/imagine prompt: A stunning AI cyborg girl stands tall amidst a bustling cityscape, her metallic skin reflecting the warm glow of the setting sun… Camera: Canon EOS R5. Lens: Canon RF 50mm f/1.2L USM. Settings: Shutter speed 1/250, Aperture f/4, ISO 400. —ar 16:9
HARPA AI is a Google Chrome extension, powered by hybrid AI-engine that combines GPT (ChatGPT, Bard, ClaudeAI) with Web Automation. Read the Get Started Guide to find out more.
HARPA automates your copywriting, web surfing, online data extraction and web monitoring tasks.
It is used to quickly write and reply to emails, post comments and DMs, summarise and “talk to” web pages, YouTube videos, PDF documents, extract and research SEO keywords, create SMM content plans and Midjourney prompts, automate form filling, write blog articles, review CVs, job offers and legal documents, plan trips, track competitors, prices and SERP pages, create learning plans, search for information online and more.
HARPA can read, understand and act upon web page content.
It can navigate, click and extract data from web pages, run it through Large Language Models, paste back to websites or trigger IFTTT chains such as Zapier and Integromat.
HARPA is the ultimate copilot and AI Agent for your web browsing and automation tasks. It provides access to AI on any website, and you can use HARPA with almost every web service or online app.
HARPA is a freemium app. It has premium paid features but the app’s core functionality is free to use. You do not need to register anywhere to start using HARPA. You can try the app before making an informed purchase.
HARPA is optimized for speed and performance and does not use CPU when it has no tasks to run. You can also limit the number of automations that are run simultaneously in the HARPA app settings.
HARPA stands for High Availability Robotic Process Automation. HARPA is designed to bridge the gap between AI and Automation and serve as your virtual assistant and a digital workforce for your business.
Generative AI features require connection to AI service providers. You can use HARPA with ChatGPT account, OpenAI API key, ClaudeAI or Google Bard. You do not need an LLM service-provider account if you only use HARPA for its’ automation and monitoring features.
Step 1: Navigate to the ChatGPT login page. Click Sign Up button. ChatGPT offers three options for new user registration:
Choose the appropriate option and create the account. ChatGPT will send a verification email to the registered email address.
Step 2: Open your email account and verify your email address by clicking the Verify Email Address button and enter your full name.
Step 3: Enter your phone number. ChatGPT will send a verification code to your phone number. Verify it by typing the code in the space provided. Finally, your ChatGPT account is ready to use!
ChatGPT Plus comes with GPT-4. It is limited by OpenAI to 50 requests per 3 hours. GPT-4 is also available for OpenAI API accounts.
To upgrade to ChatGPT Plus:
To create OpenAI account and generate a key, navigate to openai.com
Locate GPT connection button below the Chat window. You might need to expand the control panel by clicking the "triangle" button. Select a GPT connection and model in the screen that appears.
Alternatively, type -3 in chat and press Enter to switch model to GPT-3, -4 to switch to GPT-4, -c to switch to ClaudeAI, etc.
HARPA is privacy-centered. It runs on your computer, does not send or store website data on remote servers, only performs the tasks you have asked it to. We do not show ads or sell your data.
GPT prompts you enter to HARPA are transparently sent to OpenAI, Google or ClaudeAI servers for processing and not stored on our servers.
Privacy PolicyHARPA does not work with Google Sheets yet. Some domains e.g. Discord may require you to disable fingerprint protection on Brave and Edge. Click HARPA icon in the browser toolbar, instead of pressing Alt+A. If the HARPA popup does not appear:
If nothing helps, navigate to chrome://extensions, disable HARPA, quit and restart your browser, and turn HARPA back in.
This error may occur if:
You can see 🧬 chats in chat.openai.com if you're on a Browser Session connection. If you don’t want your dialogues to be cleared automatically whenever a tab is closed, you can change this in the App Settings.
If a YouTube video is fresh or has a low number of views, it may not have a transcript available. If you are using uBlock Origin, turn it off while running HARPA or add a rule to allow XHR requests on YouTube.
Yes, HARPA has an extensive library of copywriting prompts, can generate and refine short and long articles in 18 writing styles and can mimic yours.
Please click the settings (”gear”) button in the app and uncheck the box next to “Show floating quick-access bar”.
There are Pros and Cons to using OpenAI API keys over Browser Session:
HARPA provides AI search assistance for Google, Google Scholar, Baidu, Bing, DuckDuckGo, Brave, Yahoo, Naver, Yandex, Kagi, Searx.
This is a planned feature. You can follow product updates in our Discord and also vote for the features you’d like to see implemented first.
A prompt is a message (instruction) your send to Large Language Model (LLM) such as OpenAI GPT-3.5. AI Command is an automation recipe which may or may not include steps that prompt LLMs.
Think of AI Commands as a sequence of steps, e.g. “Navigate to web page”, “Click Messages button”, “Extract messages”, “Run through GPT to find important messages”, “Send as Email”.
HARPA comes with over 100+ preconfigured commands and you can create yourself to automate your workflow.
HARPA makes it possible to customise your prompts with predefined parameters, such as:
You can also create custom parameters in your commands to store data or options. Custom parameters are usually named {{p1}}, {{p2}}… You can also set up global parameters such as {{global.signature}} and reuse them in your prompts e.g. to write customised emails.
Large Language Models have a limited token window, e.g. 4096 tokens on ChatGPT 3.5. It is often impractical or impossible to fit large documents, web pages, and video transcripts into a single GPT request.
Chunking is a technique that divides content into smaller parts, also called chunks, to ensure that each part, including the prompt, fits within the GPT token limit. These chunks are then injected into prompts as {{chunk}} parameters and processed by GPT one by one.
Chunking does not possess inherent memory. Therefore, if you require context to be maintained across chunks, it must be incorporated within the prompt.
Most HARPA AI commands send a {{language}} parameter with prompt. You can set the language for commands output in the app settings. You can also send {{language}} parameter in chat along with your prompts to instruct AI to reply in a given language.
HARPA UI is currently available in English only. We will internationalize and translate the user interface once all the major features are implemented.
Support for Cloud AI API is currently under development.
This is a planned feature. HARPA does not have a Cloud API to schedule, launch or otherwise orchestrate AI commands yet. You can follow product updates in our Discord and also vote for the features you’d like to see implemented first.
Not at the moment. The Firefox version as well as UI translation are planned after all the major features are implemented.
All rights reserved © HARPA AI TECHNOLOGIES LLC, 2021 — 2023