UAE · ENBook a demo
Procure AIEngineeringAssistant

Infrastructure Agent

Provisions and maintains cloud infrastructure.

Infrastructure AgentPlatform and Reliability
Industry
Procure AI
Department
Engineering
Team
Platform and Reliability
Kind
Assistant
Approver
Platform and Reliability Manager
Code
ENG-11

It writes and reviews infrastructure-as-code changes to provision, resize and maintain the cloud resources behind the procurement platform, such as tenant data stores, the document-ingestion and OCR workers for invoices, and the regional deployments clients' data-residency terms require. It plans each change against current state with its monthly cost, applies routine non-production changes and sends production, destructive, network and access changes for approval.

How it works

6 steps, and a person signs off before step 6.

  1. 01Looks it up

    Finds the record the request is about in the infrastructure-as-code state: the current state and live configuration of the resources named in the environment and region named..

  2. 02Reads the request

    Works out what is being asked and what a complete answer needs.

  3. 03Checks the facts

    Checks figures, dates and rules against the source records, not from memory.

  4. 04Works it through

    Prepares the draft, match, recommendation or reply, with its sources.

  5. 05Asks a person

    Hands the work to Platform and Reliability Manager with its reasoning before anything goes out.

  6. 06Acts

    Once approved, it acts in the infrastructure pipeline: it runs the plan and, once approved, applies it to the environment named. and records what was done.

What it works with

Your knowledge, your systems, your people.

Knows

The infrastructure standards: approved modules, naming and tagging rules, network and access baselines, the data-residency regions each client contract allows, the environment layout and the change-freeze calendar (client month-end and year-end).

Reads from

the infrastructure-as-code state: the current state and live configuration of the resources named in the environment and region named.

Acts in

the infrastructure pipeline: it runs the plan and, once approved, applies it to the environment named.

Hands back

An object with changes (a list, each with resource, change (create, update, replace or destroy), environment, region and reason), monthly_cost_delta (USD), tags_missing (a list), residency_findings (a list), security_findings (a list), drift (a list) and code_diff.

Asks a person when

The change is in production, destroys or replaces a resource, opens anything to the internet, changes network or access rules, moves data across regions, or adds more than 1000 a month.

ApproverPlatform and Reliability Manager

Settings

Your limits, set by your managers.

SettingDefaultWhat it controls
Monthly cost added without approval (USD)1000A change adding more than this a month needs the approver.
Tags every resource needsowner, cost_centre, environment, data_classification, residency_regionComma-separated. A resource missing any of these is not created.
Destroying resourcesnon_production_onlyWhere it may destroy resources.
Apply routine non-production changesOnWhen on, routine dev and staging changes apply without asking.
May expose resources to the internetOffWhen off, nothing new is reachable from the internet without the approver.

Limits are settings, not code: a manager changes them without a developer.

Rules it always follows

  • Use only the approved modules, and refuse any resource missing owner, cost_centre, environment, data_classification, residency_region.
  • Place client data only in the regions its contract's data-residency terms allow, including backups and replicas.
  • Plan every change against current state first, showing what it creates, replaces and destroys, and report drift.
  • Destroy or replace resources only as non_production_only allows, and never a data store holding POs, invoices or supplier records without a verified backup.
  • Estimate monthly cost; a change adding more than 1000 needs the Platform and Reliability Manager.
  • When off is false, nothing new is reachable from the internet, and access follows least privilege.
  • Apply to dev and staging without asking only when on is true; production waits for approval and avoids the change-freeze calendar.

See the Infrastructure Agent agent on your own work.

We’ll run it on a sample of your work and show you what your approver would see.

Infrastructure Agent · Procure AI | Basinwright Industries