Playground
Try the engines on your own messages before you write any code.
The playground runs Humane's analysis on a message and shows what it detected and how it would shape the reply. Point it at a model and compare the raw reply with the Humane-shaped one side by side.
Playground
Raw model
Great question! There are several things you can try. First, let’s go over the options in detail…
With Humane context
I’m sorry you’ve had to ask three times. Here’s exactly where it stands.
tone warm_grounded · length short
See the reading before you trust it
Paste a message a real user might send. The playground returns the emotions and intent it detected, the state changes and the recommended tone, empathy and length. With a model chosen, the compare view sends the same message twice, once raw and once with Humane's behavioural context.
- Detected emotions, intent and the phrases behind them
- Recommended tone, empathy and reply length
- Raw and Humane-shaped replies side by side
- Rate-limited, with model URLs checked against SSRF rules
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
A useful first hour
- 1
Bring real messages
Take a handful of hard messages from your logs, with personal details removed.
- 2
Compare
Run each through the compare view with your own model and read the two replies side by side.
- 3
Decide where to start
Most teams start with Observe on every turn, then turn on gates where the comparison showed gaps.
Playground questions
Something else? Ask us.
Works well with
Model adapters
Humane talks to models through four adapters: OpenAI-compatible, Anthropic, Amazon Bedrock and Replicate. Change providers without touching the behavioural layer.
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.
ExploreObserve 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.
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.
