CREATIVE TECHNOLOGIST

I build worlds and the machines that build them.

I build open-source AI agents that do real work inside your company: they ship code, run the backlog, follow your calls and answer your customers. And they stay yours.

Let’s work together Try Xnode

A DECADE OF 3D & VFX FOR FILM, GAMES & DIGITAL FASHION

TEAM TRAINING·CUSTOM TOOLS·AI AGENTS·AUTOMATION·WEB APPLICATIONS·CHATBOTS·DASHBOARDS·VFX·3D·COMPOSITING·GAME DEVELOPMENT·BLOCKCHAIN / WEB3·HERMES / OPENCLAW·CREATIVE DIRECTION·POST-PRODUCTION·AI PIPELINES·GENAI PIPELINES·AI AGENT VPS MANAGEMENT·MESSAGING APP AGENTS·SOCIAL MEDIA CONTENT AUTOMATION·

WHAT I BUILD

AI agents that actually work. Not chatbots.

A chatbot answers questions. An agent takes real actions inside your tools, so it works with clear permissions and a human in control. A few jobs the agents I build already do:

01DEVELOPER
Keeps company code alive
Example: this website is maintained by an agent. It reads the brief, edits the code in three languages, checks the layout and prepares a private preview.
02PRODUCER
Keeps the backlog moving
Example: every morning it updates the board in ClickUp or Google Sheets, flags what is late and pings whoever owns it.
03MEETINGS + GOOGLE
Turns calls into actions
Example: it joins the Google Meet, then sends the recap through Gmail, books follow-ups in Calendar and files the documents in Drive.
04SALES + SUPPORT
Answers customers 24/7
Example: it replies on WhatsApp and Telegram, qualifies leads and hands the conversation to a person when needed. Tested in production.
05CREATIVE PIPELINE
Any media provider, no lock-in
Example: it generates image and video variants for campaigns, picking the best provider for quality and cost each time. A huge advantage for creative agencies.
06CONTENT OPS
One idea, every channel
Example: it turns a post on X into a ready LinkedIn post with its image, and publishes it on schedule.

What would you automate?

OWN THE SYSTEM

Your agent should not belong to your AI provider.

Codex and Claude Code are great for learning. But they depend entirely on one company: if it changes prices, retires a model or restricts access in your country tomorrow, you can only adapt. Learn with them, then move to an agent you control.

RENTED AGENT / FAST START

The provider chooses models, prices and availability
Supported connectors define what the agent can reach
Switching platform can mean rebuilding context and workflows

OWNED AGENT / BUILT TO MOVE

Choose OpenAI, Anthropic, DeepSeek, local models or the next provider
Keep memory, data and operating logic when models change
Connect any authorized API, internal tool or communication channel

WHAT IS HERMES?

An open-source agent that lives on your server.

Hermes Agent is an open-source framework (MIT licence) from Nous Research. It runs on your own server or computer, remembers context between sessions, learns reusable procedures (skills) and works with any model: OpenAI, Anthropic, DeepSeek, OpenRouter or local models. You talk to it from the terminal, Telegram, WhatsApp, Slack, email and more.

That is why it can become whatever you need: your work terminal, a developer, a producer, a voice for your customers or a game partner. Alternatives such as OpenClaw exist; I build on Hermes.

“I ALREADY HAVE CLAUDE CODE. WHY SHOULD I SWITCH SYSTEMS?”

Because your business should not depend on one company.

With Claude Code or Codex you depend on one company: if it raises prices you pay, if it retires a model you adapt, if a political decision blocks the service in your country you stop. Pay or switch off. The agents I build on Hermes are yours:

They run on your server, not on someone else’s account.
If a provider raises prices or cuts you off, you move to another one or to a local model and keep working.
Memory, data and workflows stay yours when you change model.
I install them, shape them around your case and maintain them over time.

You pay for a system you own, not for access to someone else’s.

AGILE WORK / THE AGENT AS THE COMPANY TERMINAL

One agent at the core. Work never stalls.

No more people chasing each other across calls, chats and email. The agent becomes the terminal at the centre of the company: everyone brings their own expertise, hands it to the agent, and work keeps circulating between people, agents and people again. Not a personal assistant: the company’s shared brain.

01A shared brain

The agent keeps the company’s memory: projects, decisions, files, history. Context no longer lives in one person’s head.

02It knows who you are

It recognises the people it works with and what each one is allowed to do: who approves, who publishes, who can only read.

03Handoffs without calls

Someone works through the night; the next morning a colleague asks the agent what was done and picks it up from there. Nobody chases anybody.

04No bottlenecks

Work flows in a loop between people and specialist agents, around the clock, instead of waiting for the next meeting.

