Emotionally intelligent AI, through an API and a platform.

Each stage builds on the one before it and has its own gate. A later stage is not promised on a date.

1. 01

   Evaluations

   Proposed first offer

   Evaluate your agent's existing transcripts. You get the failing turn, the reason, a comparator, and the uncertainty.

   Gate · Data rights and handling terms agreed before any transcript review.

2. 02

   Observe

   Later

   Score production sessions per turn and replay mishandles, so a team sees where context was lost.

   Gate · A separate product gate after evaluations.

3. 03

   State API

   After validation

   Send a conversation, receive an inspectable person state, and pass it to any model you already use.

   Gate · Only if state beats strong history, summary, and larger-model controls on real people.

4. 04

   Affective models

   Longer-term research

   Emotionally intelligent models served through the API and the Affective platform.

   Gate · Depends on the milestone results. A tie or loss for state means this stage is not pursued as described.

The aim is a first call in minutes: keep your existing model and code, add a user ID, and read back the state. This is a design sketch, not a working interface.

# Design sketch only. No public API exists today.
client = Affective(api_key="...")

reply = client.chat.completions.create(
    model="<your-model-or-affective>",
    messages=conversation,
    user="user-123",          # enables per-person memory
)

reply.state            # inspectable state, with evidence and uncertainty
client.memory.get("user-123")   # see what is remembered

Before building models, we are testing whether structured person state beats strong controls on real people. If it does, the later stages follow. If it does not, we say so. Read the current evidence.