Shared clusters, split by customer
Every tenant's share of one EKS cluster, by actual consumption.
FinOps without tagging. Even for AI.
Cloud Intelligence™ is a FinOps platform that allocates 100% of your cloud spend to the customers, teams and features that used it, without tags. It reads runtime traffic and splits shared compute, storage, network, databases and AI by actual usage. No code changes.
Patent-pending runtime attribution · Free to start, no credit card · Live in 15 minutes
Attribute every dollar: 100% of spend allocated at runtime
Allocation · last 30 days · all clouds
$2,418,906 100% allocated
eBPF sensor · live
Tagging program · 43% of spend covered18 months
Cloud Intelligence™ · 100% of spend allocated1 day
Customers 46%Teams 31%Features 14%Agents 9%
Kubernetes (shared EKS)
$1,184,310
by runtime consumption
Network egress + NAT
$342,007
traced to workload
Databases + Kafka
$596,220
split by queries
OpenAI + Anthropic
$296,369
tokens per customer
Explored 0 of 3· Insights 0/6
Day one with Cloud Intelligence™
Every dollar has an owner. Tags optional.
100%
of spend allocated
0
tags needed
15min
to deploy
you can't tag your way to 100%
Your tagging campaign will never finish. Stop waiting on it.
Cloud Intelligence™ reads runtime traffic and allocates every dollar by what actually happened. Even mature programs track 43% of cloud cost at the unit level (Gartner, 2025). The other 57% is shared infrastructure, network and AI, with no tagging surface. Keep your tags. Lose the dependence on them.
See how runtime attribution worksWhat the bill contains · where tags reach
Compute (tagged)
Tag
Managed DBs
Partial
Shared Kubernetes
Runtime
Network, NAT, egress
Runtime
Kafka, queues, caches
Runtime
LLM tokens, GPUs
Runtime
Self-managed on EC2
Runtime
Reachable by tags
43%
Reachable by Cloud Intelligence™
100%
you can't tag a customer
Know what every customer costs you. Not just what they pay.
Cloud Intelligence™ reads the customer ID in the traffic and gives every account a cost-to-serve, a COGS line and a margin, with human and agent usage split. Customers share clusters, hit APIs and trigger inference without leaving a tag behind.
Pick a customer. We'll show you what they cost youCustomer
Northwind Retail
Cost to serve
$80,761
COGS line
$68,420
Gross margin
32%
Cost by feature · detected from API paths
/api/checkout$31,210
/api/search$21,604
assistant (LLM)$18,390
/api/reports$9,557
Human usage 62%Agent usage 38%
“You can't tag a customer in a multi-tenant environment. Attribute™ finally shows us what each customer costs.”
what you get
Cloud Intelligence™ sees what billing exports and CUR miss.
Cloud Intelligence™ reads runtime traffic and answers the question the bill can't. Billing exports describe resources. They never say who used them.
Cross-AZ and cross-region, traced
The bill nobody owns gets an owner, down to the workload.
Customer cost, auto-detected
Cost-to-serve and margin per account. No tags, no code.
Chargeback by real usage
Team reports finance trusts and engineering stops disputing.
Agents and tokens, per customer
Input, output and cached tokens, by feature and customer.
Anomalies with the cause
Alerts name the customer, the feature and the change.
what to work on next
Start the week with a list, not a guess.
Cloud Intelligence™ ranks every workload by what it will add to next month's bill if nothing changes, and names the cause. New and spiking workloads surface the week they appear, so the next thing to fix is always at the top.
Which workloads need focus?1 new and 9 spiking · ranked by monthly impact · top 3 of 10
If nothing changes, the top 5 add
+$16,557per month
4 spiking Deployments are 79% of it · 1 spiking AutoscalingRunnerSet is 21%
Spiking this week
9workloads · +$4,490 vs last week
3 of 9 bill mostly to Amazon Elastic Compute Cloud
New since last week
1workload · first seen this week
From the eBPF sensor · 9 clusters
WorkloadSignalLast 14 daysWhat changedImpact / month
Where the increase came from
Claude Sonnet 5.5 (Amazon Bedrock Edition)+$862 46%
eu.anthropic.claude-sonnet-5-5
Claude Sonnet 5 (Amazon Bedrock Edition)+$437 23%
eu.anthropic.claude-sonnet-5
Claude Opus 4.6 (Amazon Bedrock Edition)+$296 16%
eu.anthropic.claude-opus-4-6-v1
84% of the change
When it started
Since Sun 4 Oct, 09:00 UTC
In a few hours of the week
Usage vs price
input tokens standard+$617newcache write tokens standard+$522newInputTokenCount+$298new
Where the increase came from
Simple Storage Service+$402 50%
s3://ci-artifacts-us-east-1
Elastic Compute Cloud+$291 36%
arm64-big runner nodes
Data Transfer+$117 14%
cross-AZ runner traffic
91% of the change
When it started
Since Mon 5 Oct, 02:00 UTC
Ramped over a day
Usage vs price
PUT, COPY, POST requests+$318newm7g.4xlarge on-demand+$291newregional data transfer+$201new
Where the increase came from
Claude Sonnet 5.5 (Amazon Bedrock Edition)+$322 49%
us.anthropic.claude-sonnet-5-5
Claude Sonnet 4.6 (Amazon Bedrock Edition)+$184 28%
us.anthropic.claude-sonnet-4-6
Claude Opus 4.6 (Amazon Bedrock Edition)+$91 14%
us.anthropic.claude-opus-4-6-v1
81% of the change
When it started
Since Sun 4 Oct, 13:00 UTC
In a few hours of the week
Usage vs price
input tokens standard+$244newoutput tokens standard+$198newcache read tokens standard+$126new
spike daysnormal daysImpact = extra cost per month at this week's rate · click a row to investigateShow 5 more ↓
Ranked by impact
Impact per month at this week's rate puts the biggest lever first.
