askbowtie logo
Observability Monitoring · Marketing Analytics Attribution

askbowtie

askbowtie joins a website's analytics, JavaScript errors, uptime, Google Search and Google Ads into one MCP server, so the owner's own AI agent answers questions with cited numbers. One cookieless script tag; the dashboard stays optional.

Beta Free plan Freemium API available 16+ Verified by Guidaio
Overview

What is askbowtie?

askbowtie is website monitoring rebuilt for the age of AI agents. It joins the five signals a site normally scatters across separate tools, namely analytics, JavaScript errors, uptime, Google Search Console and Google Ads, into one per-site database, then exposes that database over MCP so the owner's own coding agent can query it directly and answer with cited numbers. The brand mark is the relational-algebra join operator, which states the principle plainly: combine tables on what they share. Setup is a single asynchronous script tag of 9.8 KB placed in the page head. From there the tracker captures pageviews, sessions, clicks, rage clicks, form submissions, console, resource and network errors, CSP violations, Core Web Vitals and scroll depth with no configuration at all. Four JavaScript calls cover everything custom, track, converted, error and guardrail, while two declarative HTML markers classify a page's role and name the regions of a page without any API call. A server-side ingest endpoint accepts webhook-confirmed conversions straight from a backend. The MCP server exposes 39 tools in eight families: getting started, traffic and pages, conversions and performance, errors and uptime, deterministic crawling, search and ads, funnels, and actions and operations. Beyond reading, the ads tools can pause campaigns, set budgets and bids and add negatives, gated to owners and admins with every change previewed before it applies. Twenty-one published playbooks turn common investigations into procedures an agent follows step by step. The stated philosophy is restraint: store the raw signal faithfully, serve it back close to raw, and let the agent do the reasoning rather than pre-digest it into a vendor's verdict. Only two events are allowed to interrupt a human, the site going down and traffic dropping to zero. The tracker is cookieless by default and each site gets an isolated database, built on Cloudflare Workers and D1. The publisher is Benjo LLC, a Kentucky limited liability company whose founder is identified only as Ben. The product launched in June 2026, having previously been known as itbroke.dev.

What it does

  • Answer plain-language questions about a live site directly inside an AI agent, over MCP
  • Join analytics, JavaScript errors, uptime, Google Search and Google Ads into a single queryable source
  • Catch broken forms, failed requests, CSP violations and rage clicks as they happen
  • Track conversions with revenue and match Google Ads spend to the signups it actually bought
  • Run a deterministic full-site crawl that scores health and lists broken links and images
  • Build ordered funnels and measure fall-off and median time per step
  • Schedule recurring reports, alerts and funnel watches without opening a dashboard
Audience

When to use askbowtie / When not to

A quick filter to help you decide if askbowtie is the right fit.

When to use askbowtie

  • Developers and site owners who already work inside an AI coding agent such as Claude Code, Cursor or Codex
  • Solo founders, freelancers and indie makers who want one script tag instead of four monitoring subscriptions
  • Growth and performance marketers who need Google Ads spend matched to the signups it actually bought
  • Small technical teams tired of switching between analytics, error tracking, uptime, Search Console and Ads tabs
  • Privacy-conscious sites that want cookieless, first-party measurement with no consent banner for the tracker itself

When not to use askbowtie

  • Teams that want a rich, ready-made dashboard, since the product is agent-first and the screen is deliberately secondary
  • Anyone not already running an AI agent, as the hosted AI option is still closed and waitlisted
  • Organisations that require a signed DPA, published security certifications or a named hosting country
  • High-traffic sites relying on the free tier alone, capped at 100,000 events a month across every site on the account
  • Users who need native mobile apps or a contractual uptime guarantee, neither of which exists
Get started

How to use askbowtie

A typical end-to-end flow, from setup to results.

  1. Run a free crawl on any public URL first, with no account and no installation required
  2. Sign in with the one-tap magic link sent by email, with no password and no credit card
  3. Choose a path: query through your own AI agent over MCP, use the hosted dashboard, or join the waitlist for the hosted AI
  4. Add the 9.8 KB async script tag to the page head on every page, using the copy-paste guides for WordPress, Shopify, Google Tag Manager or Cloudflare
  5. Connect Google Search Console and Google Ads under Settings then Connectors, one OAuth consent per product
  6. Register the MCP server with your agent over the http transport, passing a Bearer token
  7. Ask the agent to call the playbook tool first: it returns the guided map of what to ask and which tool answers it
  8. Optionally mark up page roles and page sections with two HTML attributes, so analytics group by meaning rather than by slug
  9. Add custom events, conversions with revenue, application errors and guardrails through the four JavaScript calls
  10. For backend conversions, capture the browser session id, store it on the order, then replay it in the server-side POST
Quick read

