SRE-as-a-Service

Your own SRE fleet, running by Friday.

Thalamus investigates your incidents, finds the cause, and proposes the fix — in your own instance, in the region you choose. Provisioned in minutes, not a quarter of procurement.

No card required to start. Annual contracts, invoiced on net terms.

How it works

Four steps, and one of them is waiting.

The deployment packs are already built and tested against live accounts in all three major clouds. Provisioning runs them; it does not invent anything on the day.

Create an account

Email, password, confirm the address. No sales call to get in.

about a minute

Choose a name and a region

Your instance lives at yourname.tenant.thalamusaicloud.com, in the region closest to the systems you are monitoring.

under a minute

We provision it

Infrastructure, database, hardened defaults, DNS. You watch the actual steps — not a spinner — and we email you when it is ready.

a few minutes

Connect your first service

Point it at your monitoring and your repositories. The fleet starts watching what you onboard, and nothing else.

the same afternoon
What you get

An agent fleet, not a dashboard.

Dashboards tell you something broke. Thalamus works out why, writes it down, and proposes the change — with the evidence attached.

01

Investigation

Correlates signals across monitoring, logs, traces and recent changes, then says what it thinks caused the incident and how confident it is.

02

Resolution

Proposes a fix as a pull request against your repository, with the test that demonstrates it. You review and merge — or do not.

03

Evidence

Every claim is labelled as fact, observation, inference or hypothesis. You can tell what was measured from what was guessed.

04

Data protection

Redaction, retention and subject-rights handling are built in, not a later module. Credentials are encrypted at rest with your own key.

05

Guarded autonomy

Observe, recommend, approve — then more if you want it. The fleet does not act beyond the level you set, and the level is yours to move.

06

Your cloud, your region

AWS, Azure or GCP. Data residency is a dropdown at signup, not a professional services engagement.

Pricing

Priced by the services you actually onboard.

Not per seat, not per host, not per gigabyte ingested. A service is a production thing you want the fleet to watch and act on — its own deployment, its own golden signals, its own on-call expectation. Not every repository, and not every container.

Launch
Up to 15 services
$54,000
per year
  • Full agent fleet
  • Shared infrastructure
  • Onboarding included
Get started
Team
16 to 35 services
$108,000
per year
  • Everything in Launch
  • Priority support
  • Change-window pinning
Get started
Division
36 to 75 services
$183,600
per year
  • Everything in Team
  • Multiple instances
  • SSO and audit export
Get started
Enterprise
76+ services
$280,800
per year
  • Dedicated AWS account
  • Own database
  • Any region
Talk to us

Where your data lives

Launch, Team and Division run on shared infrastructure. Your data is stored in a shared database, separated from other customers by enforced database-level access controls, and is never accessible to another customer. Your monitoring, credentials and incident history remain private to your organisation.

Enterprise runs in an AWS account created for your organisation, with its own database. No infrastructure is shared with any other customer, and you choose the region.

Growing past a tier

We true up at renewal. Never mid-term.

Onboard what you need, when you need it

Crossing a tier does not generate a charge and does not stop anything. We do not take the fleet off a service because a counter moved.

You see the number before we send it

Your console shows peak usage and what it means for renewal, throughout the term. A true-up should never be the first you hear of it.

Questions

The ones that actually get asked.

What counts as a service?

A production service you want the fleet to watch and act on — typically something with its own deployment, its own golden signals and its own on-call expectation. Not every repository, and not every container. Only services you have actually finished onboarding are counted; a half-configured one is not billed.

How long does provisioning really take?

Minutes, not days. The deployment packs are already built and tested against live accounts, so provisioning runs something that works rather than assembling it on the day. If it fails, we roll back and tell you — and nothing is charged for an instance that never came up.

Can we run it in our own AWS account?

Yes, on Enterprise. Your instance runs in an account created for your organisation with its own database, and nothing is shared with another customer. For running entirely inside infrastructure you already own, that is our managed service rather than this product — worth a conversation about which fits.

What happens to our data if we leave?

You can export it at any time, and we delete it on request. Suspension for a late invoice does not delete anything — your data is retained and comes back when the payment clears.

Do you need production credentials?

You choose what to connect. Credentials are encrypted at rest with a key held by your instance, and the fleet operates at the autonomy level you set — starting at observe-and-recommend, where it changes nothing.

Is there a free trial?

You can create an account and see the console without paying. Provisioning a real instance creates real infrastructure, so that starts a contract — talk to us and we will scope it against your actual estate first.

Next step

Bring your incident numbers, not a shortlist.

A 45-minute look at how your estate actually runs — incident volume, MTTR, what recurs, where the hours go. If there is not enough repetitive volume to justify going further, we will say so.