utilize his general skill, knowledge and experience" when moving between jobs.
The problem is that the line between the two is a fiction. Not a blurry line, but a fiction.
Consider a senior engineer who has spent 15 years diagnosing failures in complex machinery. When a particular vibration pattern appears on a diagnostic readout, she knows, immediately and without conscious reasoning, that a specific bearing is about to fail. This knowledge was developed on company time, using company equipment, solving company problems. The company paid for every hour she spent building this instinct. But the knowledge itself? It lives in her head. It goes where she goes. And it's precisely this kind of knowledge that makes her valuable to her current employer, and to the next one.
Law professor Camilla Hrdy calls this "the general knowledge, skill, and experience paradox." The law says employees can take their general knowledge with them. But the law also says trade secrets, which are specific, valuable, and not generally known, belong to the employer. And in knowledge-intensive industries, these two categories collapse into each other. The specific becomes general over time. The general was built from the specific. You can't separate them without destroying the thing itself.
ThreadWeave changes the calculus. Radically.
In the pre-ThreadWeave world, the organization relied on Lars. Lars was the institutional memory. When someone needed to know why the Oslo office used a different engagement letter structure than Copenhagen, they asked Lars. When a client mentioned a precedent from 2011 and three senior people nodded along without actually remembering the details, Lars remembered. The knowledge was in the organization, accessible, valuable, used daily, but it was not owned by the organization in any meaningful sense. It was on loan.
ThreadWeave changes the custody arrangement. It connects to the systems people already use, including email, chat, documents, and meeting transcripts, and builds a knowledge graph passively, in the background. Every decision, every rationale, every "here's why we do it this way" gets captured, classified, and woven into a structure that persists after the person who said it leaves.
This is the explicit promise of the system. It's why organizations need it. But it also raises a question the literature on knowledge management has been wrestling with for thirty years:
If the organization now possesses Lars's knowledge, does it own it?
Knowledge management scholars Baskerville and Dulipovici (2006) identified two competing ethical frameworks that organizations bring to this question:
Knowledge created during employment, using organizational resources, in service of organizational goals, is organizational property. The employment contract, whether explicit or implicit, transfers ownership. Lars was paid for his time. His knowledge output, like his code output or his document output, belongs to the firm.
This is the theory that underlies most corporate KM initiatives. It's clean, legally defensible, and aligns with how organizations already think about every other asset they possess. Buildings, patents, client lists, brand equity, all owned. Why should knowledge be different?
Knowledge is inseparable from the knower. It's not a thing you can transfer like a file. It is, in the words of Nonaka and Takeuchi, part of the "collective self," the accumulated experience, judgment, and identity that makes a person who they are. To treat knowledge as organizational property is to treat the employee as a means, not an end. It's alienation in the literal sense.
Under this theory, ThreadWeave's passive capture, its "no forms to fill in" promise, is not a feature. It's the problem. The system extracts something deeply personal without the knower's active participation, let alone consent. It transforms a human relationship (ask Lars) into a database query (search ThreadWeave), and in doing so, it fundamentally changes the power relationship between the individual and the organization.
Davenport and Prusak, in their foundational 1998 book Working Knowledge, described organizational knowledge sharing as a market. Knowledge owners are sellers. Knowledge seekers are buyers. The currency is reciprocity, reputation, and altruism. And like any market, this one has pathologies.
The most stubborn pathology is the one that matters most here: monopoly power. When an employee possesses critical knowledge that nobody else has, that knowledge is their source of status, job security, and influence. They are, in market terms, a monopolist. And monopolists don't compete. They extract rents.
A substantial body of recent research (2019–2024) has confirmed what every manager has suspected: knowledge-based psychological ownership, the feeling that "this knowledge is mine," leads directly to knowledge hiding. The mechanism is straightforward: psychological ownership → fear of losing knowledge power → hiding behavior. People who perceive their knowledge as the source of their value will actively resist its systematization, even when the organization has every legal right to it.
This isn't irrational. In organizations that conducted layoffs last quarter, or that treat senior employees as interchangeable, or that have a history of extracting knowledge and discarding the knower, hiding is perfectly rational self-preservation. As Anand et al. (2021) argue, what managers call "hoarding" is often a reasonable response to a low-trust environment.
ThreadWeave bypasses this entire dynamic. It doesn't ask Lars to share. It doesn't require him to document anything. It watches the information flow through the organization and builds the graph passively. Lars doesn't have to participate. He doesn't even have to know.
This solves the market pathology: knowledge gets captured regardless of the monopolist's willingness to sell. But it creates an ethical pathology in its place. The system appropriates without consent. It dissolves monopoly power by making the monopolist irrelevant, not by making them willing.
Michel Foucault spent much of his career arguing that power and knowledge are inseparable, a relationship he called pouvoir-savoir, power/knowledge. You can't have one without the other. Knowledge isn't a neutral substance that sits in databases waiting to be retrieved. It's always entangled with who gets to define what counts as knowledge, who controls access to it, and who benefits from its use.
