Browse categories
Explore
Fiverr Pro
English
$
USD
I will setup openhuman neoortex memory ai agent vps mac mini hermes obsidian autofetch
Thierry Dan
About this gig
Your AI resets every session. You re-explain your stack, your calendar, your team and context evaporates. A smarter amnesiac is still an amnesiac. That's the gap OpenHuman fixes with Neoortex.
I install and configure OpenHuman's Neoortex memory AI agents so your assistant already knows you before you type a word. Autofetch syncs Gmail, Notion, GitHub, Slack, and 118+ integrations every 20 minutes automatically. Add Hermes on VPS or Mac Mini for a persistent ai agent with Obsidian vault memory and a subconscious that runs 10,000 background loops daily at under $1 compute cost.
What I build:
- OpenHuman install: VPS, Mac Mini, Ubuntu, Hetzner, DigitalOcean
- Neoortex memory tree + Obsidian vault sync
- Autofetch: Gmail · Notion · GitHub · Slack · Google Calendar
- Subconscious background loop configuration
- Hermes Agent or n8n workflow integration
- Ollama local ai agent (zero cloud dependency)
- Token Juice: up to 80% token cost reduction
- Fix: auth failures, sync errors, integration issues
Tools:
OpenHuman · Neoortex · Hermes · n8n · Docker · Obsidian · Ollama · ElevenLabs · OpenRouter · Claude API · OpenAI
Message me with your setup I'll confirm the build before you order.
Get to know Thierry Dan
Thierry Dan
Your OpenClaw AI Agents Automation, Trading Bot Developer
- FromUnited Kingdom
- Member sinceApr 2026
- Avg. response time1 hour
Languages
English, French, Spanish
AI Engineer & Trading Bot Developer.
I build production Hermes & OpenClaw multi-agent systems from Claude Code-to-Hermes Slack bot migrations and multi-agent architectures to autonomous Forex/Crypto/Polymarket trading bots (MT4/MT5, Binance, TradingView, SMC/ICT/MMC).
I configure MCP integrations, n8n & ManyChat pipelines, RAG memory, Docker sandboxing, and local/cloud LLMs (Ollama, OpenRouter, Anthropic). Stack: Python, Node.js, TypeScript, React, Pine Script. I wire Claude Code & Codex as AI engines, handle API integrations end-to-end, and ship production systems not demos. Let's build.
FAQ
How does Neoortex index context differently from standard vector similarity search in other AI memory tools?
Neoortex builds a cross-source entity graph. Gmail, Slack, and Notion data resolves to one node, not three fragments. Tiered memory (hot/warm/cool/cold) sorts by recency. Zero LLM dependency means 10M tokens indexed in 10 seconds at 10x lower cost.
How do you configure the autofetch sync loop to pull Gmail and Notion every 20 minutes without manual polling?
I connect Gmail and Notion via OAuth inside OpenHuman. No API key needed. I set the autofetch sync interval, test the delta pull, and confirm live ingestion into the Neoortex memory tree. Your handover guide covers the 2-click reauthorization if any OAuth token expires.
Can you deploy OpenHuman on a VPS so it runs 24/7 without depending on my local machine staying online?
Yes. Laptop sleep stops the OpenHuman memory tree sync, autofetch loop, and subconscious tasks. I deploy on Hostinger, Hetzner, DigitalOcean, or Contabo with a process manager for auto-restart and persistent OAuth polling so your agent runs 24/7.
How does the Obsidian vault sync work and can I browse and edit my memory tree outside OpenHuman?
OpenHuman writes data as 3,000-token markdown chunks into an Obsidian-compatible wiki folder on your device. I set up the vault path, test read and write, and confirm the agent picks up your edits on the next ingest. You can browse and edit the memory tree directly in Obsidian.
How do you configure the subconscious background loop and what does the output routing look like in production?
The subconscious runs 10,000 plus global memory recalls per day at under $1. It sweeps the full knowledge graph using recency, relevance, and randomness. I set the recall frequency, route outputs to Telegram or Slack, and tune the mirror test so results match your real priorities.
Can you run Hermes Agent alongside OpenHuman on the same VPS without state conflicts or memory overlap?
Yes. Hermes handles task execution and OpenHuman builds persistent identity context. I run both in separate Docker containers with no shared memory. An n8n routing layer sends execution tasks to Hermes and context queries to OpenHuman. Full config files included on delivery.
How does Token Juice reduce API spend and what inputs see the most compression in a live deployment?
Token Juice cleans every tool call result, email body, and web scrape before it reaches any LLM. HTML becomes markdown, long URLs shorten, and extra whitespace is removed. This gives up to 80% token reduction. I confirm Token Juice is active and include token counts in your Loom.
What is your process for setting up Ollama as the local AI model backend to eliminate cloud API dependency entirely?
I install the Ollama backend, pull your chosen model such as Llama 3.1, Mistral, or Phi-3, and connect it to OpenHuman's local endpoint. When the app starts, Ollama starts with it. No cloud API needed. Minimum 16GB RAM is recommended for models above 7B parameters.
Can OpenHuman's mascot join Google Meet as a visible participant and how do you configure the voice pipeline?
Yes. I connect Google Meet via OAuth, set up speech-to-text input and ElevenLabs voice output, and test the mascot as a live meeting participant. This needs the desktop app with a display. Headless VPS needs a virtual display setup. I check your hardware before you order.
Can you migrate an existing OpenHuman setup from a local machine to a VPS without losing the memory tree?
Yes. I export your SQLite memory tree, Obsidian vault files, Docker config, and environment files and redeploy on the new VPS. All memory graph data, entity relationships, tiered memory state, and vault notes carry over. Timeline is 24 to 48 hours based on vault size.
