A working model of your research
A memory that grows as you work, and a researcher profile built from your public research record at signup. Both are yours to read and edit.
The tenth question you ask should be smarter than the first, because by then Althea knows what you are working on, which subfield you live in, and your preferences. You stop re-explaining your context every session.
Before your first message
At signup, Althea builds your researcher profile from your public research footprint: professional databases for roles and affiliations, neural search for your papers, talks, and open source work, plus the general index for the gaps. An LLM pass fuses the fragments and resolves ambiguities. The result is structured into two fields, Professional Experience and Interests & Hobbies, loaded before you type anything. Ask a technical question and it calibrates depth against your experience; briefings filter against your interests.
| Layer | What it holds | Where it comes from | You control it by |
|---|---|---|---|
| Profile | Professional Experience and Interests & Hobbies; the structured model of you as a researcher | Built at signup from your public work | Full read/write access in settings |
| Memory | What you are working on now, preferences, context from your conversations | Grows continuously as you work together | Asking. "What do you know about me?" gets a straight answer; "forget that" works too. |
Audit it in one message
Ask "what do you currently know about me?" and Althea answers concretely: your identity, your current research focus, your preferences, and which layer each lives in. If anything is wrong, say so and it corrects or forgets it.
What it knows
A profile built from public data can be wrong, which is exactly why it is fully readable and editable rather than hidden. The profile represents your public research record; Memory retains what you have shared. What leaves your conversation, and what never does, is covered under privacy and consent.