v1.0.0#
17/03/2026
Attention
Reply CMP v1.0.0 is now available v1.0.0 consolidates the improvements and customer feedback gathered across the entire v0.x series. This release brings the biggest leap forward in AI and machine learning capabilities to date, alongside significant performance improvements and quality-of-life refinements across every module.
What’s Changed#
v1.0.0 ships six major improvements together: saved queries with homepage pinning, ML-powered cost forecasting, backend top-k grouping for faster analysis, the AI Report Agent with crontab scheduling flexibility, a critical performance fix for change history, and a completely rebuilt documentation site.
🚀 New Features#
Saved Queries & Homepage Pinning#
Save any Analyze configuration (filters, grouping, time range, chart type) as a named query. Saved queries can be pinned to the Home page, where they appear as live interactive charts refreshed on every visit. Up to 3 pinned queries are displayed.
Pin / unpin from the saved query list in the Analyze header
Share saved query URLs with teammates — the recipient sees the exact same view
ML-Powered Cost Forecasting#
Both the Forecast toggle in Analyze and the Forecast threshold alerts in Budgets now use a machine learning model for predictions up to 30 days ahead. For budget periods longer than 30 days (quarterly, annual), a linear regression model extends the forecast beyond the ML horizon.
Analyze: enable the Forecast toggle on any chart to see an ML-powered cost projection
Budgets: Forecast threshold alerts fire before a budget is exceeded, based on the ML prediction for the first 30 days of the period and linear regression thereafter
AI Report Agent#
Reports are now created through the AI Report Agent — a conversational experience embedded in the Reports section. Describe the report you need in plain language; the agent proposes a configuration, shows a live HTML preview, and saves the schedule on confirmation. Full crontab syntax is supported for scheduling, so you are not limited to daily / weekly / monthly presets.
Examples:
0 9 * * 1-5— every weekday at 09:000 8 1,15 * *— twice a month on the 1st and 15th0 8 1 1,4,7,10 *— quarterly
The template-based content model (introduced alongside this release) ensures consistent report layout across all scheduled deliveries.
Faster Cost Analysis#
Grouping and aggregation in Analyze are now performed at the backend using a top-k algorithm. Only the top-k most expensive dimension values are returned and rendered; all remaining values are consolidated into an “Other” bucket. Default top-k: 15 (primary group) and 10 (secondary group).
Significantly faster chart rendering on large accounts
Cleaner, more readable charts with no more 200-series overload
Configurable per query in advanced options
Connection Data Freshness#
Each connection card now shows when data was last updated, so teams can immediately spot stale connections without digging through logs:
When the last resource discovery scan completed
When cost data was last refreshed
How recent the reservation data is (for connections with Reserved Instances, Savings Plans, or CUDs)
New Documentation Site#
The Reply CMP documentation site has been completely rebuilt from the ground up with:
Task-oriented guides for every major feature
A dedicated AI Features section covering all six capabilities
Improved navigation structure and full-text search
All content updated for v1.0.0
🔧 Improvements#
Notifications: notifications can now be dismissed one at a time or all at once with a single action
Analyze chart: the forecast section now renders with full date context, making it easier to read projected costs alongside actuals
Discovery events: each event now carries the identifier of the discovered resource, making it easier to link audit entries to specific inventory items
Connection details: reservation data freshness is now tracked and displayed separately from cost prediction freshness
🐛 Bug Fixes#
Fixed critical performance issue in change history: single-resource event retrieval was extremely slow on large accounts; the History tab now loads in under a second, making AI Rollback Suggestions fully practical for rapid incident response
Fixed bulk notification dismissal not working correctly in all cases
Fixed individual notification dismissal occasionally failing silently
Fixed the cost analysis view including response fields that had no functional effect
Fixed AI report creation failing when no manual content instructions were provided
📚 Documentation#
Complete documentation site rebuild
Added Saved Queries & Homepage Pinning section to the Analyze guide
Added ML forecasting two-stage model description to Analyze and Budgets guides
Added AI Report Agent guide and crontab scheduling reference
Added v1.0.0 performance note to Change History guide
Added query performance (top-k) section to Analyze guide