ozeida Quality Control: your systems, explained in one letter.
ozeida Quality Control is the observability and quality product by ozeida (Saltillo, Coahuila, Mexico). It grades every system from A to F, shows every alarm across every region with its full current state, and explains what's going on in plain language with an AI assistant. It runs serverless on AWS and reads CloudWatch without replacing it.
Updated
Why it exists.What matters, in plain sight.
A–F grades
Every system gets a letter and a word (A · Very good, B · Good, C · Fair, D · Problems, F · Urgent). Anyone understands how it's doing without reading a dashboard.
Technical, made simple
Clear dashboards for the engineering team: a one-sentence conclusion on top, what matters first, detail one click away.
Alarms with no surprises
Every alarm, across every region, with its full current state; no truncation, no cut-off lists.
An assistant that explains
Ask in your own words. The AI assistant answers in plain language, with a spending cap and without sending more data than needed.
Product.Project quality at a glance.
Project quality
Illustration with sample data
- A Very good
- B Good
- C Fair
- D Problems
- F Urgent
AVery good
Website
BGood
Inventory
BGood
Billing
CFair
Mobile app
AVery good
Customer portal
DProblems
Help desk
2
Every alarm with its current state, complete and uncut.
The quality view is designed for a non-technical person: a letter, a word and the date of the last evaluation per system, with a trend. The engineering team gets its own per-service dashboards (Lambda, API, databases, CDN, queues) under the same rule: conclusion first, detail after.
Who it's for
Leadership and non-technical teams
See how every system in the business is doing without asking anyone to translate.
Small engineering teams
All alarms and per-service dashboards in one place you don't have to host, maintain or patch.
Companies on AWS that have Grafana and nobody reads it
An alternative with A–F grading and plain-language explanations that can run alongside Grafana or replace it.
How it works.Three steps, no hassle.
- 1
We connect your systems
We gather the signals your AWS projects already produce (CloudWatch metrics and alarms) with read-only permissions.
- 2
We give each one a letter
We evaluate every system and explain it on an A–F scale, with a date and a history.
- 3
You know what to do
The assistant tells you what's happening and what to look at first.
How it's built
Serverless
Static frontend behind CloudFront, API and collectors in Rust on Lambda, data in DynamoDB. Nothing to patch.
Read-only on your cloud
It reads CloudWatch; it doesn't store its own long time series or define alarms. Your alarms stay where they are.
Identity-based access
Login through your company's identity provider; each role sees only what it should, and the boundary is enforced in the API, not in the screen.
AI with limits
The assistant uses models through Amazon Bedrock, receives the minimum data, and has a monthly budget with automatic cutoff.
Product status
ozeida Quality Control is in development. We built it to operate our own systems; it is the tool we use in our SRE and observability service. If you want it for your company, request early access: we'll tell you where it stands, what it includes today and what it doesn't, and let you know when you can get in.
ozeida Quality Control FAQ
What is ozeida Quality Control?
It is ozeida's observability and quality product: it grades every system from A to F, shows every alarm in full, and explains the state in plain language with an AI assistant. It runs serverless on AWS.
Does it replace CloudWatch?
No. It reads CloudWatch and presents it in a way people understand; metrics and alarms keep living in CloudWatch.
Is it a Grafana alternative?
For teams on AWS that need non-technical stakeholders to understand system health, yes: A–F grading, complete alarms and plain-language explanations, with no instance to maintain. It can run alongside Grafana or replace it. It doesn't copy anything from Grafana.
What does the A–F scale mean?
A · Very good, B · Good, C · Fair, D · Problems, F · Urgent. Always the letter and the word, never color alone, so it reads without ambiguity.
What data does the assistant send to the AI?
Only what's needed to answer the question (state, system names, summarized metrics), never raw logs. It uses Amazon Bedrock, which per AWS public documentation does not train on customer data, and has a budget with automatic cutoff.
How much does it cost?
It has no published price yet. The serverless architecture is designed so infrastructure cost stays low and proportional to use; pricing will be published when it leaves early access.
Does it work outside AWS?
Not in this version: it reads CloudWatch metrics and alarms. Other clouds will be evaluated based on demand.