Decision design + assurance for consequential AI decisions
Decide where AI belongs in one real decision. Get an explainable recommendation, then turn it into a free Design Brief covering evidence, alternatives, human approval, and what must be proven before anyone acts.
Synthetic example · actual deterministic output
Weekly inventory replenishment
Illustration only · no real orders or outcomes
01 · Fit recommendation
Strong fit. Allow only a specifically approved action through a bounded execution path.
Fallback: DA-3. This is a provisional design recommendation, not proof of a deployed capability or permission to act.
02 · Decision Design
Export your own Design Brief, canonical JSON, technical Markdown, or print view for review.
Your free path
Complete the part you need. Payment is never required for Fit, Design, exports, or the local Assurance workspace.
Six Fit sections produce a recommended AI role, evidence gaps, human controls, and a provisional design ceiling.
Start Fit →Clarify the alternatives and safety rules. Download a Design Brief, JSON, technical Markdown, or print view for accountable review.
Continue from Fit or import Fit JSON →Import the Design into local-first Assurance to keep implementation evidence, human gate acceptance, authority, and outcomes distinct.
Open Assurance →See the evidence loop in practice. The first-party reference cycle records Decisagent’s own v0.9 release. It is observed first-party evidence, not a customer outcome or the synthetic example above.
Optional delivery planning: $129 once. The Implementation Pack adds architecture, a rollout plan, acceptance gates, backlog, and implementation-agent handoff. It does not implement, approve, or authorize your system.
Category definition
It works with evidence, alternatives, objectives, constraints, uncertainty, authority, accountable people, and observable outcomes. Choosing an internal tool while completing a task is not enough.
Read the working definition →The unit of analysis
EvidenceWhat information supports the decision—and what is missing?
ChoiceWhich real alternatives are evaluated against objectives and constraints?
AuthorityWho authorizes an action, under which boundaries?
OutcomeCan a result be attributed to the originating decision?
Accepted product intelligence
2 human-reviewed profiles.
Decision-first map
Choose a supplier or supplier portfolio for a defined requirement and operating horizon.
02Choose whether and how to intervene when supply, logistics, quality, or demand conditions threaten an operating plan.
03Choose a price, price floor, discount, or pricing policy for a product, customer, or market context.
04Choose how much operational, technical, or service capacity to provision across a defined planning horizon.
05Choose whether, when, and how to intervene in a customer risk, service, retention, or safety situation.
Capability, not permission
A level describes which decision capabilities have evidence at an exact scope. It is not a five-star rating, product ranking, operational assurance result, or permission to operate at that authority.
Inspect Methodology v0.2 →Information or analysis without a first-class external decision.
Evidence assembly for a named decision, without alternative evaluation into a recommendation.
Explicit alternatives evaluated into a recommendation or justified abstention.
A reproducible trace plus at least one explicit objective, constraint, or uncertainty signal; not complete operational governance.
A consequential external action under explicit policy or action-time authority.
Attributed outcome, expected-versus-actual comparison, and a governed later-decision update.