Pros & Cons

Pros

  • A single 9.8 KB asynchronous line to install, with no claimed render cost
  • A genuine join across analytics, errors, uptime, SEO and paid, so one question crosses all five
  • Unlimited, unmetered MCP queries included in the free tier
  • Generous free tier of 100,000 events a month, with no subscription and no credit card
  • Cookieless by default, form values never read and click text scrubbed before it leaves the page
  • An isolated database per site, with data never pooled, sold or used for advertising
  • Dense documentation, 21 executable playbooks and a machine reference written for agents

Cons

  • Very young product, launched in June 2026, with terms that label the service beta and offer it as is
  • Tiny publisher: a Kentucky LLC with a founder known only by first name and no published postal address
  • No social presence at all, which makes team size and traction hard to gauge
  • GDPR is never mentioned, and there is no DPA, no Article 27 representative and no DPO
  • No published security certification, and no hosting country named beyond Cloudflare infrastructure
  • Pricing is not settled: credits are announced while the product is currently free in full
  • No mobile apps, no documented interface languages and no availability guarantee
Pricing

Pricing & Plans

A permanent free plan is available at no cost and covers normal use: the tracker, the dashboard and the full MCP server, up to 100,000 events per month across every site on the account, with unmetered MCP queries. The lowest paid entry point is USD 10.00 for 1,000 credits, which fund AI answers in the in-app Ask tab and events beyond the free allowance. Credits do not expire, no subscription is offered, and no credit card is required to begin. At the time of this review the publisher states that the full product is entirely free.

Free — USD 0
  • tracker
  • dashboard and full MCP
  • up to 100
  • 000 events per month across all sites
  • unmetered MCP queries
Credits — USD 10 for 1,000 credits
  • AI answers in the Ask tab and events beyond the free allowance
  • a few credits per question
  • no expiry
  • with a free stack for early users
Special offers — Early bird phase: the full product is currently free, with every feature included · Early users keep a founding-user rate for good, even once credits arrive for heavier use · Early users start with a free stack of credits
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how askbowtie handles your data.

GDPR overview

There is no mention of the GDPR anywhere on the site: the acronym appears on none of the pages collected, and compliance is never claimed. What the privacy policy does describe is a GDPR-shaped architecture. Roles are set out explicitly, the site owner being the controller for visitor data and askbowtie the processor, and the lawful basis rests on the customer. Rights to access, correct, export, delete, object and restrict are offered, exercised by email to support@askbowtie.com. No EU representative under Article 27 is designated, no Data Processing Agreement is published, and no Data Protection Officer is named. The tracker sets no cookies, and the policy notes that a site running only bowtie.js generally needs no consent banner for it. The publisher is established in Kentucky, United States, and no transfer mechanism is mentioned. The policy is dated 16 June 2026.

Who owns the data?

Benjo LLC states that customers keep ownership of their data. The terms split the roles: for anything the tracker collects on a customer's own site, the customer is the data controller and askbowtie acts solely as their processor, working on instruction; for the customer's own account data, askbowtie is the controller. Each site is given its own isolated database, and the vendor commits to never selling that data, never using it for advertising and never sharing it beyond the processors named in the privacy policy. A bring-your-own-database option, writing to a database URL the customer controls, is announced as a roadmap item.

Reuse rights

Customers may reuse their own data freely: they own it, and the permission granted to the vendor is limited to what is needed to run the service for them. Export or deletion can be requested at any time by email, and on termination the data is made available for export for a reasonable period before being deleted. In return, the customer carries the obligations: instrumenting only sites they own or are authorised to instrument, holding a lawful basis to run analytics, publishing their own visitor-facing privacy notice, and keeping sensitive personal data out of the custom event payloads they control. Reselling or white-labelling the service, or building a competing tracker from its internals, requires written agreement.

Data retention & training

Retention summary
No retention period is published in figures; the policy stays qualitative. Account data is kept for as long as the account is active. Analytics and error data is retained to power the dashboard and historical comparisons, with no stated limit. Operational logs, such as uptime checks and resolved alerts, are pruned automatically on a rolling basis. Customers can ask for an export or a deletion at any time by email, and the publisher undertakes to action it. On termination, data is made available for export for a reasonable period and then deleted. At the collection layer, the session id is a random value held in the browser's localStorage for 30 minutes of activity, and the visitor's IP address is used only transiently for rate-limiting and country lookup, never written into the event record.
Trains on customer data
No
Subprocessors disclosed
Yes
GDPR contact

Hosting summary

Data is stored on Cloudflare's infrastructure, and the technical foundation is named as Cloudflare Workers and D1. No country and no region is ever specified, so the hosting jurisdiction cannot be established from the published material. Each site receives its own isolated database, and data is never pooled between customers. Two processors are named explicitly and presented as the only ones: Cloudflare for hosting and edge, and Resend for sending sign-in links, alerts and scheduled reports. Data is exchanged with Google only when a customer connects Google Search Console or Google Ads, and then solely to pull that customer's own metrics. Connector grants are stored encrypted at rest, and MCP API tokens are stored only as a hash, the raw token being shown once and never again. No advertising or data-broker services are used. A bring-your-own-database option, writing to a database URL the customer controls, is announced as a roadmap item rather than a current capability. The publisher is established in Kentucky, United States.

Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting askbowtie.

  • The service is explicitly labelled beta and offered as is, with no availability guarantee: features can change, ship or be removed
  • Only the ingest endpoint and the MCP contract are promised stable, so instrumentation built on anything else may need rework
  • Connecting Google Ads necessarily grants read and write access, because Google offers no read-only scope, and a misdirected agent can alter live campaigns
  • Calling the identify function moves tracking out of the cookieless default and shifts the consent burden onto the site owner
  • Letting an agent narrate the numbers can erode the habit of reading the data yourself, and a confident answer built on a mis-wired session id can be quietly wrong
  • Governance is thin for a data processor: no GDPR mention, no DPA, no Article 27 representative, no certification and no named hosting country
  • The publisher is a very small company with no postal address and no public presence, which concentrates continuity risk on one individual
Setup

Setup & Integrations

Technical difficulty

Low to moderate, depending on how far you go. A crawl needs nothing at all: a public URL, no account, no code. Sign-in is a passwordless magic link. The tracker is one script tag in the page head, with copy-paste guides for WordPress, Shopify, Google Tag Manager and Cloudflare, and the onboarding explicitly caters for those who must forward the snippet to a developer. Google connectors are one OAuth click each. Registering the MCP server takes a command line and a Bearer token. Server-side ingest is the genuinely demanding part, since session ids must be captured, stored and replayed.

Deployment

Web appAPI

Integrations

Google Search Console Google Ads Claude Code Cursor Codex Cline Windsurf Zed Copilot WordPress Shopify Google Tag Manager Cloudflare
Company

Behind askbowtie

Company name
Benjo LLC
Founded
06/08/2018
Country of origin
🇺🇸 United States
UBO
Ben
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇺🇸 United States
Legal contact
Support contact
Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement askbowtie.

G GA4P PlausibleP PostHogS Sentry
FAQ

Frequently asked questions

Does my site's data go to an AI model provider?
No. Data lives in a database dedicated to your site, your agent queries it over MCP, and only the answer it asked for passes through. Raw analytics are never shipped to a model provider.
Will the tracker slow my site down?
The script is 9.8 KB and fully asynchronous. It loads after your content and never blocks rendering, capturing sessions, conversions, errors and broken forms in the background.
Do I have to use MCP?
No. A hosted dashboard and an email digest expose the same joined data. A hosted AI is also announced for people who run no agent of their own, but it is not open yet and requires joining a waitlist.
What does it cost?
Normal use is free: tracker, dashboard and full MCP up to 100,000 events a month across all your sites, with MCP queries unmetered. Beyond that, and for AI answers in the Ask tab, credits cost USD 10 for 1,000 and never expire.
Do I need a credit card to start?
No. Sign in with the magic link, install the tracker, and the free tier works. Credits are added only when you want AI answers or your traffic outgrows the free allowance.
Who owns the data?
You do. Each site gets its own isolated database, data is not pooled with anyone else's, and the publisher commits to never selling it or using it for advertising.
Does it set cookies?
The tracker sets none and uses a random session id held in localStorage for 30 minutes of activity. askbowtie.com itself uses one essential first-party cookie to keep you signed in to the dashboard.
Is there a minimum age?
Yes, 16. The terms require account holders to be at least 16, and the privacy policy states that personal data is not knowingly collected from anyone younger.
Who is behind askbowtie?
Benjo LLC, a limited liability company formed in the Commonwealth of Kentucky, United States. The tool was previously known as itbroke.dev and was rebranded in 2026.
Can I see it before signing up?
Yes. A public demo runs on real crawl output from a showcase site and needs no signup at all.
Conclusion

Should you pick askbowtie?

askbowtie makes an unusually clear bet: the dashboard is not the product, the agent is. By joining analytics, JavaScript errors, uptime, Google Search and Google Ads into a single per-site database and exposing it over MCP, it lets one question cross five silos that normally demand five tabs and five logins. For anyone who already lives inside a coding agent, that advantage is immediately legible, and the free tier backs it convincingly: the full tracker, the dashboard and unmetered MCP queries up to 100,000 events a month, with no subscription and no card. The privacy posture is coherent rather than decorative: no cookies by default, form values never read, click text scrubbed before it leaves the page, one isolated database per site, and an explicit refusal to sell data or run advertising. The reservations are all about maturity and governance. This is a June 2026 product from a Kentucky LLC whose founder is identified only as Ben, with no postal address, no social presence and 43 early users on the counter. The word GDPR appears nowhere; there is no DPA, no Article 27 representative, no security certification and no hosting country named beyond Cloudflare infrastructure. The terms label the service beta and disclaim any availability guarantee, and the eventual pricing is announced but unsettled. The result is a tool worth adopting quickly on a personal site, an indie project or a small technical team's stack, where the cost of being early is low and the join genuinely saves time. For a regulated environment or a high-stakes site, test it first and revisit once the governance paperwork catches up with the engineering.