Everything between “we need a database” and “it's running, and we can prove it.”
Most of the work in changing infrastructure is not writing the change. It is the reviewing, the approving, the checking afterwards and the remembering later. Intentric is that middle — the part every team rebuilds and nobody wants to own.
Six things your team stops building for itself.
Each one is something engineering organizations construct in-house, again, at companies that do not sell infrastructure tooling.
Requests in plain language
Someone describes what they need: the service, the size, the region, the constraints.
- No vendor tooling to learn first
- No module to find first
Changes worth reviewing
One specific change comes back, with its cost, footprint and risks stated up front.
- Reviewing it is reading a decision
- Not conducting an investigation
Your rules, every time
Permitted regions, spending ceilings and what is too dangerous to wave through, applied before anyone is asked to look.
- No exception path for a persuasive request
- Your own policy code, or ready-made packs
- Yours can only tighten the platform’s
Approval that means something
A named person authorizes each change, and what was approved is what runs.
- Deletes and replacements always need a person
- Anything holding data needs an elevated approver
- Auto-approval only where your admin turns it on, for plans that only add new resources
Confirmation, not assumption
What is running is checked against what was asked for, then again on a schedule, so you hear about drift before your customers do.
- A correction shows its estimated monthly cost from list prices
- And waits for a person
A record you can hand over
Who asked, who approved, what changed, what resulted — written as it happens, readable by someone who was not there.
- Exported as CSV or JSON
- Or sent to your webhook
What your team actually looks at.
One console, built for the engineer who wants something and the person who has to be comfortable saying yes.
What you run
Everything under management, current and grouped the way your teams are.
What’s proposed
Pending changes with their cost, footprint and risk, ready to approve or refuse.
What’s happening now
Work in progress, live, without anyone tailing a log.
What’s moved
Where reality has stopped matching intent, and the correction waiting on a person.
What happened
The searchable record of every request, decision and result.
Who can do what
Members, roles, single sign-on with user provisioning from Microsoft Entra ID or Okta, passkeys, and separate environments for separate stakes.
What needs you
An inbox for the plans waiting on your decision and the outcome of yours, by email too, with Slack and Microsoft Teams channels for the team.
What it costs
The activity measured and the estimated monthly run rate, shown as two figures rather than one total.
Intentric AI
An assistant beside every screen that proposes a request or explains a deployment. It cannot approve or apply anything.
The things teams actually stand up.
The services behind most real infrastructure requests, each with the same rules, approval and confirmation. Each line says which clouds have it.
Databases
- Managed PostgreSQL on AWS, Azure and Google Cloud, sized and configured for the availability you asked for.
- MySQL on AWS and Google Cloud.
- SQL Server on AWS, Express or Standard edition: the administrator password stays in your cloud account and the platform has no permission to read the data in it. Deleting it keeps a final snapshot.
- Serverless NoSQL tables on all three, which your services reach by identity and never by a key.
- A managed in-memory cache on all three.
- A data warehouse on AWS, with its administrator password kept in your account. Deleting it destroys the data.
Compute
- Containers on all three, and scheduled jobs that run a container image to completion on a cron schedule with no way in.
- Functions that run a container image once per request, private unless you make them public, on AWS and Google Cloud.
- Linux virtual machines in a private network with no public address and no key, reached only through your cloud’s own session service.
- Managed Kubernetes on AWS and Google Cloud, private unless you name the address ranges that may reach it, and applications that run on it in a namespace of their own.
- A private registry for the images they run from, and hosting for AI agents, on all three.
Storage
- Object storage for what your application keeps, on all three.
- Network file shares that your virtual machines mount, and, on AWS, your containers.
- Secrets whose value only you ever set: Intentric creates one empty and never sees what you put in it.
Networking
- A private network: an address range cut into public and private subnets, with outbound internet through a NAT. It can also list the address ranges and ports it may reach and deny everything else outbound.
- Load balancing, DNS and content delivery on all three, and a web application firewall in front of a load balancer on all three.
- An API gateway in front of a function, on AWS.
- Private DNS zones on all three, which answer only inside the one network you connect them to and start empty.
- Private endpoints on AWS and Google Cloud, and a site-to-site VPN on AWS.
- TLS certificates on AWS and Google Cloud.
Security and identity
- Encryption keys you own, on AWS and Azure, rotated on a schedule. Intentric never chooses who may use one, and the key is not yet connected to databases or buckets.
- A user sign-in directory for your application’s own users, on AWS, with an authenticator app as the only second factor. Those users live in your AWS account, not with us.
- A pipeline role on AWS that lets one GitHub Actions environment (or one protected GitHub branch) of one repository, or one protected GitLab.com branch, deploy one container service or the code of one function without a long-lived key.
- On Google Cloud, a pipeline role trusts one workload’s service account and deploys one container service, and a pipeline can be pinned to one workflow.
- A pipeline role’s trust is exact, and every plan that carries one waits for an elevated approver.
Operations and cost
- Alerts that tell a topic when a condition you choose is met and when it clears, on AWS and Google Cloud.
- Budgets that tell someone when what a set of your resources costs crosses percentages of a monthly limit, on AWS and Azure. A notice, not a cap: nothing is stopped when it is exceeded.
- Backup plans on AWS, taken on a schedule into a vault of their own; you restore in the cloud’s own console.
- An audit log export on AWS that records every change made in your cloud account and keeps it where the platform cannot delete it.
Messaging and email
- Queues and topics for the work your services hand to each other, on all three.
- Transactional email on AWS and Azure. Google Cloud has no such service.
Something missing?
Tell us what you need: coverage grows with what customers ask for. Not every service is on every cloud, and Intentric tells you before it plans, not when it applies. The full list by cloud is in the availability matrix.
This does not ask you to abandon your process.
Infrastructure that lives in Git keeps living in Git. Environments stay separate. Roles stay enforced. Intentric fits the shape your organization already has.
Git stays Git
Bind a definition to a repository on your git host: a change there can be proposed as a plan automatically, and goes through the same approval.
Separate environments
Different parameters and different stakes, kept apart.
Sign-in you already use
Google, Microsoft, GitHub, GitLab or Bitbucket, and passkeys with recovery codes.
Single sign-on
OIDC or SAML, with user provisioning from Microsoft Entra ID or Okta, for larger organizations.
Machine access
For your own automation, held to the same rules as people.
Terraform you already wrote
Import it as a proposed definition, with a report of anything it could not carry over.
Infrastructure you already run
Brought under management as a governed change of its own: you choose what, and an elevated approver signs off before anything is claimed.
Copies you control
One-way copies of your definitions in an S3 bucket, Google Drive folder or OneDrive library. Intentric stays the record; nothing is read back.
The infrastructure code is yours the whole time.
Intentric produces standard, inspectable OpenTofu/Terraform-compatible infrastructure code — the format your team already reads and reviews — or a Pulumi project for deployments that use that engine. Export it whenever you want, in full, without asking us.
- Nothing proprietary sits between you and your own infrastructure
- Your team can read it, diff it and run it entirely without us
- Leaving costs you the workflow, not the estate
- Your cloud bill is yours, at cost — we never resell or mark it up
See it against something you actually run.
The sandbox runs the whole loop against a simulated AWS with no card and no cloud account. To see it with your own estate, book a demo.