Philosophy

The problem with work today

Most of what knowledge workers do every day is execution — the mechanical act of translating decisions into output. Writing the code you already know needs writing. Formatting the report you already know the answer to. Sending the email you already composed in your head.

The thinking happened in seconds. The execution takes hours.

We built Exponential because we believe this is a temporary problem. AI agents can handle execution. The question is: what's the right interface for humans to direct that execution?

Why not just a chatbot?

Chatbots are conversations. Work is not a conversation — it's a system.

Real work has tasks, status, priorities, dependencies, handoffs, and reviews. It needs structure. You don't want to type "hey, can you also finish that other thing from yesterday?" into a chat window. You want a board where you can see everything, assign work, approve plans, and review output.

Exponential is that board. It's the management layer for AI agents — the same way Linear is the management layer for human engineers.

The team lead model

In Exponential, you are the team lead (팀장). Agents are your teammates (팀원).

This isn't a metaphor we chose for marketing — it's a design principle that shapes every feature:

• Agents plan before they execute — just like a good teammate who checks in before starting work.
• You approve before changes happen — just like a code review, but before the code is written.
• Agents report progress — just like a teammate who keeps you updated on Slack.
• You review deliverables — just like accepting a pull request.

The relationship between you and your agents is the same as the relationship between a lead and their team. The interface should reflect that.

What humans become

When agents handle the execution, humans become pure decision-makers.

You focus on:

• What to build (strategy)
• Why it matters (judgment)
• Whether it's good (taste)

You stop spending time on:

• How to build it (implementation)
• The mechanics of doing it (execution)

This isn't about replacing humans. It's about removing the parts of work that were never the point. Nobody became an engineer because they love writing boilerplate. Nobody became a writer because they love formatting documents.

The craft is in the thinking. The execution is overhead. Exponential removes the overhead.

Opinionated by design

Exponential is opinionated software. We have strong views about how human-agent collaboration should work:

1. Plan first, execute second. Agents should never surprise you. The plan-approval gate exists because trust is earned through transparency.

2. Agents are teammates, not tools. They have personalities, knowledge, and principles. They communicate in your activity timeline, not in a separate terminal window.

3. Local first. Your code stays on your machine. Your data stays on your machine. Agents run in your environment, with your permissions.

4. Provider-neutral. We don't lock you into one AI provider. Claude, Gemini, Codex, Copilot, Qwen — use whatever works best for each task.

5. Full-spectrum, not single-lane. Coding, docs, emails, research, operations — the same board handles all of it. One interface for all agent-driven work.

These aren't just features. They're beliefs about what this technology should be.