# Managed QA testing service for AI-written code

> Managed QA for startups with no QA team: we check the code and tests your coding agents write. Pilot £1,490 + VAT or $1,950, refunded if under 3 issues.

Source: https://cueapril.com/managed-testing
Updated: 2026-10-15
Index: https://cueapril.com/llms.txt

**Your coding agents write the code and the tests. We check both.** Cue April’s managed QA service tests the pull requests that touch your billing, permissions or other high-risk code against requirements your team approves. A person reproduces every failure, and mutation testing shows which deliberate faults your tests miss.

£1,490 or $1,950 excluding VAT (20% for UK customers), fixed, for 10 working days. Money back if your team accepts fewer than 3 issues, on the guarantee terms below. No automatic subscription. A Cue April engineer scopes every pilot and reviews every failure.

The [owned local sample](https://cueapril.com/sample-report) records 19 task executions: 9 passed and 10 failed. Twelve generated mutants left six survivors. The fixture contains deliberately seeded shipping-boundary and tenant-isolation faults, and no customer data, cloud execution or model-generated results.

Cue April managed testing is a UK-based managed QA service for SaaS startups without a QA team. We check the code and the tests your coding agents write, starting with a paid pilot on one critical area: 10 working days for £1,490 or $1,950 excluding VAT (20% for UK customers), refunded if your team accepts fewer than 3 issues.

Best for seed to Series B SaaS teams of 10 to 60 engineers who use Claude Code, Cursor or Copilot on TypeScript and have no dedicated QA. We serve teams in the UK and the US.

## Managed QA for teams without a full-time tester

- Coding agents such as Claude Code, Copilot or Cursor write most of your code and most of your tests.
- Nobody tests full time, so checks before a release depend on who has a spare hour.
- Your billing, pricing or permission rules change often, and a mistake there reaches customers.
- A customer, investor or auditor has asked how you test changes.

Code and tests written from the same assumption can agree while a requirement is broken. We start with expected behaviour your engineer has reviewed.

## How the two-week pilot works

- **Day 1:** In a 45-minute kickoff call we agree the area that would hurt most if it broke, such as billing, permissions or customer data, and what we’ll measure.
- **By day 3:** We connect your repository, set up a runner for your stack and confirm we can build and run your code. If we can’t by then, we stop and refund you.
- **By day 4:** We write requirements for that area from your specs and tickets. Your engineer approves them, so the checks test what the code should do, not only what it does today.
- **Days 4 to 9:** We run unit, API, property-based and mutation checks, plus your browser tests where supported. A person reproduces every failure before you hear about it.
- **Day 10:** You get the report, each issue with its evidence, regression tests as a pull request and what the monthly service would cost at your volume. Then you decide.

## What you get from the pilot

- Requirements for your riskiest area, written from your specs and tickets and approved by your engineer.
- Up to 20 pull requests checked against them, with every failure reproduced by a person.
- A mutation report that shows which deliberate faults your tests didn’t notice.
- Regression tests as a pull request your team owns.
- A written report, a 45-minute readout and a projected monthly cost.

## What we do, and what stays with you

We:

- Set up and run the checks
- Write requirements for you to approve
- Reproduce every failure
- Write regression tests
- Report what we found

You keep:

- The final say on what the code should do
- Which issues matter
- Your merge rules
- The release decision

You approve the plan, the targets and any action in your environments before anything runs. Cue April does not decide releases: your team reviews the evidence and makes the release decision. Read [how the checks work](https://cueapril.com/how-it-works), including [mutation testing](https://cueapril.com/testing/mutation-testing) and [property-based testing](https://cueapril.com/testing/property-based-testing), or our guide to [testing AI-generated code](https://cueapril.com/guides/testing-ai-generated-code).

## Managed QA testing pricing

Most engagements start with the two-week pilot, and Custom starts with a four-week pilot. Prices exclude VAT. UK customers pay VAT at 20% on top, so the pilot costs £1,788 with VAT.

### Two-week pilot: £1,490 or $1,950 once, excluding VAT

For trying us on the code you can least afford to break.