It is like having a creature that works with the team: the more you use it, the better it knows the company.

REAL WORK

Real projects, in production.

Some of the main companies where I brought in and built AI systems, and the world I come from.

HOUSE OF CREATORS / CREATOR MANAGEMENT AGENCY

Technical and creative AI pipelines, end to end

I ran the agency’s AI pipelines end to end, from infrastructure to creative output. I designed workflows for adult content where mainstream models stop: purpose-picked models and providers, with consent and usage rights checked. I also built an MCP server for one of their launch products, so agents can use it directly.

Technical + creative AI pipelines
NSFW workflows on providers that allow it
MCP server for a launch product

AFFILROI / BETTING AFFILIATE NETWORK

From one poster to endless on-brand posts

They started from a single poster and random prompts written in ChatGPT. I built a social pipeline where you change a few variables (event, copy, format, channel) and generate unlimited posts and variations, always inside the brand identity.

Variable-driven templates
Unlimited variations
Brand identity locked in

BEFORE AGENTS / 3D, VFX, WEB3

Still active in 3D, games and Web3

I keep working with Decentraland, in 3D and in games, including on blockchain: smart contracts, digital assets and Web3 experiences. The 3D and VFX work for brands such as Versace, Nike and The North Face is in the archive below.

Decentraland
Smart contracts + Web3
3D, VFX, game development

HOW I HELP

From random experiments to a production pipeline.

AI that works inside your business, not beside it.

01 / TRAINING

Training & creative workflows

Hands-on training for image, video and audio production. Build reusable workflows together, with the knowledge your team needs to run them.

02 / CUSTOM TOOLS

Custom applications

Turn a complex process into a tool people can use: creative interfaces, internal apps and dashboards built around your needs.

03 / AUTOMATION

Agents & automation

Connect knowledge and tools to handle repeatable tasks, with clear permissions, human review and visible results.

ADVANCED OPTIONS & INTEGRATIONS

GENAI PIPELINES

Models & LoRA training

Custom LoRA training and self-hosted models where appropriate. Adult-content workflows only where the model and provider permit them, with consent, usage rights and moderation requirements defined upfront.

INFRASTRUCTURE

Agent VPS management

Deployment, monitoring and maintenance of always-on agents on your own servers, with secrets, logs and permissions handled properly.

CHANNELS

Messaging & social automation

Connect messaging, email, CRM, databases and APIs. Drafting, scheduling and publishing depend on channel support and approved permissions.

CUSTOM TOOLS / FROM METHOD TO APPLICATION

Your workflow. A tool built around it.

Applications are built around how your business actually creates content.
Xnode is one example: a visual tool for workflows your team can reuse, share and automate.

Xnode.

WHAT YOU CAN DO WITH XNODE

01
AI GENERATION
Images and video
Generate images and video with the models and providers you prefer.
02
LORA
Your style, every time
Use custom LoRA to keep faces, products and style consistent.
03
LLM
Text and data
Chain multiple LLMs and analyse data inside the same workflow.
04
AUDIO
Voice and sound
Speech, TTS, voice processing, equaliser and pitch.
05
COMPOSITING
Post‑production finish
Compositing, masking and colour correction without leaving the graph.
06
DEPTH + 3D
Depth and motion
Depth Anything and 3D-to-mocap (reverse mocap).
07
FOR EACH
One generation, a hundred variants
Automate generations across many inputs: prompts, images, products.
08
WEBHOOK
Your pipeline becomes a service
Run the workflow via webhook: it becomes an HTTP microservice.
09
NETWORK NODE
Connect it to anything
Control your workflows from n8n, Pabbly or Zapier.
10
DELIVERY
From graph to channel
Post straight to social media or deliver to cloud storage.
11
JSON
Reusable workflows
Save every workflow as JSON and reuse it whenever you need.

REPEATABLE

The process is the asset

Once a flow works, it is saved as a graph, not as one person's memory. Anyone on the team can open it, run it and improve it.

BYOK

Your keys, your costs

Bring your own API keys (BYOK) and pay providers directly. Choose models with their pricing in view; costs depend on usage.

AGENTS + PUBLISH

From graph to channel

Connect workflows to agents and Pabbly automations for repeatable runs and supported publishing channels.

PRO TOOLS

Control the finishing steps

Use LoRA, upscaling, compositing and export steps to shape the result, not just generate a first draft.

NSFW READY

Full creative freedom. Including NSFW.

Connect models and providers that permit adult content, or self-host your own. Xnode runs the same visual workflow for everything you're allowed to make.

Try Xnode

WHERE I COME FROM