Vote on what we ship next
A two-board feedback system where you can file bugs, request features, and vote on what we ship next. Plus a brainstorm timeout fix you may have hit on long Opus runs.
The feedback system is live. Two boards: one private to your workspace, one shared across all customers. File a bug, request a feature, propose an improvement, and vote on anything you want to see prioritized. Admin triage is built in so you can watch a request move from open to in-review to shipped.
How it works
Each entry on the public board carries a vote count and a status. The private board is for things you'd rather not surface (workspace-specific friction, sensitive bugs, customer-account questions). You choose visibility on submission. Everything else is the same: comments, vote tallies, status transitions.
We'll start posting roadmap candidates to the public board as PA's own entries so you can vote them up or down before we commit. Lower the chance of us building the wrong thing in the wrong week.
Brainstorm sessions don't time out anymore
Long-running Opus reasoning would sometimes hit a 60-second proxy timeout on the brainstorm endpoint and the session would die mid-thought. The stream now holds open for 270 seconds and surfaces a Thinking… heartbeat so you see the model is working rather than the page just hanging. Anthropic adaptive reasoning is the case where this matters most; the fix applies to all providers.
Quietly underneath
Improvements & fixes
- Stripe seat counts sync on every subscription change. Adding or removing a seat in the Stripe customer portal now reflects in the workspace invitation flow within seconds. Previously the portal change took until the next webhook event to apply.
- Email send paths now honor your notification preferences. A few internal-trigger emails were skipping the unsubscribe check on transactional paths that should have respected it.
Related entries
Demo mode, cost drill-down, and the GPT-5 lineage
The biggest week of the run-up to launch. New users without a key get five free AI-pair runs. Every run shows what it cost. The model registry now ships GPT-5.x and Gemini 2.5 Flash-Lite.
Schedule eval batches without watching the timer
The new Schedule batch action queues eval-suite runs through Anthropic's Batches API so you can walk away. Results post back when the batch returns.