01
You are interacting with AI
Conversation analyses, scores, comparisons, translations and suggested replies are generated by AI. HIL™ structures the request and the result around communication intelligence; it does not turn an inference into a verified fact.
02
What the system uses
Hilho analyzes the text, images or documents you choose to submit, together with the context and language preferences you provide. It may recognize text inside a screenshot. It is not designed to identify a person from a face, voice or other biometric feature, and it does not independently investigate the people in a conversation.
03
Permission before third-party AI processing
Before the mobile app sends a conversation, selected files, context or reply preferences to OpenAI, it identifies what will be sent, names OpenAI as the recipient, explains the purpose and asks for explicit permission. Declining blocks the request. Permission can be withdrawn in Account, and the Hilho server enforces that choice before a new AI request is made.
04
What HIL™ means
HIL™ — Human Intelligence Layer™ — is Hilho's name for the orchestration layer that combines eight specialized analysis functions: IntentMap™, ToneTrace™, TrustLens™, PressureRadar™, ContextGraph™, PersonaMap™, ReplyCraft™ and LinguaBridge™. These names describe Hilho's product architecture; they are not claims of scientific validation, independent certification or patented technology.
05
How to read the output
A plausible communication goal inferred from the available words and context.
A language-based interpretation of tone—not a measurement of a person's internal state or a clinical assessment.
A measure of how much usable evidence the conversation provides. It does not judge truthfulness, identity, safety, or a person's character.
Patterns that may merit attention. It is not proof of abuse, deception or wrongdoing.
A comparison of stated preferences and conversation patterns, not a prediction of relationship success.
An editable draft. You decide whether, how and when to communicate.
06
Known limitations
- missing context, sarcasm, cultural norms, dialect and screenshots can change meaning;
- an AI model can produce confident but incorrect explanations;
- scores can imply more precision than the evidence supports;
- the person whose messages are analyzed has not necessarily provided their side of the story; and
- translation can preserve meaning imperfectly.
07
Human oversight
No Hilho output should automatically trigger a consequential decision. Read the original conversation, consider alternative explanations and check facts independently. For safety, health, legal or high-impact situations, use a qualified human professional.
08
Uses we do not allow
Hilho must not be used for social scoring, biometric emotion inference in workplaces or education, sensitive-trait inference, covert surveillance, or automated/high-impact decisions in employment, education, credit, insurance, housing, health, law enforcement, immigration or legal services.
09
Report a concern
If an analysis appears harmful, discriminatory, misleading or unsafe, do not rely on it. Send the relevant context—after removing unnecessary personal data—to hilho@hilho.com.