Kriyagniai

Context Engineering

[Kriyagniai @languageofai]

Next Frontier is here.

Mandana AI

A Reasoning orchestration.

Built in-
1. Quick Reasoning
2. Deep Reasoning

Quick Reasoning gets you there fast.
Deep Reasoning slows down to get you there right.

Built for Complex Problems.

- Researchers
- Product Managers
- Founders

[Kriyagniai @languageofai]

Launching Soon.

#ai #stealth #launch #startup #India

[Kriyagniai @languageofai]

@0xCodila We said it earlier - Start from the limitation first. Understand where AI breaks then see the magic.

[Kriyagniai @languageofai]

@sama I start distilling models the moment they are launched with my 600 Agentic workflows across different domains. Results are promising. Already an acquisition offer from a SFT startup. Would like to join OpenAI. Thank you.

[Kriyagniai @languageofai]

Do we really need a multi agent system or does one system can simulate multi agent behavior just on the basis of agentic prompt engineering?

End of Framework era. It’s time to rethink fundamentally.

#ai #agenticai #prompt

[Kriyagniai @languageofai]

Those who are newly joining college- Start Asking the right questions. Because Knowledge has no value left.

Those who are in school - Focus on Maths.

World Order is changing.

Kriyagni AI

#ai #newworld #education #transformation #India #memoryofdada

[Kriyagniai @languageofai]

Chinese universities cut more than 12,000 degrees from their universities.

Reason - AI.

What’s for us ?

Those are graduating in 2026 - Start building AI projects. Not GitHub copy paste projects. Think & Build.

[Kriyagniai @languageofai]

structured, confidence aware & ethical.

Replacing the RLHF with AI agents.

Ethical | Reliable | Confidence aware AI
KRIYAGNI AI

#ai #trainingdata #llm #agents

[Kriyagniai @languageofai]

@heynavtoor We solved it bro.

[Kriyagniai @languageofai]

China is already distilling Claude fable.

Government has strictly warned them not to distill after 22 June.

#ai #claude #fable

[Kriyagniai @languageofai]

Solving the biggest bottleneck in AI world - Training data.

Agentic workflow for training data generation.

KRIYAGNI AI is coming out of stealth soon.

Train AI that thinks.

Stay Tuned.

#ai #trainingdata #agenticai #llm

[Kriyagniai @languageofai]

One more.
Query : What is Zepto’s current valuation and funding status? Give me a confidence score on each finding.

[Kriyagniai @languageofai]

@elonmusk You need SLM data instead of Hormuz. IPO is coming

[Kriyagniai @languageofai]

Worst voice AI assistant @bankofbaroda.

Hire better engineers or handover me the workflow.

I will make it best.

[Kriyagniai @languageofai]

We built an AI tool that maps daily current affairs directly to the UPSC syllabus.

Less time decoding news.
More time understanding and revising.

Built for serious UPSC aspirants.

KRIYAGNI AI

#UPSC #AI #CurrentAffairs

[Kriyagniai @languageofai]

Building in AI era is addictive if you know what to build.

Domain expertise + Product thinking + Agentic AI workflows + Storytelling = New Solutions for the world

Mahesh

#ai #building #startup #product #agentic

[Kriyagniai @languageofai]

I build AI scientists @elonmusk.

I am the best persona prompt engineer in the world rn from India.

You can test.

Thank you.

Mahesh

[Kriyagniai @languageofai]

Who says AI isn’t conscious ?
@claudeai

[Kriyagniai @languageofai]

Can explicit reward criteria in prompts replicate RLHF-trained behavior?

Mahesh

#ai #question #prompt #help

[Kriyagniai @languageofai]

The future of AI training is not human labeling.

It’s Parametric Knowledge Agents.

Kriyagni AI

#ai #trainingdata #llm