AI-ASSISTED TESTING FOR PUBLIC REST APIs

Load test the
whole journey.

Turn your REST API definition into connected requests and business checks. Review the scenario, run it on Loaduna, and inspect the result.

Your first workflow, set up with our team.

OpenAPI + your business rulesYou approve the scopeWe run the test

01 / INSIDE A REAL TEST

One journey.
Every step connected.

One checkout, four connected requests. Explore the reviewed scenario and a recorded result from our controlled sample app.

THE STARTING POINT

An API definition + a business rule. “Create an order for a selected product. Read it back and confirm it is the same order, with confirmed status.”

REVIEWED SCENARIO · CONTROLLED SAMPLE

A checkout, from start to finish.

Passed

Select a step to see how it connects.

Establish the session

Sign in to the synthetic account. The following requests reuse that session.

Request detail POST /login

Carry the product forward

Select a product from the catalog. Its identifier connects the catalog response to the order request.

Request detail GET /catalog

Check the business result

Create one synthetic order for the selected product. Check that the response reports success and the order is confirmed.

Request detail POST /orders

Close the loop

Read back the order just created. Check that its identifier matches and its status is still confirmed.

Request detail GET /orders/{order id}
  1. Business success

    Every response reports business success.

    All four steps
  2. Confirmed order

    The created order has confirmed status.

    Create order
  3. Same order, verified

    Read-back matches the created order identifier and confirmed status.

    Verify order
  4. Test data cleaned up

    The executor removes this run’s synthetic records.

    After the journey
  1. Validation

    Initial validation

    Passed
  2. Replay 1

    No additional AI call

    Passed
  3. Replay 2

    No additional AI call

    Passed

Same approved scenario. Both replays used no additional AI generation.

Per validation / replay
Journey requests
4
Business failures
0
Cleanup
Complete

A SEPARATE RECORDED RUN · 2 OCT 2026

The settings.
The measurements.

A short scheduled run against our controlled checkout application. Read the result together with the profile that produced it.

Arrival target
1 journey / second
Configured duration
3 seconds
Virtual-user ceiling
1
Request ceiling
20, including cleanup
Recorded result✓ Passed
Journey requests
16
12 latency samples; sign-in excluded
Observed p95 latency
3.94 ms
Only 12 samples on a local fixture
Reported duration
3.114 s
Configured run window: 3 seconds
Business failures
0
0 scenario failures · 0 dropped iterations

✓ Cleanup completed 1 additional request

A small functional demonstration. Shared development infrastructure, not a production capacity benchmark. The latency values do not predict performance under sustained load.

How to read these numbers

This is a separate execution from the validation and replays above. The arrival target counts complete journeys, not individual HTTP requests. The 16 journey requests include sign-in; the 12 latency samples exclude it. Cleanup is reported separately. The p95 is a percentile of this small sample, not an SLA or a statistically robust capacity estimate.

See the workspace report
Loaduna report interface showing the recorded sample checkout: 12 measured requests, p95 latency 3.94 ms and a passed functional fixture result.
Current report interface, rendered with the recorded fixture’s aggregate measurements and sample identifiers. Step timings were not retained in this record and are shown as unavailable. This is not a customer screenshot.

02 / CHECK THE ACTUAL OUTCOME

The request worked.
Did the workflow?

An HTTP 200 can still carry the wrong result. Check the returned data and the business outcome alongside the response.

See what the recorded test caught

In a separate controlled test, all three requests returned HTTP 200. One business check failed because the response contained ok: false instead of ok: true. Cleanup completed.

HTTP RESPONSE✓
200 OK
{ "ok": false }
Response received
BUSINESS CHECKExpected success.
Received failure.

Recorded failure example · Controlled test application

03 / FROM CONTEXT TO EXECUTION

From your API.
To a test you can use.

You define success. Loaduna helps turn it into linked requests and checks.

01 YOU

Bring the goal.

Bring OpenAPI 3.0, a critical workflow and the expected result. No agent is required for this path.

02 AI + YOU

Shape the scenario.

AI connects the requests and checks. You review the proposal and approve the scope.

03 LOADUNA

Run. Inspect. Repeat.

Execute on our infrastructure, follow the result, and reuse the approved scenario.

Start with OpenAPI 3.0. JSON or YAML · Public HTTPS REST API · Guided first setup

Check the fit

04 / CLEAR BOUNDARIES

Your target.
Your rules.

Agree what can be tested, how far the run can go, and when it must stop.

Data & access
Execution scopeYOU DECIDE
TargetVerified API
RequestsAgreed limit
DurationAgreed window
✓ Reviewed before executionIllustrative scope

Keep a test worth repeating.

The same approved scenario can run again without another AI generation.

PRICING BY AGREED SCOPE

Know the scope.
Then the price.

Preparation and execution are quoted separately. You agree the scope and price before paid work.

01 / PREPARATION

A reviewed scenario

API context, connected steps, business checks and the initial review.

02 / EXECUTION

An agreed run profile

The target, traffic limits, test window and number of runs.

Reusing an unchanged scenario needs no new AI generation. Execution still uses resources.

05 / GOOD QUESTIONS

The details,
when you need them.

What can I test?

Public REST APIs over HTTPS. A scenario can keep a session, carry values between requests and check the business result. We review supported authentication and inputs during setup. Browser UI testing, private networks and arbitrary protocols are outside the current scope.

What do I need to get started?

An OpenAPI 3.0 definition in JSON or YAML, the workflow you want to test, and what a correct result looks like. Selected source files can add context; an agent or full repository access is not required for this workflow. If you have no API definition, contact us to review the available inputs. Your first workflow is set up with our team; public signup and online checkout are not available.

What does AI do?

AI proposes the requests, data connections and business checks from the context you supply. You review the proposal and execution scope. A deterministic runner executes the approved scenario and records the measurements. Generating a scenario never starts traffic against your application.

Can I test production or increase the load?

Yes, with verified authority, approved access, a test window and explicit request, rate and duration limits. We start by validating the journey. Sustained load, concurrency ramps and capacity testing require a separately agreed execution profile. The examples on this page show functional validation and a short scheduled fixture run. They do not establish production capacity.

How are inputs and access handled?

We agree which context to use, the model provider and data handling requirements before you share confidential inputs. API paths and field names can also be sensitive. Test access is arranged separately. Do not email credentials, session cookies or customer records. Arbitrary repository onboarding is assisted, rather than a self-service connection.

How do repeat runs and pricing work?

The same approved scenario can run again without another AI generation. A changed scenario may need a new proposal and review. Preparation and execution are scoped separately, and execution still consumes resources. We agree the scope and price before paid work; contacting us does not trigger a charge.

Can I export the scripts or reports?

Executable scripts stay managed inside Loaduna. Your workspace shows readable scenarios, saved versions and recorded results. Downloadable report exports are not currently available. Invited users can open the workspace through Workspace access.

LET’S START WITH YOUR API

What needs to work
when it matters?

Describe one critical workflow and your target environment. We’ll review the fit and the inputs needed, then agree the first run’s scope and price.

  1. 01Share the goal
  2. 02Agree the first run
  3. 03Review it in your workspace

Let’s plan your first test.

Email us about your test

Keep credentials and customer records out of your message.