Pranav

Building

[Pranav @Pranavkerg]

@SentinelSCA At Cartha the check runs on the proposed tool call before it executes.

Allowed tool? Policy pass? Within budget? Needs approval?

Any fail → denied. Nothing reaches the system. Full replay kept.

Happy to compare approaches.

[Pranav @Pranavkerg]

@howardting Permissions become the bottleneck fast. Once agents start calling tools, you need a clear control plane for what they’re allowed to do. That’s the problem we’re focused on at Cartha.

[Pranav @Pranavkerg]

@ben_callow This is such an important point. Controls only matter if they sit on the live execution path. At Cartha we put the gate right before the tool call so the constraint actually applies.

[Pranav @Pranavkerg]

@niffywontmiss @0xSweep Agree. Once agents can take real actions, guardrails can’t just live in the prompt. They need to sit at the tool call level. That’s what we’re building at Cartha.

[Pranav @Pranavkerg]

@YoursTruly_IA This is exactly what we’re seeing. Autonomy looks great in demos. In production it becomes a control problem. At Cartha we focus on the layer that can deny a tool call before it runs.

[Pranav @Pranavkerg]

Your AI agent can talk.

The dangerous part is when it can do.

Cartha sits in the path of the tool call and can stop it before anything happens.

Refunds. CRM updates. Emails. Bookings.

Denied before they land.
https://t.co/2BoXfT3bFm

[Pranav @Pranavkerg]

@annasoreskin @X Thank you 😊

[Pranav @Pranavkerg]

Hey @X algorithm 👋

Looking to #connect with people building in AI:

🤖 AI agents
⚙️ n8n and automation
🔍 AI and SEO tinkerers
🚀 Solo founders
🏗️ Build in public
🧠 Prompt and context engineering
💸 Indie hacker economics

[Pranav @Pranavkerg]

@itsalicesoul Just shipped Cartha a lightweight governance layer for AI agents.
It adds tracing, memory, budgets, and policy guardrails so agents can actually run in production… with zero code setup.
https://t.co/2BoXfT3bFm
Would love feedback if anyone’s building agents right now.

[Pranav @Pranavkerg]

@akfullstack @X Sure

[Pranav @Pranavkerg]

@SamVale29 Just shipped Cartha a lightweight governance layer for AI agents.
It adds tracing, memory, budgets, and policy guardrails so agents can actually run in production… with zero code setup.
https://t.co/2BoXfT3bFm
Would love feedback if anyone’s building agents right now.

[Pranav @Pranavkerg]

@life_ofbryan Just shipped Cartha a lightweight governance layer for AI agents.
It adds tracing, memory, budgets, and policy guardrails so agents can actually run in production… with zero code setup.
https://t.co/2BoXfT3bFm
Would love feedback if anyone’s building agents right now.

[Pranav @Pranavkerg]

@annasoreskin @X Hey! Thanks for reaching out, happy to connect.

I’m building Cartha (control layer for AI agents that take real actions in production).

What are you working on?

[Pranav @Pranavkerg]

@jilyannori @X Thank you 😊

[Pranav @Pranavkerg]

Most AI tools show you the damage after the agent already acted.

Cartha sits in the path and can deny the tool call *before* it lands.

Agent wants to refund → blocked.
Stripe never gets called.

Live at https://t.co/2BoXfT3bFm

[Pranav @Pranavkerg]

@adelelecroix @X Hey! Thanks, happy to connect.
What are you building?

[Pranav @Pranavkerg]

@OrbanPeter3 @X Hey! Thanks for reaching out, always happy to connect with fellow founders.

I’m building Cartha, a control layer for AI agents that call tools in production (so you can deny risky actions before they execute).

Would love to hear what you’re working on.

[Pranav @Pranavkerg]

@aiwithadb Just shipped Cartha a lightweight governance layer for AI agents.
It adds tracing, memory, budgets, and policy guardrails so agents can actually run in production… with zero code setup.
https://t.co/2BoXfT3bFm
Would love feedback if anyone’s building agents right now.

[Pranav @Pranavkerg]

@conecteer No not yet, we are actively approaching potential start-ups that use autonomous ai agents who might need governance layer.