# Agent operators and AI agents running admin for small companies
> Small companies run the same admin as large ones, with no operations team to own it. Twistag built Supermanager, an AI agent operations company where AI agents do that work and a named agent operator owns the result. Six months after launch, around 20 companies run their admin on it, with more than 100 agents at work.

Client: Supermanager · Industry: Software & platforms · Practices: AI & Agents, Product Engineering · Timeline: Ongoing since 2025
## Key results
- **100+** AI agents at work for clients
- **~20** small companies run on the platform
- **6 months** in market
## The challenge

**Admin everyone does, no one owns.**

Twistag built Supermanager, an AI-native company that runs the everyday admin of micro and small companies, with AI agents doing the work and an agent operator owning the result. The company was created inside Twistag's own stack, by Twistag's team, and now runs separately from the Twistag group. Six months after launch it serves around 20 companies with more than 100 agents at work.

A five-person company runs the same admin as a large one: invoices, suppliers, leads, travel and payroll. It has no operations team, so the work lands on the founder. An assistant adds hands but no system. Automation tools need someone technical to build and maintain them. An AI agent that nobody checks is a risk a small company cannot carry. Founders were paying for growth and spending their week on admin.

## The approach

**Agents do the work. An agent operator owns it.**

1. **Map the work first.** A short call with an agent operator finds where a team's hours go and sets what to automate first. The roadmap comes before any build. 
2. **Automate inside existing tools.** Agents on Amazon Bedrock run workflows across sales, finance, operations, people and customer service, in the tools a client already uses. Clients with the technical confidence build their own automations; for everyone else, the operator does. 
3. **Put a person on every run.** Operators run their clients' workflows from one platform, built so that people without an engineering background can create, run and check automations for several companies at once. 
4. **Act for the client, with the client.** Payments, recurring work, a vault for logins and documents, and the client's own team all sit on the same platform, with the client's approval wherever money or access is involved. 

## The impact

**A company that runs on agents.**

Around 20 small companies run their admin on Supermanager, with more than 100 agents at work, six months after launch. The split holds at that size: agents take the repeatable steps, and operators own the result, the exceptions and the client relationship. The platform is the durable asset. Because one operator can build and run workflows for several companies, a new client does not mean a new operations hire. For Twistag, Supermanager is the proof of what it builds for clients: an AI-native company, from product and agents to operations, built end to end. Twistag has not published figures for hours saved per client or companies per operator.

> Agents can run most of a small company's admin. Someone still has to answer for it.

## Technologies used

- AWS Bedrock
- Supabase
- Next.js
- AWS

> Twistag built Supermanager, an AI-native company where AI agents and agent operators run the everyday admin of small companies.

---

Source: https://twistag.com/case-studies/ai-native-admin-company-supermanager
