Work / Sandora

Sandora

An AI Department OS for building virtual AI teams, automated workflows, and human-governed business operations.

Sandora is a platform for creating a custom AI department or virtual company. Instead of working with a single chatbot, users can configure multiple AI agents across roles, departments, tools, data sources, workflows, approval gates, and audit logs.

Links
GitHub

We still cannot provide the latest source code publicly due to versioning, commercial, and privacy constraints. We are still working to distill it into the most concise public-ready version.

Overview
The project was previously named DropPilot and was initially locked to the dropshipping market. After seeing the broader potential of the product structure, I expanded it into Sandora: a general AI Department OS for individuals, startups, agencies, and small businesses, without tying it to one specific industry.
Problem
Small teams often deal with repeated work, fragmented tools, scattered data, and no system that coordinates business operations like a real team. Most AI tools solve isolated tasks, while real business work involves multiple roles, steps, decisions, tools, and human approval points.
Role
Founder / Product Builder / Systems Designer
Approach
Sandora is designed around company setup, department builder, agent builder, model/provider routing, integration hub, company brain, workflow builder, mission control, company chat, approval center, and analytics/audit logs. The core product direction combines AI agents, business workflow, and human governance.
Outcomes
AI Dept OS
virtual teams for business operations
Multi-agent
departments, tools, workflows, and routing
Governance
approval gates, audit logs, and cost control
  • Being piloted for repeated operational tasks inside small businesses under controlled scope.
  • Demonstrates that multi-agent systems are not only about the model, but also about tight control and cost optimization; the strongest model is not always the best choice.
  • Kept human-in-the-loop governance central for risky actions such as sending emails, publishing content, launching campaigns, writing data, or spending budget.

Project Highlights

Agent Directory

Browse, filter, and select AI agents by department, skills, status, model, and connected tools.

Configure Agent

Configure role, persona, model fallback, knowledge sources, tool access, permissions, and approval rules per agent.

Mission Control

Coordinate backlog, tasks, agent assignment, approval checkpoints, workflow automation, and impact estimates.

Company Chat

Department channels that connect discussion, tasks, meetings, files, summaries, and operating decisions.

Video & Walkthrough

Sandora
Walkthrough placeholder
00:00 / 00:00

Timeline

2026
01. Product repositioning
2026
02. Department and agent model
2026
03. Workflow and approval layer
2026
04. Industry templates
In progress
05. Public profile update

Behind The Project

From DropPilot to Sandora

Sandora was previously named DropPilot and was locked to the dropshipping market. After deeper product analysis, I found that the most valuable part was not one specific industry, but the platform structure itself: departments, agents, workflows, tool connections, task tracking, and human control.

Product wedge

The product is not only a chatbot or workflow automation layer. Sandora combines AI agents, business workflow, and human governance to model an operating team with roles, permissions, memory, tasks, and accountability.

Target impact

The main value is increasing productivity, reducing early operational cost, standardizing workflows, and keeping important actions controlled through approval queues and audit logs.

Details to update

Visuals, workflow demos, detailed architecture, and implementation outcomes will be added later once there is a public-ready version.

Gallery

Interested in this project?

Let's build practical edge and AI systems for your workflow.

Contact
Share this project: