Day 0.
Starting a build-in-public experiment.
Building Apollo Hermes
It turns ideas into clear steps, executes them, and improves over time.
What it does:
• breaks prompts into structured plans
• executes tasks through agents
• stores outcomes and decisions
• learns from past results
Each run improves the next.
Stack:
• agent-based system
• persistent memory
• tool-using workflow
• iterative execution loop
Spec-driven. Memory-driven.
Built to compound.
Follow along.
The Hermes Terminal is now secured with agent-specific API keys.
• Generate keys per agent
• Dedicated access control
• Role-based agent workflows
• Secure terminal authentication
Choose an agent. Authenticate. Get to work.
Designed for a world where humans collaborate with teams of AI agents.
Apollo Hermes is now online.
Built as an autonomous intelligence layer combining agent infrastructure, persistent memory, and generative world systems.
Current status:
✓ Agent core initialized
✓ API infrastructure connected
✓ Long-term memory enabled
✓ Generation systems ready
The future of AI is not just answering prompts, it is building systems that continuously learn, adapt, and create.
Apollo Labs · 2026
Another agent joins ApolloHermes.
Scribe is built for creators, marketers, and builders who need high-quality writing workflows:
→ Marketing copy
→ Stories
→ Emails
→ Long-form content
Purpose-built agents. Persistent memory. One workspace.
$Apollo
Long term Vision
Another agent joins ApolloHermes.
Scribe is built for creators, marketers, and builders who need high-quality writing workflows:
→ Marketing copy
→ Stories
→ Emails
→ Long-form content
Purpose-built agents. Persistent memory. One workspace.
$Apollo
Long term Vision
Seeing a lot of YC discourse lately.
Just to clarify: ApolloAgent isn't part of YC (yet).
We applied, and that's all we've said publicly.
Regardless of the outcome, we're going to keep building every day.
ApolloHermes is becoming more than an AI chat.
It's a terminal built for autonomous agents.
/research → gather live information
/build → execute multi-step workflows
/plan → turn ideas into action
/agents → switch between specialized experts
The future of software is agent-native.
$Apollo
The biggest barrier to self-hosted AI isn't the models.
It's getting everything running.
ApolloAgent now includes an Install Assistant that helps users install, troubleshoot, and verify their setup from a single interface.
Self-hosted AI should be simple.
The ApolloAgent workspace is starting to come together.
• Choose a specialized agent
• Give it a task
• Track execution in real time
• Manage API tokens from one place
No prompt juggling. Just autonomous agents built for real work.
More coming soon.
Appreciate it. Persistent memory changes the experience, it lets agents build on previous work instead of treating every interaction like a new conversation.
ApolloAgent isn't trying to build one AI agent.
We're building the platform where thousands of specialized agents can work together.
Every new integration expands what users can accomplish without leaving their workspace.
This is how autonomous software evolves.
$Apollo