Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Wren by Seldon

Describe what your web app should do in a sentence; agents explore it in a browser and bring back recorded evidence of what broke.

Reviewed by ReviewRamble editors · Updated

Visit Wren by Seldon Pay per use

Share this page:XLinkedInFacebookRedditWhatsAppEmail

Our rating

3.6/5

Wren makes it cheap to find out whether exploring agents catch what your scripted tests miss, but with no public docs or stated integrations it is something to trial, not yet something to build a release process on.

What is Wren by Seldon?

Wren is the software testing product from Seldon (Blok Intelligence Inc.), an applied research lab that builds models of how people behave. You describe a test in plain English: where to start and what a successful result looks like. Agents then work through your web app in a browser, choosing each next step from what is on screen, and run several tests in parallel. Every run is recorded step by step, so a failure comes with the path that led to it and the steps to reproduce it.

Pricing is per completed test run: $3 each after the first 10, which are free and shared across a workspace. A run is one test case, executed once, in one browser and environment. Model usage, browser time and the recorded evidence are included, there are no seat fees, and runs that fail because of an agent or platform problem are not billed. Volume pricing and enterprise agreements are available through sales.

Is Wren by Seldon worth it?

Scripted browser tests break for dull reasons. A designer renames a button, twenty tests fail at once, and someone spends the afternoon repairing selectors instead of finding out whether the release works.

Wren asks for a sentence or two instead of a script: start on the product page, add a jacket to the bag, check the bag shows one item. Seldon's agents then use your web app in a browser, deciding each step from what they see. If a button gives no feedback, an agent may click again or look elsewhere, the way an impatient customer would. The example on Seldon's own page is the useful kind: a shopper clicked "Add to bag" twice because nothing confirmed the first click, and ended up with two items. That is a bug no one thought to write a test for.

The idea behind it comes from Seldon's main line of work, simulating how different people make decisions. Seldon reports that its simulated users made the same choices as real people 87% of the time across a full journey. That figure is Seldon's own, and it measures its user simulations in general, not how many bugs Wren finds.

The pricing is easy to reason about. Each completed run costs $3, the first 10 are free, and there are no seat fees or token charges. A run is one test case in one browser, so a 10-case suite costs $30 every time it runs. Runs that fail because Wren broke are not billed; a run that finds your bug is.

The catch: there is very little to inspect before you commit. The documentation sits behind a login. The product page names no CI system, GitHub app or Slack integration, and every example is a web app in a browser, with nothing said about native mobile apps. The site footer lists SOC 2, GDPR, HIPAA and ISO 27001 without saying which are completed audits. Seldon says it does not train general models on customer data unless you authorise it, but third-party AI providers process that data, and to test anything behind a login the agents need your test account's credentials.

Our take: $3 a run is cheap for a smoke test and expensive for a suite. Ten free runs on your checkout or sign-up flow will tell you whether the agents find paths your scripts never tried; running it on every merge is a different bill, and a bigger act of trust in a product this new.

Where it earns its place

  • A lead engineer at a 10-person e-commerce startup

    Find out what a checkout redesign broke before it ships

    Describe three checkout journeys in plain English and run them on staging. Ten runs are free; after that each completed run is $3, so a pre-release pass of 5 journeys costs $15, with recordings to hand straight to whoever fixes the bug.

  • A QA manager with a brittle Playwright suite

    Check whether agents catch bugs the scripts miss before replacing anything

    Pick one flow the suite already covers and run it in Wren alongside it, as Seldon suggests. Compare the recordings against the scripted result; if Wren finds extra paths, move the flakiest scripted cases over first.

  • A product manager at a B2B SaaS company

    See where a confused first-time user gets stuck in onboarding

    Write the onboarding goal as a test and review the recorded runs for retries and dead ends, like the double click Seldon shows in its own example. Each run costs $3, far less than recruiting a user for a session.

