Your AWS estate, with a senior engineer watching it
Torque Assist is an AWS-native AI CloudOps assistant. Connect read-only — one account or your whole organisation — and it maps what you're running, where the money goes, what's exposed and what broke, the way a senior engineer would, without the headcount.
Read-only · No changes to your AWS account · No card required
Sees your whole estate
It reads your accounts, workloads and configuration and builds the picture your team never has time to draw — inventory, dependencies, and what's drifted.
Finds the waste and the exposure
Recoverable spend and security risk in the same pass. 15–25% of typical AWS spend is recoverable, and most accounts have at least one thing open that shouldn't be.
Investigates when things break
Point it at an alarm or a real issue and it gathers context, reasons about likely causes and leaves a reviewable timeline — instead of a blank console at 2am.
Most teams running on AWS don’t have a CloudOps function — the cloud belongs to a few engineers who have other jobs too. Torque Assist covers the ground nobody has time for: the bill, the security posture, the alert that fires at 2am. The pilot is the cheapest way to find out what it sees in your account.
What it does
Torque Assist connects to your AWS accounts read-only and works across three jobs a senior cloud engineer would do — continuously, rather than whenever someone gets a free afternoon.
- Cost. It reads your bill alongside your estate and flags the waste worth chasing — idle and oversized resources, and spend nobody can put a name to.
- Security. It scans configuration for exposure — public storage, open ports, over-broad IAM, missing MFA and key hygiene — and ranks what it finds worst-first.
- Incidents. Give it a real issue or a noisy CloudWatch alarm and it investigates: pulls the relevant logs, metrics and recent changes, reasons about likely causes, and lays out next steps.
Beyond the review, it runs day to day: alerts from CloudWatch, Alertmanager or any webhook land in one inbox, deduplicated and grouped into issues. You can ask it questions mid-investigation or weeks later. And when it can see a fix, it proposes one — and waits for a named person to approve it.
How the pilot works
- Connect, read-only. Launch a CloudFormation stack in one account, or roll it out across your whole AWS organisation. Torque Assist can read logs, metrics and configuration — it never makes changes.
- It goes to work. The assistant maps the estate, then runs its cost and security passes. You can also point it at a live issue and watch it investigate.
- We walk you through it. You get a prioritised report — what it found, ranked — plus a 30-day action plan. We go through it together on a call.
- You decide. Keep the findings and act on them yourself, or continue with Torque Assist running. If you continue, there’s an optional AWS Marketplace private offer.
Who it’s for
Teams already on AWS where the cloud is owned by people who have other jobs too — a CTO, Head of Engineering, DevOps lead or platform engineer. Usually somewhere between “the bill keeps creeping up and nobody knows why” and “we’d find out about a public bucket from a customer.” If that’s recognisable, the pilot will tell you where you actually stand.
If you want people behind it, not just software
Plenty of teams don’t have an SRE, a DevOps hire or anyone whose job is AWS — and the honest answer isn’t always a tool. The pilot is the free first step either way: it tells you what’s actually in your account before anyone quotes you for anything.
If what you need is ongoing cover, that’s our managed AWS service — the same assistant, with AWS-certified engineers behind it, business-hours or 24/7. Start with the pilot; decide afterwards.
What the pilot typically surfaces
- Where the money actually goes — by workload, service and tag, not just a bill total
- Idle and over-provisioned EC2 and RDS, and steady-state compute with no Savings Plan cover
- Public S3 buckets, root accounts without MFA, and security groups open to 0.0.0.0/0
- Over-broad IAM roles and stale access keys that should be scoped down or removed
- Reliability gaps — single-AZ NAT gateways, missing failover, unmanaged snapshot sprawl
- For a live issue: the likely cause, the blast radius, and what to fix now vs. harden later
What you get back
A practical, prioritised report — not a 200-page data dump. Below is an illustrative example of the one-pager you'll receive.
Torque Assist
AWS Account Review
15–25%
Recoverable spend
18
Findings
Public S3
Top risk
Top findings
- High Public S3 bucket exposed to the internet
- High Root account without MFA; access key unrotated for years
- Medium Idle / over-provisioned EC2 & RDS — right-sizable
- Medium No Savings Plan / RI coverage on steady-state compute
- Medium Single-AZ NAT gateway — no failover path
- Low Untagged spend that can't be allocated to a team
30-day action plan
| Action | Effort | Impact |
|---|---|---|
| Enable Public Access Block on all S3 buckets | Low | Closes data exposure |
| Enforce root MFA and rotate stale access keys | Low | Closes account takeover |
| Apply Savings Plan to steady-state EC2 | Low | 8–15% of spend |
| Right-size idle EC2 / RDS instances | Medium | 10–20% of spend |
Illustrative example. Cost figures are shown as ranges; your report uses your actual account data.
Read-only by design
Torque Assist connects to a single AWS account with a read-only role. It can read your data but never change it, and you can revoke access at any time. If you'd rather continue afterwards, there's an optional AWS Marketplace private offer.