AWAIWorkflow
Alchemy
independent systems lab / 2026

A field lab for creative automation

Make the messy
stuff useful.

AIWorkflowAlchemy designs the layer between human imagination and machine execution: practical Python tools, agentic workflows, MCP systems, creative memory, and asset intelligence.

Enter the lab For creators, builders
and strange workflows.
FIELD NOTE
001 / 004
Creative materials flowing through a luminous system into organized digital assetsraw material → useful result
“The point is not to make the machine look intelligent. The point is to make the work feel lighter.”
PYTHON AUTOMATIONMCP SYSTEMSCREATIVE MEMORYASSET INTELLIGENCEHUMAN JUDGMENTPYTHON AUTOMATION

Less magic. More leverage.

Build the layer
that makes work flow.

Most creative tools stop at generation. The interesting work starts after the idea: naming, sorting, checking, connecting, remembering, documenting, and getting the thing out into the world.

A / 01

Agentic Python

Small, inspectable workflows that research, transform, classify, generate, and hand work back with context intact.

A / 02

MCP infrastructure

Deliberate connections between models, files, APIs, creative tools, and services—with permissions and audit trails.

A / 03

Creative memory

Project-aware memory with provenance, confidence, correction, and expiration instead of an undifferentiated context pile.

A / 04

Asset intelligence

Turn folders, images, audio, video, prompts, and exports into searchable, explainable creative inventory.

THE
MANIFEST
02

Good automation does not erase the human.

It clears the path to the part only the human can do: choose, notice, connect, decide. Every system here is designed to be understandable at zero, useful at one, and extensible when the work gets serious.

01 Explainable by default02 Reusable over clever03 Human checkpoints04 Documented handoff

Research with somewhere to go.

Ideas, but
with handles.

Short dispatches from the edge of agent architecture, memory, metadata, and creative operations—translated into decisions and experiments.

RESEARCH / 08.14.26

Your AI assistant doesn’t need more memory. It needs better memory governance.

Provenance, confidence, expiration, and correction are the difference between a useful creative partner and accumulated context.

Read the note ↗
BUILD LOG / IN PROGRESS

From a messy folder to a searchable creative inventory.

A practical Python pipeline for normalization, metadata inspection, duplicate detection, classification, and review.

Read the build log ↗
UP NEXT / 2026

The creator’s asset library is already an agent environment.

Every filename, prompt, revision, and export is a potential tool, signal, or piece of memory.

Browse the archive ↗

03 / the person behind the apparatus

Systems for people who have too many ideas and not enough hours.

AIWorkflowAlchemy is the public notebook and studio front door of Steven Chaplinski, a creative automation engineer building systems that create, organize, understand, and publish content at scale.

The work lives at the intersection of Python, AI integration, APIs, generative media, storytelling, and the gloriously unglamorous details that make an idea deliverable.

Start with the mess

04 / open channel

Bring the workflow
that keeps failing.

Folders, repetitive tasks, half-built tools, impossible briefs, or a system that almost works. There is usually something useful hiding in the mess.

hello@aiworkflowalchemy.com

Artifacts from the apparatus.

Not promises.
Working shapes.

Independent systems, prototypes, and strange little machines built to make creative operations less fragile.

CASE FILE / 001

Media Asset
Organizer

Because “final_final_v7” is not a metadata strategy.

Open file ↗
CASE FILE / 002

Creator
Tool Server

Memory, files, APIs, and agents in a room together. Carefully.

Open file ↗
CASE FILE / 003

Research-to-Content
Pipeline

Source collection → reasoning → review → something worth publishing.

Open file ↗

One studio / many surfaces

Everything talks
to everything. Eventually.

FILESimages · audio · video
prompts · exports
PYTHONscan · transform
classify · report
MCPtools · permissions
context · actions
MEMORYdecisions · preferences
provenance · expiry
HUMANchoose · notice
connect · decide