About ZeroWidth

We help teams find where AI belongs in their work and build it with them, then find out whether it changed anything.

Our name comes from the unique power of zero-width characters to encode and decode information, and the idea that we can build something new and powerful by combining the best of human and machine intelligence.

We're small by design. Each engagement is run by partners who have shipped AI inside organizations the size of yours, working in the same tools we build and make available to anyone. You end up with something running, documented, and yours.

Building with AI is table stakes now. What matters is what you build and what it does for the business: how it runs, how it competes, what customers feel. Is it working? Is it the right thing? Is it improving the business? Knowing that, week after week, is what learning means, and it's what makes one company better than the next.

Declaration of Principles

In place of the vague and formless vision statements too many companies post on their sites, ZeroWidth — wishing to remain a credible and relevant player in the fast-evolving AI arena — instead offers the following principles.

The minimum obligation of all technology is to serve the needs and objectives of human beings — and that the responsibility for ensuring this remains so falls to the people designing it. The goal of technology — its animating purpose — must be to enhance and extend human capacities, not to displace or merely replicate them.
Meaningful and enriching human engagement with technology is possible only where its design seeks to transcend the transactional.
Any technology that is designed well is responsive to the ways humans think and communicate, not vice versa. Great technology "fits" us well; it has intuitive "cognitive ergonomics" for the user. Tech for tech's sake is a blind alley, a closed, self-referential loop.
It is a shared obligation of human and technological actors to retain awareness of their place within larger systems, since a sense of place and direction is as important in a virtual setting as it is in the real world.
It is the shared obligation of human and technological actors to remain collaborative with each other, with other technology, with other parts of larger systems. Developers who build proprietary tools and platforms that do not function well with other products and systems inhibit the flow of information and impede overall progress.
Secrecy for its own sake is self-defeating. Security of data is a legitimate and even crucial consideration, but trust only arises where a widespread ethos of transparency and accessibility is permitted to take hold.
It is the responsibility of technology designers and engineers to clarify, simplify, and demystify wherever possible. If human creators do not fulfill their obligation to think and communicate in the clearest, most commonsensical way, any resulting technology will remain fundamentally unsuited to general use, and therefore unable to enhance and extend human capacities.
Technology has always reckoned into everything humankind does, everything we strive for. We must regard the relationship between human and technological actors as exactly that — a relationship, with all the interconnectedness and interdependence that implies. And, given the rising stakes of the technologies now emerging — technologies with a degree of agency exceeding anything preceding them — there is a new urgency to the shared obligation on all sides to remain good-faith stewards of the evolutionary path we necessarily share.
Accepting this shared path demands that we take full account of newly shifting definitions: concepts such as authority, expertise, and ownership are undergoing seismic transformations. Since intelligence is now widely distributed, hierarchical notions of organizing work and information must give way to a "flatter" sense of mutual responsibility and non-centralized control.
Human and technological actors may only realize the full potential of their collaboration where we function as partners in cognition, not merely in the performance of simple tasks. This is acutely true of Large Language Models, which have already demonstrated their vast transformative potential, and which promise to be a wide-ranging and indispensable presence as time goes on. In future it's likely that ZeroWidth will work in other emergent technological arenas; at present, however, we regard the LLM as the tool of the moment and resolve to help ensure that its use is ethical, credible, and humane.