KM researchers have been slow to engage with this. Heizmann and Olsson (2015) argue in "Power Matters" that the knowledge management field has "systematically avoided the power dimension," treating knowledge as a resource to be optimized rather than a site of political struggle.
ThreadWeave is, from a Foucauldian perspective, a technology of power. It reorganizes who knows what. It transforms personal, contextual, situated knowledge into organizational, standardized, searchable data. It makes knowledge visible to the organization in a way it wasn't before, and visibility, in Foucault's framework, is the first mechanism of control.
Consider what happens when a manager can search ThreadWeave and see, for the first time, exactly which teams are making decisions, who's influencing whom, where the bottlenecks in organizational knowledge flow actually are. This is valuable information. It's also surveillance. And the people being surveilled, the ones whose email threads are being classified as ANSWER or DECISION, whose meeting transcripts are being ingested, whose private expertise is being externalized, may not welcome the transparency.
Workplace monitoring has become ubiquitous: keystroke logging, screen capture, email scanning, productivity scores. Knowledge management systems that ingest passively sit somewhere on this spectrum, and the ethical questions are similar.
A 2024 study on knowledge workers' perceptions of AI-mediated communication found that privacy concerns were not about the technology itself but about the power asymmetry it created. "This might sound like I'm wearing a tinfoil hat, but..." was how participants expressed their unease. The fear wasn't that the AI would do something malicious. It was that information they considered personal, their communication style, their decision-making patterns, their professional judgment, was being captured and systematized without their meaningful consent, and they had no visibility into how it was being used.
ThreadWeave is on-prem. Your data never leaves. The PII gate strips emails, phone numbers, and personal identifiers before anything reaches the knowledge store. These are genuine privacy protections, and they matter. But they address data privacy, the question of whether sensitive personal information leaks out. They don't address knowledge sovereignty, the question of whether an organization should be able to capture, classify, and retain someone's professional judgment without their active participation.
This article has spent a lot of time on the tension. That's deliberate: the tension is real, and pretending it isn't would be dishonest. But it's also worth being clear about what ThreadWeave actually does, and doesn't do:
ThreadWeave does not extract knowledge from people's heads. It captures what they already chose to communicate, in emails, in chat messages, in documents, in meeting transcripts. The "tacit knowledge" it captures is tacit to the organization, not to the individual. Lars knew what he was saying when he explained the Stockholm-Copenhagen difference in a Teams thread. He just didn't know that was the only place the explanation existed.
ThreadWeave does not make Lars replaceable. It makes the organization resilient to Lars's absence. Lars still has thirty years of pattern recognition that no database can replicate. What ThreadWeave preserves is the record of Lars's contributions, the decisions, the rationales, the context, so that the next person doesn't start from zero. Lars is still the expert. ThreadWeave just means the organization doesn't collapse when he retires.
ThreadWeave's design respects a distinction the law already makes. The law says explicit communications, emails, documents, messages created during employment, belong to the employer. ThreadWeave ingests these. The law says general knowledge, skill, and experience belong to the individual. ThreadWeave doesn't touch these, because it can't. Nobody's building a database of instincts.
ThreadWeave is on-prem, open-source, and MIT-licensed. This matters for the custody question. The knowledge doesn't leave the organization's walls. It doesn't sit in a vendor's cloud. It can't be held hostage by a pricing change or an acquisition. The organization that owns the knowledge also owns the infrastructure that stores it, and the MIT license means the code stays free to use and modify, forever.
The literature on knowledge ownership converges on a single unresolved tension: knowledge is both personal and organizational at the same time. It is created by individuals, using organizational resources, in service of organizational goals. It is inseparable from the person who holds it, but it would not exist without the organization that enabled it. You can't split the difference without losing something essential.
ThreadWeave doesn't resolve this tension. It makes it visible. It forces organizations to confront a question they've been able to avoid because the knowledge infrastructure to ask it didn't exist:
When the organization remembers what Lars knew, who owns the memory?
The law has a partial answer: explicit communications belong to the employer. Philosophy has a partial answer: knowledge is not property in the way a building is property. The KM literature has a partial answer: psychological ownership matters as much as legal ownership, and systems that ignore it will face resistance.
But the full answer, one that balances organizational resilience against individual sovereignty and honors both the investment the organization made in Lars and the investment Lars made in himself, doesn't exist yet. It's being built. One thread at a time.
Based on a literature scan of knowledge ownership research: Baskerville & Dulipovici (2006) on ethical conflicts in KM; Davenport & Prusak (1998) on the knowledge market; Hrdy on the general knowledge, skill, and experience paradox in trade secret law; PepsiCo v. Redmond (1995) on inevitable disclosure; Heizmann & Olsson (2015) on Foucault and power in KM; Peng (2013) and Isaac et al. (2020–2024) on psychological ownership and knowledge hiding; Nonaka & Takeuchi (1995) on organizational knowledge creation; Drahos & Braithwaite (2002) on information feudalism; Anand et al. (2021) on the rhetoric of knowledge hoarding.