Research noteEditorial
Drift control
How changes in models, workflows, policy, evidence, or context affect continuing authorization.
Scalebotics blog
Notes from the value gap.
Long-form writing from Scalebotics research and operating work — research notes, field memos, case studies, and playbooks. Each piece states its evidence boundary up front.
01Editorial index
The running record.
New pieces are added as they are published, newest first. Four kinds: research notes, field memos, case studies, and playbooks.
Research noteEditorial
Hallucination control
How operating conditions, review, and evidence manage unsupported AI outputs.
Research noteEditorial
The governed agent runtime
How agent actions can be evaluated against customer-defined policy with explicit outcomes and reviewable records.
Research noteEditorial
Bounded autonomy
How agents can operate independently inside approved limits while routing consequential exceptions to people.
Research noteEditorial
Policy-based AI governance
How customer-defined rules, decisions, human review, and records support AI operation.
Field memoEditorial
What evidence supports — and what it does not
The trust and evidence boundary: what Sable evidence contains, and what it does not certify or guarantee.
PlaybookEditorial
Questions for evaluating governed AI products
The buyer and procurement brief: policy, authority, evidence, access, support, deployment, and operating limits.
PlaybookEditorial
Find uncovered agent actions
Inspect a connected surface to distinguish governed actions from policy and instrumentation gaps.
03Next decision
Reading is the start. The decision is the point.
If a piece raises a question about your own AI uses — the value case, the evidence, the operating conditions — start an inquiry. A reply should clarify fit and the most sensible next step.