Kiru

Hey Product Hunt! 👋
I built Kiru because businesses need a flexible, low cost tool that injects intelligence into what they are already working on.

Kiru is a Windows desktop application that can transcribe any phone call (including over iPhone or Android) and take a screenshot when anything in the conversation is in question. The screenshot feature can be used on its own to process text on the screen or photographs.

The program doesn't have "control" of the mic, system audio, or the desktop. It transcribes the audio when you press 'start record' and only takes a screenshot when the 'take screenshot' option is selected. It's not an AI that 'takes control' of any system. In terms of privacy or system control this application doesn't do anything more than Siri, Alexa, or the ChatGPT voice feature and it is only prompted with the screenshots you take. But this is more than an AI voice screenshot app. My goal is for this to be the AI app to end the AI agency custom software industry.
The current state:
Billions going into model benchmarks and parameter counts
AI agencies charging six figures to build the same assistant tools over and over for every company
Meanwhile, we are not taking advantage of context created by the converstaions we are already having and the context created by what we already have on our screens —we have 1m+ token windows but businesses are still typing prompts into empty chat boxes or paying to build custom software for an LLM pointed at one narrow part of a workflow.
Small business owners are left out entirely because every solution requires either a PhD or a blank check. Medium and Large businesses are overpaying for brittle agentic workflows that are not future proof.
What actually matters: Getting the model into a rich context environment where it can see and hear what's actually happening in your business.

Chatting with LLMs can only yield a limited amount of insight from the because the model doesn't have the context of what you are hearing, saying, and looking at. If you are not giving the AI all the context of your conversation, along with a well crafted prompt, along with a fully loaded RAG —the answer you get will be broad and hedged. Generic, in other words. If you give the model rich context, only then will it be able to give a very specific insight for your situation.

You likely already know this because you have probably had the experience of typing paragraph after paragraph into one prompt after another, back and forth with an LLM. Back and forth and then after 20 or 30 prompts in the thread have you finally corrected the LLMs false assumptions about the situation you are asking it about. Meanwhile, according to reports only 10% of businesses have tried integrating AI into the workplace but well over 90% of employees are using the commercial LLMs at work, (ChatGPT, Gemini, etc.). This is too much typing at work! Typing to an AI that can only respond with the most generic answers.
How Kiru works:
Continuously transcribes system audio (calls, meetings, any sound through your computer)
Captures screen context on-demand when you query the AI
RAG knowledge base fed by your actual conversations and documents
Salesforce integration + API control to take action

These extra features are just the beginning. If you use a customer service phone call software on Windows and it is somewhat common I want to integrate it into Kiru (if they have adequate developer support). If you try Kiru for the 7 day free trial and don't like it because it does not have Jira, Google Calendar, or some other popular program please tell me about it at support@stratamos.com and I will likely built it into Kiru. That's the whole point of this project. To build the AI integration once so everyone can have it for a flat fee instead of some people paying tens of thousands for a brittle integration here or there. Integrations from AI that does not even have the full context.

My plan is launch Kiru first and start adding integrations upon request. Use me. Let me be your AI agency developer and I will make the custom connections from your AI queries to your software. The best thing about it is we don't have to haggle. $14 flat fee or 7 day free trial like everyone else. We don't even have to talk or meet. Just look at how the Salesforce integration works then write me an email. Explain what other platform you want connected and why.

Back to the app itself... Written in Rust. Perfoms so well it can run on a $500 Windows laptop—local Whisper transcription, Groq API calls for lightning-fast inference
$14/month flat rate (plus groq API cost - usually pennies) No cloud computing fee. No six-figure implementation contract. No AI expertise required.
The agencies are rebuilding the wheel every time. We're just giving you the wheel.
I'm here all day—ask me anything about the architecture, why context matters more than model size, or how we keep costs this low.
7 day free trial at stratamos.com