Reason attached
Each row names the services and models that moved the number.
New and spiking, flagged
A workload shows up the week it first runs or jumps.
One click to the owner
Open a thread on any row and hand it to the team that runs it.
showback and chargeback
Chargeback nobody argues with.
Cloud Intelligence™ allocates shared platforms, databases and clusters to teams by real usage, so engineering and finance work from the same numbers from day one. Even splits, revenue shares and headcount ratios get disputed every quarter.
Start showback from day oneShared EKS cluster · prod-us-east
$412,880 split by consumption
Even split (disputed)
Platform$103,220
Payments$103,220
Search$103,220
Data$103,220
Observed usage
Platform$61,932
Payments$194,054
Search$107,349
Data$49,545
PlatformPaymentsSearchData
“Within weeks, Attribute™ transformed our understanding of cost structures and shaped our pricing decisions.”
every workload. one method.
Attribute everything the bill contains.
Cloud Intelligence™ sees who used every workload you run, with one sensor and one method.
Coming soon
Don't see your stack?
Tell us what you run and we'll prioritize it.
Custom providersMore AI gatewaysMore data clouds
Request an integration48 shown · all integrations
observed, not inferred · patent-pending method
How attribution works without tags
1
Install in 15 minutes.
A lightweight eBPF sensor deploys to your cluster via Helm, plus a billing integration. No code or configuration changes.
2
The sensor reads what actually happened.
It observes traffic to every cloud resource, from databases to LLM endpoints. Customer IDs, headers and model calls are read directly.
3
Cost lands where it belongs.
Every dollar is attributed to the customer, team, feature or agent that drove it. Shared resources split by consumption, not estimates.
SOC 2 Type 2, ISO 27001, FinOps Certified Platform.
What does each customer cost me?Book a demoCloud Intelligence™
Stop reporting on waste. Start removing it.
Cloud Intelligence™ unifies senior cloud architects, AI agents, and deep multi-cloud data in a single platform. Run FinOps at the speed of AI.
This sprint · Payments
−31%NAT egress
✓Mapped cross-AZ traffic to 6 servicesDone
✓Moved checkout cache into same AZDone
•Reviewing S3 lifecycle rulesIn progress
FDForward deployed engineer, embedded with your team
ai spend
See what AI costs you before the invoice does.
Cloud Intelligence™ shows Anthropic, Cursor, Cowork and the models running on your cloud providers in one view. Track AI as a share of your cloud bill, see where the month is heading, and compare against companies your size.
How much of our bill is AI?Anthropic, Cursor, Cowork and AI models running on your cloud providers
How much of your bill is AIAI spendShare of cloud billPeers using AI
8.2%
of September’s cloud bill was AI, up from 0.3% in November. At this pace it reaches 8.3% in October. The median among companies your size using AI is 4.7%.
NNov
DDec
JJan
FFeb
MMar
AApr
MMay
JJun
JJul
AAug
SSep
OOct
12 months · Oct projected at this pace · band = middle half of peers, dashed = medianView AI spend
This month
$12,136
spent so far · day 7 of 31
On pace for
$56,809
+2% vs September
Of your cloud bill
8.3%
projected · 8.2% in Sep
▲claude-opus-5-thinking-high spiked to $42 on Sep 29. See the anomaly
Where you stand
8.2%On par with peers
Biggest AI service
$22kClaude Sonnet 4.6
Biggest change vs August
+$3.2kClaude Opus 5
Via your cloud
51%
Cost per task
$1.31Coming soon
Every AI source, one view
Direct providers, coding tools and cloud-hosted models together.
Share of the bill
See AI as a percentage of cloud spend, month by month.
Pace and projection
Know where the month will land by day seven.
Against your peers
Compare with companies your size that use AI.
Frequently asked
questions
What is Cloud Intelligence™?
Cloud Intelligence™ is a FinOps platform that allocates 100% of your cloud spend to the customers, teams, features and agents that drove it, without relying on tags. It reads runtime traffic and splits shared Kubernetes, network, databases and AI by actual usage.
How do you allocate cloud costs without tags?
A lightweight, read-only eBPF sensor observes traffic to every cloud resource, from databases to LLM endpoints. It reads customer IDs, headers and model calls directly, so shared resources are split by consumption, not estimates.
Do I need to change my code or remove my existing tags?
No. The sensor installs in 15 minutes via Helm alongside a billing integration, with no code or configuration changes. Your existing tags are ingested as one more signal.
Can I see cost per customer and margin?
Yes. Cloud Intelligence™ reads the customer ID in the traffic and gives every account a cost-to-serve, a COGS line and a margin, with human and agent usage split.
Does it track AI spend?
Yes. Anthropic, Cursor, Cowork and the models running on your cloud providers appear in one view, so you can track AI as a share of your cloud bill, see where the month is heading, and compare against companies your size.
You know what every customer pays. Not what they cost.
Four inputs, no signup, for the estimate. Forty-five minutes for the real number, by customer and by service.
Prefer to try it yourself? Start for free. No credit card required.
