Bilateral Spaces is an artificial intelligence research lab. We build agents that understand how civilizations coordinate: through titles, mandates, reporting lines, and chains of command.
Today’s training loop runs almost entirely on text, and the most ambitious efforts aim to carry it into the physical world. We believe a third space matters as much as either. Every person casts a shadow, a tuple recording where they stand physically and where they stand societally. A person in a meeting is a body in a chair and a node in a chain of command, and the second determines most of what they will do today.
Physical intelligence forms in the physical space. Societal intelligence forms in the web of institutions we live inside. Bilateral intelligence emerges where the two collide. A model that understands how civilizations are created, how they are influenced, and how they thrive is what we mean by general intelligence.
Horizontal agents
Across, not down. Most agents today are vertical: a coding agent, a dispatch agent, a support agent, one job done deeply. The frontier for large organizations is the agent that moves across departments the way a chief of staff does. It carries a mandate from the top through every unit it touches, works out which programs and teams are affected, drafts the instruction for each, and reports back on what happened.
Where command is clear. This works best inside institutions with an explicit chain of command: governments, large enterprises, and sovereign wealth funds.
Long horizons through structure
Nobody holds long work alone. A person who carries a task for a week carries it through structure: the standing check-in, the report that comes due. Institutions turn long horizons into chains of short, accountable intervals.
We give agents the same scaffolding an employee gets. A mandate it returns to. Checkpoints where a human reads the work. An escalation path for when it is stuck. A case file that persists while the loop sleeps. A long-running agent owns a case the way an officer owns a docket: it wakes when something moves and reports on schedule.
Sovereign by design
Every government is understaffed against its own mandate. A decree takes months to become instructions in the regions, because the mapping from decision to work is done by hand. That gap is exactly the shape of work a second workforce of agents absorbs first.
For a nation, that workforce has to be sovereign. The agents run where the data lives, speak the country’s language, follow its own mandates, and are evaluated by someone who does not sell the model. An institution that owns its evaluations, its learning loop, and its chain of command remains sovereign whichever model sits underneath. Models are becoming interchangeable parts. The chain of command is not.
The first agent employees
These are not different products. They are one kind of employee holding different mandates, and each reports to a named human who signs off on the work.
Decree. Reads a new directive, maps which programs and regions it touches, drafts the instruction for each, and follows up until every region has moved.
Reporting. Reads operational data across every region overnight and puts the morning brief on the director’s desk.
Disbursement. Follows money from decision to repayment, and flags risk in week two instead of month six.
Writing
Join us
We have taught machines to write and to code. We have not yet taught them to read the room. If that is the problem you want to work on, write to us.