mobile conversion intelligence in your ide.

Connect tasu to Claude, Cursor, Codex, or any MCP-compatible agent. Choose your paywall, price your plans, sequence onboarding, and cut churn, without leaving your editor.

Works with Claude Code · Cursor · claude.ai · Claude Desktop · VS Code · Windsurf

Setup

Pick your tool. Paste once. Done.

One URL, every MCP client. No SDK, no dashboard, no data connection.

  1. Get your key at checkout. It also lives in your account.
  2. Run this in your terminal.
  3. Ask while you build: "hard paywall or free trial?"
claude mcp add --transport http tasu https://tasu.ai/api/mcp --header "Authorization: Bearer tasu_your_key"

Lost your key? Log in to your account and copy it again. Your usage lives there too.

Tools

Two tools. Zero noise.

Intent-level tools your agent discovers at runtime. Both declare their output shape, so agents know exactly what comes back.

ask

Any onboarding, paywall, pricing, or retention question. Returns sourced claims with their dataset and year, plus the deciding condition when experts disagree.

counts toward your monthly quota
list_topics

The knowledge domains the brain covers, with the claim count per domain. Useful for discovery.

free, never metered

1,000 ask calls a month included. Track live usage in your account.

How it works

Expert answers, inside the build loop

Your coding agent writes the screen. tasu makes sure it converts.

STEP 1

Ask while you build

Your agent asks tasu about the screen it is writing: paywall type, trial length, notification timing, pricing structure.

YOU, IN CURSOR

"Hard paywall or free trial for my fitness app?"

STEP 2

Get sourced claims, not vibes

Every answer carries real benchmarks with their dataset and date. When experts disagree, tasu shows both sides and what decides.

TASU, OVER MCP

"Hard paywalls convert ~5x better. 90% of conversions land within 24h of install (Adapty 2026, 20,000+ apps). Trials win once your aha moment fits the window."

STEP 3

Ship the fix in the same session

Answers end as paste-ready fixes. The agent applies them to your code before the leak ever ships.

#1Move the review ask after first valuefixed
#2Embed the notification ask in a questionfixed
#3Show yearly per-week pricingfixed
Just ask your agent

Every money screen, answered

The moments your agent stops guessing and pulls a real benchmark instead. A few of the hundreds it can ask.

Choosing your paywall

Hard paywall or a free trial for my app?

Hard paywalls convert about 5x better: 10.7% vs 2.1% download-to-paid at D35 (RevenueCat 2026, 115,000 apps).

Setting the trial length

How long should my free trial be?

14 days on the annual plan won Headspace's A/B test. 17 to 32 day trials convert 42.5% vs 25.5% for 4 days or fewer (RevenueCat 2026).

Pricing your plans

Weekly, monthly, or annual, and at what price?

Structure beats price. Trial-structure changes drove a 59.6% LTV uplift. Gate the trial to annual and the first charge is $60, not $9.

Building onboarding

How many screens, and in what order?

Run the six jobs of onboarding, then a 3-screen trust sequence before the paywall. It is how Cali ($2-3M/mo) kills bill-shock fear.

Writing the paywall copy

Write my paywall CTA and subtitle.

Add a 'cancel anytime' subtitle and a step-by-step trial timeline. Blinkist did this and got more sign-ups with fewer complaints.

Adding friction that pays

Should I require a card to start the trial?

Outsider did. Sign-ups fell 50%+, but conversion rose 5x, netting 2x the paying customers. Friction that filters intent wins.

Pricing

One MCP. 9€ a month.

The price of a coffee for the conversion expert that lives in your coding agent. Cancel anytime.

The MCP
9€/mo

The conversion expert in Claude Code, Cursor, claude.ai, and any MCP client.

  • Skip the guesswork on onboarding, paywalls, and pricing
  • Real benchmarks with their source and date, ready to cite
  • Paste-ready fixes your agent ships in the same session
  • 1,000 ask calls a month, with live usage in your account
FAQ

Questions, answered

Everything you need to know before you add the URL.

Model Context Protocol, the open standard that lets AI agents connect to external tools. Claude, Cursor, and most coding agents speak it. tasu is a remote MCP server: your agent connects to one URL and can ask it questions while it works.

Ship screens that convert

One URL in your agent config. The expert answers while you build.

Get the MCP →
Mobile App Intelligence

We reverse-engineer the top apps

Two series, free. App teardowns of apps making $100k/mo to $1B ARR, and deep dives on the subjects that decide conversion. Every benchmark the MCP serves comes from here.