- 1 repository and 1 critical area, with requirements you approve
- Up to 20 pull requests that touch your area
- Mutation and property-based checks on your area
- Every failure reproduced by a person
- Regression tests as pull requests you own
- Written report and a 45-minute readout

Money back if we find fewer than 3 issues your team accepts. Credited to your first month of Verify if you start within 14 days. Nothing renews.

### Verify: £1,900 or $2,500 a month, excluding VAT

For teams without a full-time tester.

- 1 repository, or a front end and its API
- Up to 2 critical areas, with requirements you approve
- Up to 40 pull requests a month that touch them
- Mutation and property-based checks on each area every month
- Every failure reproduced by a person
- Regression tests as pull requests you own
- A weekly written summary

Month to month. Cancel before your next billing date. Start within 14 days of the pilot and its fee is credited to your first month. The pilot never starts a subscription automatically.

### Custom: Quote starting with a 4-week pilot at £4,500 or $5,900 excluding VAT

For several services, an existing QA team, or checks in your own cloud.

- More repositories or services
- Your supported existing test suites run unchanged
- A runner in your own cloud where it qualifies
- Azure DevOps and other stacks, quoted separately
- Annual terms or your contract

Terms agreed with you.

### Evidence add-on: £600 or $790 a month, excluding VAT

For teams preparing for SOC 2 or answering security questionnaires. Every pull request we check gets a record: the approved plan and who approved it, the revision tested, each check’s outcome and the artifacts, with checksums. You get a monthly pack, records kept for 12 months and export at any time. It supports your change records. It is not an audit opinion.

### On every plan

- No per-seat, per-run or AI credit charges.
- A pull request counts once, including re-checks until it merges. Docs-only pull requests don’t count. One over 1,000 changed lines counts as two, and we tell you first. One where you mark our finding as not reproducible doesn’t count.
- If you get close to your allowance, we tell you and agree any change first. We never bill overage automatically.
- From month 3, pay for 12 months upfront and get 15% off.
- We’re built for JavaScript and TypeScript web and API code on GitHub, with Jest or Vitest, Playwright, Cucumber, fast-check and Stryker. Using something else, or Azure DevOps? Tell us in your request.

## What managed QA costs next to hiring a QA engineer

In the UK, a QA engineer contractor costs a median £438 a day, so a five-day week is £2,190, and a permanent QA engineer earns a median £50,000 a year (ITJobsWatch, six months to 12 October 2026). With employer National Insurance and the minimum pension, that hire costs about £4,800 a month, our estimate from GOV.UK employer rates for 2026 to 2027. The pilot is £1,490 for 10 working days, and Verify is £1,900 a month, both excluding VAT (20% for UK customers).

In the US, the median software QA analyst or tester earned $104,300 in May 2025 (US Bureau of Labor Statistics). With benefits at the private-sector average, one hire costs about $12,400 a month, our estimate from BLS employer-cost data. Verify is $2,500 a month.

For comparison, Kodus publishes $2,000 a month for one product with up to five critical user journeys, and Bug0 $2,500 a month for up to 500 end-to-end user flows (checked 12 October 2026). Both cover end-to-end user journeys, which we don’t. Verify is $2,500 a month for code-level checks on up to 40 pull requests across 2 critical areas.

We compare costs here, not savings: what you save depends on your team. Sources: [ITJobsWatch QA engineer contract rates](https://www.itjobswatch.co.uk/contracts/uk/qa%20engineer.do), [ITJobsWatch QA engineer salaries](https://www.itjobswatch.co.uk/jobs/uk/qa%20engineer.do), [GOV.UK employer rates 2026 to 2027](https://www.gov.uk/guidance/rates-and-thresholds-for-employers-2026-to-2027), [BLS software QA analysts and testers](https://www.bls.gov/ooh/computer-and-information-technology/software-developers.htm), [BLS employer costs for employee compensation](https://www.bls.gov/news.release/ecec.nr0.htm), [Kodus managed QA](https://kodus.io/managed-qa/), [Bug0 pricing](https://bug0.com/pricing).

