Core Concepts
The product primitives behind Oriyn: behavior, personas, patterns, and experiments.
Oriyn models user behavior as evidence, derives personas and patterns from that evidence, and uses those personas to simulate product decisions.
Behavioral Graph
The behavioral graph is Oriyn's unified view of what users did across product surfaces and providers. Events, sessions, revenue markers, and product context are resolved into user-level evidence that can be searched, clustered, and cited.
- Events show what happened in the product.
- Sessions show how people moved through the experience.
- Revenue and account signals show business impact.
- Product context helps agents understand the surface being changed.
Personas
Personas are behavior-derived user segments. They are not fictional archetypes. A persona should have source users, traits, representative evidence, and a size estimate so teams can inspect why Oriyn believes the segment exists.
Patterns And Hypotheses
Patterns are mined observations about funnels, paths, drop-offs, repeated behaviors, and bottlenecks. Hypotheses turn those observations into testable product changes.
Good hypotheses are specific
Use one user-visible change and one expected outcome. For example: 'Move pricing before signup to reduce low-intent trials' is easier to simulate than 'Improve pricing'.
Simulated Experiments
A simulated experiment asks persona-grounded agents to reason through a product change using evidence from the behavioral graph. The output is decision support, not a replacement for live measurement.
| Output | Meaning |
|---|---|
| Verdict | A directional recommendation such as ship, revise, or reject. |
| Convergence | How consistently persona-grounded agents reached similar conclusions. |
| Persona breakdown | Where the decision helps, hurts, or confuses specific user groups. |
| Caveats | Weak evidence, missing context, or risks that should be checked before shipping. |