Astra now analyzes a sample of 10k conversations
We’ve expanded Astra’s sample analysis capacity from 2k to 10k conversations per run. Analyze organic and paid, owned and earned mentions, plus social listening, in one pass—so trend lines are cleaner, long-tail themes surface, and you don’t need to split batches.
Use cases:
Audit a full month or a high-volume campaign in one go
Compare campaign vs. BAU with enough volume for a statistical signal
Spot weak-signal risks or opportunities hiding in long-tail comments
AI Safeguards for automated + pending approval AI replies
Guardrails for auto-replies in Agent+. They run on the incoming message and on the drafted reply. Flagged items are held for review or auto-archived.
Use cases:
Enforce brand voice and block legal/medical claims or guarantees
Filter off-topic or spammy messages before they reach queues
Choose oversight: review flagged drafts or auto-archive during spikes
Ready-to-send queue for approved saved replies
You can now set up workflows to pre-draft replies using saved replies, and have them land in a dedicated ready-to-send queue inside the composer for batch insertion. Usage is tracked for reporting.
Use cases:
Stage seasonal/campaign replies ahead of spikes and go live instantly
Keep brand oversight while cutting response time during surges
Standardize FAQs without hunting through large libraries
Agent+ dashboard and reporting
Agent+ is a performance hub built around how your team actually works. See Agent+ involvement and assist rate, plus approve/edit / reject rates for assisted replies. Track how many conversations were tagged as engageable, which engageable tags are driving volume, and how often knowledge was cited vs. where knowledge gaps appeared. Review a knowledge gaps feed, and usage of saved replies and prompts.
Use cases:
Coach with evidence: use edit/reject patterns and assist rates to target training
Strengthen your knowledge base using the gaps feed
Align staffing to actual workload and tag mix; compare teams or campaigns
CSAT per agent
CSAT is now broken down by agent with a sortable table showing average score, trend deltas, and the full 1–5 distribution.
Use cases:
Link QA and coaching to measurable satisfaction outcomes
Identify topics or time windows where CSAT dips and fix root causes
Recognize high performers and replicate their best practices
Post analysis context added to earned mentions for better replies
Your reply engine now receives full post analysis context for earned post types as well, including caption mentions, hashtag mentions or comment mentions. We pass through creative text from OCR, a short description of the content, and the detected intention of the post. This gives the AI replies a clearer understanding of what the user actually meant when they tagged the brand, especially when looking at the caption alone is not enough.
Use cases
Improve accuracy on caption mentions where the true intent is in the creative
Reduce misfires on comment mentions that lack context
Speed up approvals with more grounded suggested replies
One click try again for suggested replies
When a workflow creates a ready to send suggested reply, you can now hit Try again to instantly generate a new option without leaving the thread. You can also delete a suggestion from the card if it is not useful. Approve & send, Reject reply, and Edit continue to work as before.
Use cases:
Quickly cycle through alternatives until the reply fits your brand voice.
Remove low-quality suggestions to keep queues clean.
Speed up moderation by staying on the same comment or DM.
