Rudran

The autonomous agent platform

Autonomous AI agents, each in its own isolated cloud sandbox.

Rudran runs every agent as a long-lived process on its own sandboxed machine, 24/7. Agents plan first and wait for your approval before anything happens. Then they work, and leave a timeline of every action they took.

Request accessSee how it works

Accounts are admitted in batches while sandbox capacity is limited.

One
isolated sandbox per agent
24/7
scheduled and unattended runs
Per agent
model choice, no house default
June 2026
running in production since

The problem

Agents need somewhere to actually live.

An agent that does real work needs a persistent machine, hours of runtime, tools that can change things, and a human who can stop it before it does. A chat session gives it none of those.

Three things break the moment you try to run agents for real:

  • Agents that can run commands and edit files need isolation from each other, not a shared process.
  • Work that matters needs a human decision before it happens, not an apology afterwards.
  • When an agent finishes, you need to see what it actually did, not just what it says it did.

How a run works

Plan first. Approve. Then execute.

The approval gate is not a setting you can forget to turn on. A sensitive action never executes without a recorded human decision, and the gate fails closed even when Rudran itself is degraded.

  1. You describe the work

    In a thread, in plain language. Pick the agent, or let a standing one pick it up on a schedule.

  2. The agent returns a plan

    And stops. Nothing has run yet. You read what it intends to do before it does any of it.

  3. You approve

    It executes across as many tool-calling turns as the work needs, pausing again for anything sensitive.

  4. You check the work

    Every action lands on a timeline you can read afterwards, with the result attached.

Architecture

One sandbox per agent.

Each agent gets its own isolated machine with its own filesystem, its own shell and its own process space. Agents hibernate when idle and wake when you or a schedule calls them.

How a Rudran run flows: from your workspace, through a per-agent isolated sandbox running the agent with its own tools, past an approval gate that holds sensitive actions for a human decision, out to the model you chose for that agent, and back to a run timeline that records every action.

Your workspace

Threads and schedules

You direct agents, or a schedule does.

One sandbox per agent

Agent — isolated sandbox

Own filesystem, own shell, own process space. Hibernates when idle.

Agent — isolated sandbox

Each agent gets its own machine, so one agent's files and processes are not another's.

Before anything sensitive

Approval gate

Fails closed. A sensitive action never runs without a recorded human decision.

Model per agent

Model gateway

Routes each agent to the model you chose for it.

Run timeline

Every step above writes to one readable record — the plan, each tool call, each approval and its result. It stays after the run ends, which is how you check the work rather than trusting it.

Supervision

You decide what an agent is allowed to do.

Autonomy is a dial, set per agent. Not a switch you flip once for everything you own.

Three supervision tiers
Approve everything, approve only sensitive actions, or let it run unattended. Set per agent, changed any time.
A gate that fails closed
A sensitive action never executes without a recorded human decision — including when the platform itself is degraded.
A hard safety floor
A backend-side floor refuses a class of destructive commands outright, whatever the agent or its instructions ask for. It has fired in production.
A readable trail
Every tool call, approval and result is recorded on the run timeline and stays there after the run ends.

What an agent can do

A real machine, not a text box.

Inside its sandbox an agent has the tools a person would have. Each one can be turned off per agent.

  • Web search

    Search the web and read page content.

  • Web browser

    Navigate and interact with websites — click, type, and fill forms.

  • Terminal

    Run shell commands inside the agent's own sandbox.

  • Run code

    Execute Python to compute results or call tools programmatically.

  • File access

    Read, write, and search files in the agent's workspace.

  • Sub-agents

    Split a plan across parallel sub-agents, each on a model you choose.

  • Text to speech

    Convert text to spoken audio.

  • Webhooks

    Receive work from any system that can POST JSON to a URL.

Models and infrastructure

Choose the model for each agent.

A research agent and a code agent do not want the same model. Rudran has no house default and takes no position on which model you should use. You pick, per agent, and change it later without rebuilding anything.

Moving to Google Cloud

Rudran runs today on a container sandbox provider with managed MongoDB and Redis tiers. The agent runtime is deliberately separated from its infrastructure behind a provider interface, so the platform underneath can change without the product above it changing.

The next provider behind that interface is Google Cloud:

  • GKE Agent Sandboxper-agent isolation with gVisor and snapshot-based suspend and resume for idle agents
  • Cloud Runlong-running and scheduled agent processes
  • Gemini Enterprise Agent PlatformGemini models as a first-class choice in the model picker
  • MongoDB Atlas on Google Cloud, and Memorystorethe run spine and its queue

This is planned work, not work already done: Gemini reaches the model picker in Q3 2026, and sandbox provider certification on GKE Agent Sandbox is targeted for Q4 2026.

Always on

Agents keep working when you are not there.

An agent can run on a schedule, or when a webhook fires. It does not need you watching, and it does not need a browser tab left open.

When an unattended run reaches something sensitive, it stops and emails you. The work waits for your decision rather than guessing.

The product

What it looks like in use.

These are the real screens, in the order you meet them.

Rudran agent creation screen: name, description and system prompt fields, an icon and theme picker, and a summary panel showing the model, supervision level set to approve sensitive actions, tools enabled, skills, knowledge and schedule.
Create an agent: a name, a playbook, a model, and how much supervision it needs.
Rudran run view with an agent paused at a plan for building a Python calculator module. The plan lists five unchecked tasks, an approve-plan action is pending, and the full plan document with its overview, design decisions, assumptions and risks is open alongside on the canvas.
It writes the plan, then stops. Nothing runs until you approve it.
Rudran run timeline showing a completed run as a sequence of tool calls — a read completing in 365 milliseconds and a shell command exiting zero in 78 milliseconds — each with its result, next to the document the agent produced on the canvas.
Every action lands on a timeline you can read afterwards, with its result and how long it took.

Pricing

Prepaid credits. No overage.

Work is paid for in credits you buy up front. When the balance reaches zero, work pauses. Rudran never bills past your balance, so there is no invoice you did not expect.

Full pricing
Free
A daily allowance while you try it. No card.
Prepaid credits
Buy credits, spend them across every agent you run.
No seats
You are charged for work done, not for people added.

Built by

Rudran Labs Inc.

Founder details not filled in yet.

This block is the “team” half of what a program reviewer validates, and it is the most likely single cause of an automated rejection. Fill in FOUNDER in apps/web/src/components/marketing/FounderCard.tsx — name, two-sentence bio, LinkedIn URL and a real headshot — before this site is submitted anywhere.

Give an agent a machine and a job.

Accounts are admitted in batches while sandbox capacity is limited. Ask for one and we will tell you where you are in the queue.

Request accessTalk to us

Rudran Labs Inc. is incorporating in Canada.