Model role requirements
Separate responsibilities, capability, hard gates, preferences, and practical constraints for team confirmation.
Public methodology
How Aimme produces candidate-match conclusions that teams can challenge and review
Aimme does not aim to compress a person into a deceptively precise number. It helps recruiting teams inspect evidence faster. The method separates job capability, eligibility gates, practical feasibility, and evidence quality so teams can understand a conclusion, identify unknowns, and choose the next validation step.
The recruiting problem
Candidate matching has no context-free ground truth. The same person can warrant a different judgment for another role, team, or stage. A methodology must state its inputs, identify non-compensable requirements, preserve evidence provenance, and admit what the model does not know.
How Aimme handles it
Break recruiting judgment into inspectable steps
Assess capability
Evaluate direct and transferable evidence related to work outcomes while recording support and counter-evidence.
Separate eligibility and feasibility
Use pass, fail, or unknown for hard gates and keep practical constraints outside capability scoring.
Evaluate evidence quality
Record provenance, freshness, completeness, and conflict; low-trust information cannot become a certain fact.
From question to reviewable result
Preserve context—and the opportunity to stop and review
-
01
Version criteria
Create a new version when a consequential role requirement changes and retain the reason.
-
02
Collect evidence per criterion
Find direct evidence, adjacent evidence, counter-evidence, or an explicit unknown.
-
03
Apply bounded reasoning
Summarize within stated rules and never invent experience to fill missing candidate data.
-
04
Produce review material
Show track results, strongest evidence, key risk, missing information, and proposed validation questions.
Why the result is reviewable
Fluent model output is not treated as fact
- A conclusion cites at least one locatable source; without one, it is marked insufficient evidence.
- A failed hard gate takes precedence over capability score, while an unknown gate triggers review.
- Conflicting sources are preserved together rather than resolving toward the more favorable claim.
- Human edits retain a reason and may calibrate later work without being presented as the original model judgment.
Operating boundaries
Be explicit about what the system does—and does not do
- Aimme does not recommend candidates based on protected traits such as name, gender, age, or ethnicity.
- Model output is recruiting support, not a hiring, rejection, or compensation decision.
- Missing evidence, conflicting sources, and unknown hard requirements are surfaced for human verification.
- Customers control candidate data. It is not used to train public models, and teams remain responsible for lawful, platform-compliant use.
Frequently asked questions
Answer the important questions first
Why not use only one match score?
A total score helps prioritization but can hide failed gates, weak evidence, and practical infeasibility. Aimme retains an overview while requiring track-level review.
What counts as reliable evidence?
Information is stronger when it is directly relevant, locatable, current enough for the decision, and explicit. Self-description, inference, and stale records require lower confidence.
How is human feedback used?
Team edits to criteria and conclusions can calibrate later missions. Historical input, model output, and human decisions remain distinguishable and traceable.
Aimme
Start with one real role
Describe an active role. Aimme clarifies the requirements, searches and matches candidates, and organizes the evidence. Your recruiting team confirms every progression decision.