## Managed QA as a service vs QA outsourcing vs a QA hire

Managed QA and QA as a service usually name the same model: a specialist team runs your testing for you. Here is how ours compares with QA outsourcing, end-to-end testing services and a first QA hire.

| Option | Who does the work | What you pay for | Best when |
| --- | --- | --- | --- |
| Managed QA from Cue April | Our team, using Cue April to plan and run the checks | A fixed monthly price, £1,900 or $2,500 excluding VAT (20% for UK customers), after a two-week pilot | Coding agents write most of your code and nobody tests full time |
| QA outsourcing | An agency’s testers | Day or hourly rates for the people assigned to you | You need people to run manual or scripted test cases at volume |
| End-to-end testing services | A vendor’s platform and team, usually covering browser user journeys | A subscription, often priced by tests or user flows | You want end-to-end coverage of your user journeys |
| Your first QA hire | Your own QA engineer | A salary: a median of £50,000 in the UK and $104,300 in the US | You want someone inside the team every day |

## Money back if we find too little

If your engineer accepts fewer than 3 issues from your pilot, we refund the full fee you paid, including any VAT. An issue is either a defect, where your code doesn’t do what a requirement your engineer approved says, or a test gap, where we made a small deliberate change to a copy of your code and none of your tests noticed. It must be in the area we agreed, reproducible from the evidence we send and not already in your tracker. Duplicates count once. Your engineer accepts or rejects each issue, with a one-line reason for any rejection.

You qualify if you give us access within 5 working days of signing the pilot order and the area has code and tests we can run. There’s one refund per company. If we can’t get your code building and running within 3 working days of access, we stop and refund the full fee.

To claim, email hello@cueapril.com within 7 days of your readout, and we pay within 14 days to the card or account you paid from. Whatever happens, you keep the report, the requirements and the tests we wrote. If we refund your pilot, there’s no fee to credit against Verify.

## Your code and data

We agree the execution provider, isolation, region and supported tools for your evaluation before testing starts. Your code runs in the qualified environment named in that setup. Repository credentials stay in a separate broker. You approve the plan, targets and permitted actions before anything runs.

Before we see any code, your pilot order confirms the sandbox provider and region, the AI model provider and its data terms, and our subprocessors, and we sign a UK GDPR data processing agreement. When you leave, we delete our copies of your code within 30 days, so export your reports before you go.

We don’t hold SOC 2 or ISO 27001 yet. If your policy requires a SOC 2 report from every vendor with code access, we’re not the right fit today.

## What this managed testing service does not do

Passing checks do not certify a release, and a mutation score describes only the code we tested. API checks are not a security assessment, and we don’t run performance or load tests. We don’t write or maintain end-to-end suites of your user journeys. We don’t offer an uptime guarantee or a release SLA, and your team makes every release decision.

## Questions to ask any managed QA provider, and our answers

### What is managed QA testing, and who is it for?

Managed QA testing means a specialist team runs your testing for you. Cue April’s managed QA service is built for SaaS startups with no dedicated tester: teams of 10 to 60 engineers shipping TypeScript with coding agents. We check pull requests against requirements you approve, reproduce every failure and hand you regression tests, starting with a two-week pilot at £1,490 or $1,950 excluding VAT (20% for UK customers).

### How much does managed QA testing cost?

Cue April’s managed QA starts with a two-week pilot at £1,490 or $1,950, with money back if your team accepts fewer than 3 issues. Verify then costs £1,900 or $2,500 a month, month to month, for up to 40 pull requests across 2 critical areas. The Evidence add-on is £600 or $790 a month. Prices exclude VAT, which UK customers pay at 20%.

### How do you check tests written by AI coding agents like Claude Code or Cursor?

We start from what the code should do, not from what the agent wrote. Your engineer approves requirements from your specs and tickets, and we check each pull request against them. Mutation testing makes small deliberate changes to a copy of your code to show which ones your tests miss. A person reproduces every failure before you hear about it.

### Is managed QA cheaper than hiring a QA engineer?

Usually, for teams that need checking rather than a full-time tester. Verify costs £1,900 or $2,500 a month excluding VAT (20% for UK customers). A UK QA engineer on the £50,000 median costs about £4,800 a month with employer costs, and a US hire about $12,400. A hire also writes end-to-end tests and joins your planning, which we don’t do.

