Timing engine
Reply at a human pace, and show the working.
Humane recommends how long to wait before each reply, based on the user's mood, energy, the intensity of their message and the pace of the conversation. Every delay is broken into labelled contributors, so you can see and chart why.
Timing · one reply
delay_ms
13,857- base+800ms
- low_mood+1,500ms
- emotional_weight+8,648ms
- conversation_pace+500ms
- jitter+2,409ms
mode: low_mood · floor 200ms · non-crisis ceiling 30s
An instant paragraph can feel wrong
A long reply that lands half a second after someone shares bad news reads as mechanical. The timing engine picks a dominant mode from the user's state and message, then adds contributors on top. Crisis and urgent messages skip the random jitter so they feel precise.
- Modes: crisis, urgent, low mood, low energy, emotion-specific or conversational
- Emotional weight adds 5 to 30 seconds for intense messages
- Pace mirroring keeps a quick back-and-forth quick
- A busy signal from social load, and a 30-second dependence cooldown
Context · user_42
- system_personakept
- behavioral_constraintskept
- relational_context
- recent_conversation
- memories1 dropped over budget
- active_tasks
- time_situation
- response_instructionskept
tone
warm_supportive
modifier
cautious (grudge)
empathy
0.82
formality
0.35
response_length
short
Rules you can see in every response
- A 200 ms floor, because instant replies feel mechanical
- A 30-second ceiling for anything that isn't a crisis
- Low mood adds 1,500 ms and low energy adds 1,000 ms
- Jitter is tunable per tenant and skipped for crisis and urgent messages
- Typing indicators and pauses between chunks on messaging channels
- timing.reason lists every contributor and its value
Use the recommendation, or don't
r = client.process(user_id="user_42", message=text)
print(r.timing.delay_ms) # e.g. 4200
print(r.timing.reason) # the contributors, in order
time.sleep(min(r.timing.delay_ms, 8000) / 1000) # cap it if you prefer
send(r.response)Timing questions
Something else? Ask us.
Works well with
Tone 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.
ExploreUser state across sessions
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.
ExploreTelegram and WhatsApp
Register a Telegram bot or a WhatsApp Business number and Humane receives messages, runs them through the full pipeline and delivers the replies. Inbound webhooks are verified, and outbound replies are split, paced and tracked.
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.
