We’re building a client portal on top of the Pickaxe Studio API and have run into a limitation with the memory endpoints.
What we can do today:
-
GET /studio/memory/list— list memory templates -
GET /studio/memory/user/<email>— read a user’s collected memory values -
PATCH /studio/memory/user/<email>/<memory_id>— update an existing memory value
What we can’t do:
- Create a memory value for a user when one (memory value) doesn’t exist yet
The PATCH endpoint returns 404: "Memory not found for the user" when the user has no existing value for that memory category. This means we can only edit values that the AI has already collected during a conversation — we can’t seed them beforehand.
Our use case:
We manage onboarding for enterprise clients and want to pre-populate memory profiles for users before their first conversation. For example, we know a user’s role, department, and communication preferences from our CRM data. Being able to write those values via the API before the user ever interacts with a Pickaxe would make the first conversation immediately personalized instead of starting cold.
What would help:
A create/upsert endpoint for user memory values — something like POST /studio/memory/user/<email>/<memory_id> with { data: { value: "..." } }, or making the existing PATCH endpoint behave as an upsert (create if not exists, update if it does).
Thanks for considering this — the memory feature is great and this would make it even more powerful for teams managing users at scale.
