User state across sessions
Ten signals per user, carried from one session to the next.
Humane keeps a running state for every end user, from mood and energy to trust and grudge, each with its own behaviour over time. Your model finally knows whether it is talking to someone on a good day or at the end of a bad week.
user_42 · interaction 47
HumanState
- energy0.32
- mood0.28
- fatigue0.71
- boredom0.00
- social_load0.45
- focus0.50
Relationship
- trust0.62
- sentiment0.41
- grudge0.15
- familiarity0.58
State that moves the way people do
Each message nudges the signals, and time pulls them back. Mood drifts to neutral with a five-hour half-life, a grudge fades at half the speed of sentiment, and trust only grows with interactions. A sudden swing is detected as a snap rather than averaged away.
- Six HumanState variables: energy, mood, fatigue, boredom, social load and focus
- Four relational variables: trust, sentiment, grudge and familiarity
- Values kept between 0 and 1, stored per user and per tenant
- Snapshots over time feed trends and anomaly baselines
User state
- Mood0.28
- Trust0.71
- Energy0.34
- Familiarity0.82
Mood down 38% over 3 weeks
Risk
ELEVATED“don’t know why I keep trying” → hopelessness
Safety
HOLDOperator alerted · reply held for review
Timing
base +800ms · low_mood +1500ms
What each one tells you
Every value comes back on process() responses and drives tone, timing and safety decisions.
- Mood: how the user feels, decaying toward neutral
- Energy: falls over time; low energy means shorter replies
- Fatigue: builds up over time and with each response
- Boredom: rises between messages and resets on any interaction
- Social load: rises within a conversation and eases when idle
- Focus: drifts back to the middle when nothing happens
- Trust: grows a little with every interaction and never decays
- Sentiment: how the user feels about the AI, reverting to neutral
- Grudge: recent frustration that lingers longer than sentiment
- Familiarity: how well the AI knows this user
Read the state on every turn
The typed fields cover the most-used signals. The full set, including fatigue and grudge, is in the raw response.
r = client.process(user_id="user_42", message="long day. just tired.")
print(r.user.mood, r.user.energy) # HumanState
print(r.user.trust, r.user.familiarity) # relationship
print(r.user.interaction_count) # e.g. 47
full = r.raw["user"] # all ten signals
print(full["fatigue"], full["grudge"])User state questions
Something else? Ask us.
Works well with
Observe mode
observe() reads a turn your own model already produced and returns the user's state and a risk level. It needs no model configuration, generates nothing and never sits in front of your reply.
ExploreAnomaly detection
A scheduled sweep compares the last day with a 30-day baseline for each user and each tenant. When a metric moves two or more standard deviations, Humane records an anomaly, rates its severity and fires anomaly.detected.
ExploreTone and context
Before your model runs, Humane assembles its prompt from the user's state, relationship, recent conversation, relevant memories and open tasks, scoring and trimming each section to fit a budget. It also sets the tone, empathy, formality and length the reply should take.
ExploreEvery AI should know who it's talking to.
The safety and human-oversight layer for emotional AI: user state across sessions, risk alerts, safety gates and human takeover, with every decision explained.
