Recede · Always-on FinOps
Cloud spend, acted on. Not just reported.
Recede runs autonomous FinOps agents that act on AWS, Azure, and GCP around the clock — flagging idle resources, rightsizing the oversize, switching non-prod off at night — and posts a weekly ledger of what was reclaimed. Reports stay static on dashboards; Recede ships the change inside your guardrails.
20–45% returned on addressable cloud spend, typical first quarter. Start with a readout — no cloud connection or credit card required.
Weekly ledger · sample
14 unattached EBS volumes, 890 GB
Idle Hunter · AWS us-east-1 · Snapshot + release
−$612.40/mo
6x D8s v3 at <5% CPU for 30 days
Rightsizer · Azure · Downsize to B4ms
−$1,184.18/mo
Non-prod analytics cluster, nights + weekends
Schedule Keeper · GCP · Stop 19:00 → 07:00 PT
−$2,307.55/mo
3 stale NAT gateways, no traffic since May
Idle Hunter · AWS eu-west-1 · Deprovisioned
−$198.72/mo
Reclaimed this week
−$4,302.85/mo
Every entry: what was found, what changed, how much spend it reclaimed.
Start with the evidence
Find the savings before you connect anything.
Request a savings audit or demo and we'll help turn a vague cloud-cost concern into a scoped first readout. No cloud role, account, or credit card is needed to ask.
- Share the clouds you run and a rough monthly spend.
- Point us at the waste, risk, or recurring problem area.
- Tell us what a useful first readout would help you decide.
We'll reply within one business day with a practical next step — whether that means a demo, a scoped audit, or a clear reason to wait.
Low-commitment first step
Shipped this sprint
Recede acts in the clouds your platform team already runs — across AWS, Azure, and GCP, with an audit log for every change.
- Continuous multi-cloud monitoring
- Idle / oversized rightsizing
- Scheduled shutdowns by env tag
- Weekly savings reports
- Anomaly alerts
- Agent audit log
- Custom policies
- Spend forecast
- Team invites
- Per-tag cost breakdown
How it works
Four stages, none of them “wait for a quarterly review.”
Recede connects read-only, scopes itself against your tags and caps, watches continuously, and acts inside the boundaries you gave it. Out-of-bound findings sleep until you widen the rule.
01
Connect
Read-only IAM role to AWS, Azure, GCP.
02
Scope
Tag filters, accounts, regions, dollar caps.
03
Watch
Agents scan continuously, score, propose.
04
Act
In-bound changes land; out-of-bound sleep.
The agents
Three agents. They act.
Each agent stays inside its lane. Idle Hunter doesn't touch production. Rightsizer doesn't touch storage. Schedule Keeper doesn't touch anything outside tagged environments.
Idle Hunter
Finds resources that exist without being used.
Unattached volumes, idle load balancers, stale NAT gateways, orphan snapshots. Snapshots first, release second, all reversible inside a 24-hour window.
Rightsizer
Matches instance shape to actual load.
Reads 14 days of CPU, memory, and network shape on every instance. Recommends downshifts one tier at a time so RDS-class workloads and bursty services are not treated the same.
Schedule Keeper
Turns non-production off when no one is using it.
Dev, staging, QA, and analytics clusters run only during the hours engineering is on them. Schedules learn from your tag taxonomy and respect regional working hours.
Guardrails
The agents act. They don't go rogue.
Recede refuses to be useful in the bad sense. Every change is bounded by a tag filter, a dollar cap, and a rollback plan wired before the change ships. Platform teams keep control; the routine work leaves the calendar.
01
Tag filters
Agents only act inside the tag boundaries you set. Prod workloads are excluded by default — opt in explicitly per environment.
02
Per-action dollar threshold
Anything above the threshold pauses for human review. You set the number; the agent stops at it, every time.
03
Kill-safe and dry-run
Every change ships a rollback plan. Run in dry-run mode indefinitely — the ledger still accrues, but nothing moves.
Realised savings
20–45%
returned on addressable cloud spend, typical first quarter.
That headline is realised, not recommended. The weekly ledger carries the receipts: every entry names what was found, what changed, and how much it reclaimed.
< 14 days
First savings posted to the ledger
From connection, not from analysis.
0 hrs
Engineering time per week
Recede posts the weekly report itself.
FAQ
The questions a CFO or platform lead actually asks.
If something below doesn't cover what you need to know, the inbox is one click away.
Next step
Stop reading reports. Recede acts.
Start with a low-commitment savings audit. Tell us what you run, what feels expensive, and the outcome you want; we'll reply within one business day with a scoped first readout. No account or cloud connection is required to begin.