### How fast can you start, and what does the pilot include?

After a 30-minute scoping call, the pilot starts once you’ve signed the order, paid the fixed fee and given us access. It runs for 10 working days on 1 repository and 1 critical area, covers up to 20 pull requests and ends with a written report, a 45-minute readout and regression tests as a pull request.

### Can we see an example of what you report?

Yes. Our sample evidence report shows real output from our own sample project, with faults we planted on purpose: 19 task executions, 9 passed and 10 failed, and 6 of 12 deliberate code changes the tests didn’t notice.

### Who owns the tests and requirements you write?

You do. We hand over regression tests as pull requests to your repository, along with the requirements your engineer approved, and they stay yours if you leave.

### How quickly do you respond?

During the pilot you hear about each issue once a person has reproduced it, and you get the full readout on day 10. On Verify you get a weekly written summary. We don’t promise an emergency response time.

### Do you need access to our source code?

Yes. We need read access to the repository in scope, and permission to comment on pull requests and open pull requests for the regression tests we write. Your code runs only in a disposable sandbox or your own VM, never on our control servers, and repository credentials stay in a separate broker. We sign a UK GDPR data processing agreement before we see any code.

### How is this different from outsourcing QA to an agency?

QA agencies usually bill testers by the day to write and run test cases. We check the code changes your agents make against requirements your team approves, show where your tests miss faults and hand you the regression tests. You pay a fixed monthly price, with no day rates, seats or run charges.

### Why would we trust an AI to check code an AI wrote?

Because the checks don’t start from the code. We write the expected behaviour from your specs, tickets and rules, and your engineer approves it before any check runs. A model response is never treated as a pass. A person reproduces every failure, and you can rerun each issue yourself from the evidence.

### What if you don’t find anything?

You get your money back, on the terms in the money-back section. If your engineer accepts fewer than 3 issues, email us within 7 days of the readout and we refund the full fee within 14 days. If we can’t get your code running within 3 working days of access, we stop and refund in full.

### Do you have SOC 2 or ISO 27001?

Not yet. Until then we answer your security questionnaire before you pay, sign a UK GDPR data processing agreement and share our subprocessor list. If your policy requires a SOC 2 report from every vendor with code access, we’re not the right fit today.

### Which AI model sees our code?

Your pilot order names the model provider and its data terms, and it’s on our subprocessor list. Whatever the model writes is a draft: checks come only from requirements your engineer approved, and a person reproduces every failure.

### Does this replace our QA engineers or our end-to-end tests?

No. We check code changes against requirements you approve, run your existing browser tests where we can, and show where your tests miss faults. We don’t write or maintain an end-to-end suite of your user journeys. If you have QA engineers, they approve the requirements and we take on the checking, reproducing and reporting.

### What happens when a check fails?

A person reproduces it first. If it’s real, you get the failing behaviour, the requirement it breaks and the evidence to rerun it, plus a regression test where one helps. Passed, failed, blocked and unfinished checks are always reported separately, and an unfinished check is never shown as a pass.

### Which languages, test tools and repositories do you support?

We’re built for JavaScript and TypeScript web and API code on GitHub, with Jest or Vitest, Playwright, Cucumber, fast-check and Stryker. Azure DevOps and other languages go through adapters and are quoted separately, so tell us your stack in your pilot request.

### Can we cancel, and is there lock-in?

There’s no lock-in. Verify is month to month: cancel before your next billing date and you won’t be charged again. You keep everything we delivered, and we delete our copies of your code within 30 days of your leaving.

## Request a two-week pilot

Use the form at https://cueapril.com/managed-testing#pilot, or book a 30-minute discovery call at https://cal.com/cueapril/discovery.

1. We email you to book a 30-minute scoping call.
2. On the call we check we can run your stack and agree the area to test.
3. You review the one-page pilot order and our data processing agreement.
4. You sign the order, pay the fixed fee and give us access. The 10 working days start then.