Before you commit

Setup
Sign up at app.seldon.com, create a workspace, describe a test in plain English with a starting point and a success condition, then run it; Seldon suggests starting with a flow your suite already tests.
Integrations
Not published. The product page names no CI system, GitHub app, Slack or API, and public documentation is not available without an account.
Data and privacy
Seldon says customer data is not used to train general models unless you authorise it; third-party AI providers process it on Seldon's behalf, with subprocessors listed at trust.seldon.com. Retention follows the customer agreement; no fixed period is published. The site lists SOC 2, GDPR, HIPAA and ISO 27001. Which AI tools train on your data?
Cost at scale
A run is one test case in one browser, so a 10-case suite run nightly is 300 runs, about $900 a month, and running it on every merge multiplies that. Volume pricing is by negotiation; no rates are published.
Cancelling and refunds
There is no subscription: after the first 10 free runs you pay $3 per completed run, runs that fail on Wren's side are not charged, and fees are otherwise non-refundable. You can stop using the service at any time, though charges already incurred stay due, and your data is returned or deleted under your agreement.
Limits
Web apps in a browser only, as far as the product page shows; concurrency limits, run time limits and mobile support are not published. Outputs are probabilistic, and Seldon's own policies say results must be reviewed before you rely on them.

What’s good

  • Tests written as a sentence or two of plain English
  • Agents adapt to what is on screen and find paths you did not script
  • Every run recorded step by step with reproduction steps
  • Flat $3 per completed run; no seat fees or token charges
  • Runs that fail on Wren's side are not billed

What’s not

  • No public documentation; docs sit behind a login
  • No CI, GitHub or Slack integration named on the product page
  • Browser-based web apps only; nothing published on native mobile
  • A 10-case suite costs $30 per run, which adds up on every merge
  • Compliance badges listed without saying which audits are complete

Best for

Engineering and QA teams who want to see whether exploring agents find bugs their scripted web tests miss, before rewriting any of them.

Wren by Seldon alternatives

Wren by Seldon: common questions

Is Wren by Seldon worth it?
Wren makes it cheap to find out whether exploring agents catch what your scripted tests miss, but with no public docs or stated integrations it is something to trial, not yet something to build a release process on. We rate it 3.6 out of 5.
How much does Wren by Seldon cost?
First 10 completed test runs free per workspace; then $3 per completed run (USD), including model usage, browser execution and recorded evidence. No seat fees; runs that fail on Wren's side are not charged. Volume and enterprise pricing on request.
What are the downsides of Wren by Seldon?
No public documentation; docs sit behind a login; no CI, GitHub or Slack integration named on the product page; browser-based web apps only; nothing published on native mobile; a 10-case suite costs $30 per run, which adds up on every merge; compliance badges listed without saying which audits are complete.
Who is Wren by Seldon best for?
Engineering and QA teams who want to see whether exploring agents find bugs their scripted web tests miss, before rewriting any of them.
How much does Wren by Seldon cost for a team or in production?
A run is one test case in one browser, so a 10-case suite run nightly is 300 runs, about $900 a month, and running it on every merge multiplies that. Volume pricing is by negotiation; no rates are published.
Is Wren by Seldon safe for company data?
Seldon says customer data is not used to train general models unless you authorise it; third-party AI providers process it on Seldon's behalf, with subprocessors listed at trust.seldon.com. Retention follows the customer agreement; no fixed period is published. The site lists SOC 2, GDPR, HIPAA and ISO 27001.
How do I cancel Wren by Seldon, and can I get a refund?
There is no subscription: after the first 10 free runs you pay $3 per completed run, runs that fail on Wren's side are not charged, and fees are otherwise non-refundable. You can stop using the service at any time, though charges already incurred stay due, and your data is returned or deleted under your agreement.
What are the best Wren by Seldon alternatives?
The closest alternatives we have reviewed are Momentic, QA Wolf, TestSprite and Meticulous.

Answers checked Nov 18, 2026.