Why we built an operating system, not another SaaS tool

Every company we've ever worked at has the same problem. You open your laptop in the morning and within an hour you've context-switched between your email client, your project tracker, your CRM, your document editor, your chat tool, your finance software, and three different dashboards that all show slightly different numbers.

This isn't a productivity problem. It's an architecture problem.

The integration tax

The modern SaaS stack was supposed to give us best-in-class tools for every job. And it did — at the cost of creating an entirely new category of work: keeping everything in sync. Companies now spend more time connecting their tools than using them.

We've seen teams with dedicated "integration engineers" whose entire job is maintaining Zapier workflows and custom API glue. We've watched sales teams manually copy contact data between their CRM and their invoicing software. We've sat in meetings where nobody could agree on the numbers because every tool had its own version of the truth.

The best tool for any job is the one that doesn't make you leave the place where you're already working.

This is what convinced us to build Tedo. Not because the world needed another note-taking app or another CRM. But because nobody had tried to make all of these things actually live in the same place, sharing the same data, governed by the same permissions, accessible through the same interface.

What "operating system" actually means

When we say Tedo is an operating system for work, we mean it literally. The same way macOS provides a file system, a permission model, and a set of conventions that all applications share — Tedo provides a workspace with nodes, a permission model, and a set of conventions that all business applications share.

Every piece of data in Tedo is a node. A note is a node. A contact is a node. An invoice is a node. A chat room is a node. They all live in the same hierarchical file system. They all have version history. They all support the same permission model. And they can all reference each other.

This means that when you create an invoice for a client, the invoice already knows who the client is — because the client's contact record exists in the same workspace. When an AI agent processes a document, it can create tasks, update contacts, and trigger automations — because it has access to the same data layer as everything else.

The hard parts

Building an operating system is significantly harder than building a single application. Here are three problems we had to solve:

1. The data model has to be universal

Every application has different data requirements. A note is freeform text. A table has typed columns. A CRM contact has structured fields. An automation has triggers and steps. But all of these need to coexist in the same node tree, be searchable through the same interface, and be accessible through the same API.

We solved this with a layered architecture. The node layer handles identity, hierarchy, permissions, and versioning. The application layer handles kind-specific data. A note node has a corresponding record in the notes table. A table node has corresponding records in the tables, fields, and records tables. But both are nodes first.

2. Applications need to be truly independent

Each application in Tedo is a self-registering module. It declares its resources, its routes, its UI templates, and its capabilities. The operating system discovers applications at boot time and wires them into the workspace automatically.

This means you can enable or disable applications per workspace. A design studio doesn't need the finance app. A solo developer doesn't need the CRM. But when they do need it, it's already connected — no integration setup required.

3. AI has to be a first-class citizen

Most tools bolt AI on as an afterthought. A chat window here, an autocomplete there. We built Tedo with AI agents as a core concept from day one. Agents are workspace members. They have permissions. They can be assigned to teams. They can read documents, search the web, manage contacts, and execute multi-step workflows.

This isn't a chatbot. This is a colleague that happens to be software.

What we've shipped so far

Tedo currently includes over 40 applications:

Notes & Notebooks — Rich text documents and multi-cell notebooks
Lists — Todo lists with nested items and checkboxes
Tables — Relational databases with typed fields
CRM — Contacts, organizations, roles, and contact bases
Sales — Pipelines, deals, and leads
Finance — Double-entry bookkeeping, invoicing, expenses, bank reconciliation
Chat — Real-time messaging rooms
Mailbox — Email management with Gmail integration
AI Studio — Build and manage AI agents
Automation — Workflow triggers, conditions, and actions
Search — Regional search engines powered by our own crawled index

And more are coming. Because the operating system already exists, adding a new application is a matter of defining its data model, its UI, and its capabilities. The node system, permissions, versioning, search, and AI integration come for free.

Where we're going

The traditional approach to software quality relies on massive scale. Google needs billions of searches to know which results are good. Salesforce needs millions of users to know which features matter. This creates an insurmountable moat for incumbents.

We believe AI agents can replace that feedback loop entirely. An agent that reads a webpage, extracts structured data, and evaluates content quality does what millions of user clicks would reveal — but with actual semantic understanding. From day one.

This is the foundation of everything we're building. An army of AI agents that discover, validate, enrich, and improve data continuously. Not because billions of users told us what's good — but because the agents can figure it out themselves.

If this resonates with you, we'd love for you to try Tedo. It's free to get started, and we think you'll be surprised how much simpler your workday becomes when everything lives in one place.