Satish

Backend & MongoDB Senior Advisor @ Cigna | Scalable Systems, Databases, AI Agents | Building Arasthoo | Practical tech threads for devs

[Satish @sattylangayan]

Feeding LLM real data is a big security risk, breaks SOC2 & HIPPA compliance . We developed "andi-ai" as a secure Agentic AI data gateway sitting between AI agent and database.
Github: https://t.co/6e5bJk123K
@MongoDB #AgenticAI #BuildInPublic #TechCommunity

[Satish @sattylangayan]

I Benchmarked ANDI-AI Against `sample_mflix`: Strong Query Workflows, Incomplete Firewall Controls. Working on Firewall fix.
Feel free to test: pip install andi-ai
https://t.co/6e5bJk123K
@MongoDB #AgenticAI #BuildInPublic #TechCommunity

[Satish @sattylangayan]

I just published I Benchmarked ANDI-AI Against MongoDB’s sample_mflix. Here’s What Broke.
https://t.co/3WapavxOjp

[Satish @sattylangayan]

@anupamrjp https://t.co/6e5bJk123K

[Satish @sattylangayan]

@lightsilver323 Agentic Gateway

[Satish @sattylangayan]

pip install andi-ai

Github link: https://t.co/6e5bJk123K

[Satish @sattylangayan]

I failed to establish risk free communication between mongodb database and cloud llms. So I build andi-ai. A simple python package that act as AI firewall to connect with llms. 5 step deterministic query engine that generate queries just by analysing collection medatadata.

[Satish @sattylangayan]

Package name update: VectorDBA is now ANDI-AI.
ANDI: Advance Natural Language Database Interface. It is a Secure Agentic AI Data Gateway for Enterprise MongoDB.
Install: pip install andi-ai
@MongoDB @pypi #agenticai

[Satish @sattylangayan]

I just published Stop Writing 50 CRUD Endpoints: Building a Dynamic, Single-Endpoint Backend with VectorDBA and… https://t.co/xYxISuaoP8

[Satish @sattylangayan]

I just published VectorDBA(Agentic MongoDB NLP Database Interface) vs. Mongo MCP Server: What’s the Difference? https://t.co/2HAhIotosk

[Satish @sattylangayan]

@ShubhamRanga_ there you go https://t.co/GxfEAfDsg7

[Satish @sattylangayan]

Building AI apps is fun until you have to manage vector DB schemas and messy syntax.
I built VectorDBA, a Python management layer that handles:
- Schema inference & local caching
- Clean, abstracted query execution
- Instant boilerplate DB setup
It essentially acts like an ORM for your vector data.

[Satish @sattylangayan]

@ShubhamRanga_ Then we will soon have enlightened OshoAI channel on YouTube ?

[Satish @sattylangayan]

Karpathy-to-Anthropic would break AI Twitter—the first part is a given, the second we wait and see. But no matter who wins, we’re moving toward a world of riches or rags. AI will either make you very rich or take your job.
#Anthropic #xAI #Claude

[Satish @sattylangayan]

@Iamjuscelino Sure true. More than that they are becoming architect now.

[Satish @sattylangayan]

Last night’s AI signal: agents are graduating from demos to infrastructure. Anthropic bought SDK-automation startup Stainless, Google’s ADK/Genkit push agent pipelines, and Claude is becoming a front door to SandboxAQ drug-discovery models. #AIAgents #AgenticAI #GenAI

[Satish @sattylangayan]

Chatbots were the warm-up. The hotter trend: multimodal AI agents that can see, hear, code, call tools and automate workflows—still early, but moving fast as on-device NPUs meet cloud AI infrastructure. #EmergingTech #TechTrends #AiAgents

[Satish @sattylangayan]

Synthetic data is emerging as a game-changer for AI, helping businesses save costs and improve privacy. By creating artificial datasets, companies can train AI models faster and more safely, addressing the growing data shortage and ensuring fairer outcomes in sectors like healthcare and finance.
#AI #healthcare #Finanace