Everywhere trust needs to be earned

Companion robots that greet every person with identical warmth feel uncanny and quickly lose credibility. Hard-coded personality states are obvious and brittle.

CCF accumulates coherence per person-context pair. The robot is genuinely more reserved with new faces and warmer with familiar ones — without a single scripted rule.

After 10 evening interactions with the same child, the robot shifts to Quietly Beloved — softer voice, warmer LED, slower movements. With a new adult, it stays in Shy Observer until trust is earned.

Smart speakers and ambient devices treat every household member identically and reset on reboot. There is no memory of who earned what level of trust.

A CCF-enabled device accumulates per-user coherence across sessions via persistent snapshot. Context keys encode time-of-day, location, and acoustic signature.

The device speaks more quietly and offers unsolicited help to a user whose morning context coherence is high — behaving like a thoughtful household member, not a voice-activated appliance.

Field robots operating near humans in warehouses or construction sites have no safe way to modulate their proximity and communication style based on familiarity with specific workers.

CCF provides a mathematically bounded social proximity model. The robot maintains separate coherence per worker-context and adjusts movement speed, alert thresholds, and interaction style accordingly.

Bob works every day with Evie, the packing robot. Over weeks, their interaction develops a kind of shorthand — less preamble, fewer warnings, faster handoffs. The kind of working-buddy fluency that only earns itself through time. A new temp on the floor gets the full cautious version of Evie. Same robot. Different history.

Game NPCs have reputation systems that players immediately reverse-engineer and exploit. Relationship meters are visible, gameable, and emotionally unconvincing.

CCF's dual-threshold gate prevents simple grinding. Instantaneous coherence requires in-context engagement; the long-run accumulator requires sustained history in that specific game context.

A village elder NPC won't share quest-critical information after one generous gift. The player must build genuine in-town context coherence over multiple sessions before the gate opens.

Health wearables deliver alerts and coaching with one-size-fits-all intensity regardless of the user's current context or their established pattern of engagement.

CCF tracks coherence per activity-context key (morning run, post-meal, sleep wind-down). Coaching tone and alert intensity adapt to the user's context-specific trust baseline.

During a context the device has seen 40 times (Thursday evening run), it speaks in Quietly Beloved mode — brief, warm, non-intrusive. In a novel context, it defaults to Shy Observer.

LLM-based agents call tools and take actions with no sense of whether this context has been safe and productive before. Every session starts from scratch.

CCF acts as a social gate for agentic tool calls. The agent's action request includes a context vector; CCF returns a permit or defers until coherence is earned in that specific operational context.

An agent that has successfully executed file-write actions in a familiar dev environment (high coherence) is granted permits faster. In an unknown production environment, the gate stays conservative.

Care robots that introduce themselves with maximum friendliness to vulnerable adults cause distress. Familiarity that hasn't been earned feels intrusive, not supportive.

CCF enforces a graduated relationship arc. The robot begins in full Shy Observer — observing, minimal contact — and earns the right to more intimate interaction over days and weeks of positive episodes.

A resident with dementia who is initially startled by the robot sees it become, over three weeks, an unremarkable part of the morning routine — always present, never surprising, earning its place in each room.

Tutoring robots and AI tutors that probe difficult subjects on the first session frighten students off. Depth of engagement should reflect depth of relationship — not just topic difficulty.

CCF accumulates per-student, per-subject-context coherence. The tutor earns the right to challenge, push back, and go deep only after the student's accumulated engagement warrants it.

In maths, after 20 sessions, the tutor confidently disagrees with the student's method and explains why. In a brand new subject like coding, the same tutor stays gentle, scaffolded, and non-presumptuous — same student, different accumulator.

Hotel robots and airport assistants deliver identical scripted interactions to first-time visitors and regulars alike. Repeat guests who expect recognition feel the absence of it sharply.

CCF accumulates coherence per guest-context across stays. A returning guest in a familiar context gets warmer, less procedural interaction — earned, not programmed.

A guest checking in for the fourth time finds the concierge robot skips the orientation speech. It greets by preference, not by script. The guest notices. That noticing is the product.

In-car AI assistants that offer unsolicited suggestions, commentary, and warnings from day one are muted within a week. They presume a relationship that doesn't exist.

CCF gates the car AI's expressive range by context — this driver, this route, this time of day. Suggestions, music, and comfort interventions are earned over journeys, not assumed from the first one.

After six months of Monday morning commutes on the same route, the car knows not to suggest alternatives unless traffic is severe. It's learned this driver's tolerance. A rental driver on the same road gets the full cautious version.

AI therapy tools that probe emotional depth before a therapeutic relationship is established can cause harm. Regulatory bodies require demonstrable safety architecture, not just policy.

CCF provides an architectural safety guarantee: the AI cannot go deeper than its accumulated relational coherence warrants. It is not a policy layer — it is structural. Cannot be fine-tuned away.

After 15 careful sessions, the AI earns the right to name a pattern it has observed. Before that threshold, it asks, listens, and scaffolds — never probes. The constraint is in the mathematics, not the prompt.

Farm robots working alongside operators in variable field conditions have no model of the accumulated trust between machine and operator. Every morning is a cold start.

CCF accumulates coherence per operator-context (field, weather band, task type). The robot's assertiveness, speed, and communication style calibrate to the relationship — not just the conditions.

A harvesting robot that has worked alongside the same farmer for three seasons coordinates with minimal explicit instruction. With a new seasonal worker, it slows, signals more, and takes the conservative line until coherence is established.

Customer service AI that opens every call with maximum helpfulness and familiar tone is immediately identified as a machine. Calibration to relationship history is absent.

CCF accumulates per-customer coherence across contact episodes. Tone, depth of proactive help, and level of assumption about the customer's situation adapt to the earned relationship.

A customer who has called eight times about the same account type gets offered the advanced fix immediately — the AI has earned enough coherence to skip the diagnostic script. A first-time caller gets the full guided flow.

Security robots that challenge every person identically — regardless of familiarity — create friction with regular staff and fail to calibrate alertness to genuine novelty.

CCF maintains coherence per personnel-context key. Familiar staff in expected contexts get frictionless passage. Novel contexts, unusual hours, or unfamiliar presence patterns elevate to Protective Guardian.

At 2am, a cleaner who is on the regular overnight roster — whose context coherence for this building, this hour, this badge signature is high — moves freely. An unrecognised badge at 2am triggers a different response entirely.

Surgical assistance robots that behave identically with a senior surgeon of 200 joint procedures and a registrar on their first assisted case create unnecessary risk.

CCF accumulates coherence per surgical team context — surgeon, procedure type, theatre environment. Anticipatory behaviour, instrument handoff timing, and error tolerance adapt to earned familiarity.

After 50 shared procedures, the robot anticipates the surgeon's next instrument request with the quiet fluency of a scrub nurse who has worked that theatre for years. In a new hospital or with a new lead, it reverts to conservative, fully-signalled behaviour.

Robots in isolated, high-stakes environments (space stations, deep sea facilities, Arctic research stations) must develop deep trust with a small, fixed crew over months — and maintain it through stress.

CCF's interaction-count-proportional decay floor means deep trust, once earned, is resilient to isolated incidents. The robot survives a bad day without losing the relationship. It recovers, as any long-term companion must.

After four months on a research station, the crew robot has earned Quietly Beloved status with three of the six crew members in their respective contexts. During a high-tension equipment failure, it moves into Protective Guardian — calm, purposeful, drawing on the deep accumulated familiarity to know exactly who needs what.