AI Power Apps

For every AI assistant and LLM app

One-click upgrades for your AI assistant.

AI Power Apps gives Claude, ChatGPT, Cursor, or your own app live online abilities and a second-opinion reviewer, through MCP or a direct API with one key. Sign up, get free usage tokens, and connect in under a minute.

  1. 01Sign up and get free usage tokens
  2. 02Enable the upgrades you want
  3. 03Pick a Sharpen tier
  4. 04Paste one key into your assistant

Two pillars

Everything an assistant is missing, in one connection

Online abilities give the model the live world. Sharpen gives it a second opinion before it answers.

Online abilities

Research, scriptures, classics, books, music, YouTube, weather, navigation, hotels, flights, GitHub, social, web search and page reading, all through one interface: discover, inspect, execute, follow up. Every search returns a search ID and record IDs the model can page through or drill into, so it never repeats a query to get the next page or the details.

Sharpen

A second-opinion model from a different family first approaches the task blind, then reviews the assistant’s draft and returns a standalone critique. Four tiers, from a fast check to two frontier reviewers in parallel. The assistant decides what to keep.

Online abilities

Fifteen capabilities, one interface

Each one is a search the model can inspect before calling, then page or follow up on by ID. Free vendors cost nothing; paid ones cost a few usage tokens.

  • Web search

    Web

    Google-quality results with country, language and time filters; read any hit in full.

    web.search

  • Web read

    Web

    Fetch a page as clean markdown or text, trimmed to the length you ask for.

    web.read

  • Academic search

    Academic

    Papers from OpenAlex with abstracts, citation counts and open-access PDFs.

    academic.search

  • Scripture

    Scripture

    Bible and Quran passages by reference or phrase, with the surrounding chapter.

    scripture.search

  • Classics

    Classics

    Public-domain literature from Gutenberg and Wikisource, with excerpts and downloads.

    classics.search

  • Books

    Books

    Open Library titles, editions and descriptions by query, author or subject.

    books.search

  • Music

    Music

    Tracks, artists and albums from MusicBrainz and Deezer.

    music.search

  • YouTube

    YouTube

    Video search with duration and date filters; pull the transcript of any result.

    youtube.search

  • Weather

    Weather

    Up to 16-day forecasts for any place name or coordinate, with air quality and marine data.

    weather.forecast

  • Routes

    Navigation

    Driving, walking and cycling routes with waypoints, alternatives and turn-by-turn steps.

    navigation.route

  • Distance matrix

    Navigation

    Travel times and distances between many locations in one call.

    navigation.matrix

  • Hotels

    Hotels

    Google Hotels availability with price, rating, amenity and property filters plus booking links.

    hotels.search

  • Flights

    Flights

    Google Flights itineraries by date, cabin and stops, with booking links.

    flights.search

  • GitHub

    GitHub

    Repositories, code and issues, with README heads, file trees and stats on follow-up.

    github.search

  • Social

    Social

    Posts and profiles on X and Reddit.

    social.search

Sharpen

A second opinion before the answer ships

Pick the tier once in your dashboard; the assistant calls one endpoint. Costs are typical usage tokens per review.

  • Sharpen

    A fast reviewer model approaches the task blind, then critiques the draft.

    ~10 tokens per review

  • Sharpen Plus

    A stronger frontier reviewer for work where the details matter.

    ~40 tokens per review

  • Ultra

    The strongest reviewer available, for high-stakes output.

    ~120 tokens per review

  • Ultra 2x

    Two Ultra reviewers from different model families, run in parallel.

    ~240 tokens per review

Sharpen sends the task and the draft to the reviewer model providers through OpenRouter. Transcripts are kept for 30 days for support and quality checks.

How it connects

MCP for assistants, REST for everything else

The same key works for both. Hosts that cannot set headers use the key-in-URL form.

MCP

Claude Code, Claude Desktop, claude.ai, ChatGPT, Cursor and VS Code all speak MCP over HTTP. One command or one JSON block and the abilities appear as tools.

Claude Code
claude mcp add --transport http ai-power-apps https://api.powerups-ai.store/mcp --header "Authorization: Bearer <YOUR_API_KEY>"
claude.ai / ChatGPT connector URL
https://api.powerups-ai.store/mcp/k/<YOUR_API_KEY>

Direct API

Plain JSON over HTTPS. Start with the catalog, inspect a capability to get its schema, execute, then follow up by search or record ID.

curl
curl -H "Authorization: Bearer <YOUR_API_KEY>" https://api.powerups-ai.store/v1/catalog

The AI-readable guide lives at https://api.powerups-ai.store/v1/guide.

Pricing

Usage tokens, monthly or yearly

Every paid plan is a monthly allowance of usage tokens, billed monthly or yearly with two months free. Free vendors cost nothing; paid vendors and Sharpen reviews draw from the allowance.

  • Free

    Free

    1,000 usage tokens, once

    Every account starts here. Enough to wire up your assistant and try every ability.

  • Starter

    $9/month

    5,000 usage tokens per month

    or $90/year (2 months free)

    For daily use by one person with a few assistants.

  • Pro

    $29/month

    15,000 usage tokens per month

    or $290/year (2 months free)

    For heavy research, Ultra reviews and team-shared keys.

  • Max

    $99/month

    60,000 usage tokens per month

    or $990/year (2 months free)

    For teams and always-on agents that search and review all day.

Give your assistant the upgrade

Free usage tokens on every account. No card needed to connect.