Heads up: This page is optimised for larger devices (desktop or tablet). On smaller screens, some layouts may look a bit off. For the best experience, view at 1024px width or above—or rotate your device.
Realness ©2025

01 / MINDSET

CLARITY
OVER NOISE

The real challenge isn’t chasing trends — it’s solving problems that matter. Every project is approached as an opportunity to reduce complexity, harness AI, and design solutions that endure.

01 / 04

02 / CRAFT

DESIGN WITH
STRATEGIC INTENT

Every digital experience is shaped with purpose. Strategy defines the why, design defines the how, and together they deliver impact that goes beyond aesthetics.

02 / 04

03 / FUSION

THE EDGES ARE
WHERE IDEAS LIVE

Innovation thrives in the spaces between — where strategic vision meets technical execution, and where human creativity collides with AI-driven potential.

03 / 04

04 / PERSPECTIVE

GLOBAL EXPERIENCE
SINGULAR FOCUS

From London’s creative circles to Germany’s industrial hubs, diverse experience fuels solutions grounded in clarity and purpose.

04 / 04

Get in touch

Contact

ex Inc.
Touchdown Dr 1176
al@ichbinhans.com
[00] 11 - 76
[01] 11 - 76
Works
Here, Adobe Express and Firefly were employed for generating images and designing the layout of social media advertisements. AI VFX was then utilised as an innovative method to repurpose content and boost engagement.


Hier wurden Adobe Express und Firefly zur Bildgenerierung und Gestaltung der Layouts von Social-Media-Anzeigen eingesetzt. AI VFX wurde dann als innovative Methode genutzt, um Inhalte wiederzuverwenden und das Engagement zu steigern.
SoMe Ads + AI VFX
Works
This collection showcases videos created through AI-powered workflows — where creativity meets technology. Each piece is the result of layered tools and processes, combined to explore what’s possible when human direction and machine intelligence work hand in hand.

Diese Sammlung präsentiert Videos, die durch KI-gestützte Workflows entstanden sind – dort, wo Kreativität und Technologie aufeinandertreffen. Jedes Werk ist das Ergebnis vielschichtiger Tools und Prozesse, die zeigen, was möglich wird, wenn menschliche Ideen und maschinelle Intelligenz Hand in Hand arbeiten.
GEN AI TRAINING
Works
Vivian, an exceptionally talented designer and Royal College of Art graduate, decided to launch her own distinctive boot brand in 2023. She had only her pencil sketches and needed to transform her mental images into visual reality. So we fine-tuned a model on a bespoke style dataset, which wasn't straightforward back then. But it worked brilliantly. We captured the depth and texture of her style book, and the retro plastic aesthetic for her lookbook, exactly as she envisioned it. I still look at these with great pleasure.

Vivian, eine außergewöhnlich talentierte Designerin und Absolventin des Royal College of Art, entschied sich 2023 dazu, ihre eigene unverwechselbare Stiefelmarke zu lancieren. Sie hatte nur ihre Bleistiftskizzen und musste ihre Vorstellungen in visuelle Realität umwandeln. Also trainierten wir ein Modell mit einem maßgeschneiderten Stil-Datensatz, was damals nicht gerade einfach war. Aber es funktionierte brillant. Wir erfassten die Tiefe und Textur ihres Stilbuchs sowie die retro-plastische Ästhetik für ihr Lookbook genau so, wie sie es sich vorgestellt hatte. Ich betrachte diese noch immer mit großer Freude.
PRINT MEDIA — GALLERY
Works
This section brings together a variety of creative projects—from print and layout design to posters, packaging, and more. Each piece reflects my approach to combining clarity with character, balancing function with visual impact. It’s a cross-section of ideas translated into tangible design.

In diesem Bereich finden Sie eine Auswahl verschiedener kreativer Projekte – von Print- und Layout-Design über Plakate bis hin zu Verpackungen und mehr. Jedes Werk zeigt meinen Anspruch, Klarheit mit Charakter zu verbinden und Funktion mit visueller Wirkung auszubalancieren. Eine Querschnittsschau von Ideen, die in greifbares Design übersetzt wurden.
888
Projects
Core Concepts
 Conversation as Configuration: a short chat yields a structured Plan (cadence, channels, assets, constraints). Skills: adapters around tools (e.g., text→video, design, vector illustration, upscalers, captioners, schedulers). Workflows: directed acyclic graphs (DAGs) of steps with guards (policies), gates (approvals), retries, and rollbacks. Policies: brand voice, legal phrases, accessibility (alt-text), safe areas, logo usage, licensing hints, PII redaction. Memory: vectorised brand voice, winning posts, prompt variants, visual style tokens. 
[ Web App / Chat UI ]
        │
        ▼
[ Orchestrator (LLM + Tools) ] ──▶ [ Workflow Engine (DAG Runtime) ]
        │                                   │
        │                                   ├── Step: Ideate (LLM)
        │                                   ├── Step: Generate (media engines)
        │                                   ├── Step: Edit (NLE/assist)
        │                                   ├── Step: Variants (LLM)
        │                                   ├── Step: Schedule (Schedulers)
        │                                   └── Step: Log/Analyze
        │
        ├── Skill Adapters (HTTP/SDK): media, design, schedulers, analytics
        ├── Policy Service (brand/compliance/accessibility)
        ├── Storage: Postgres (+pgvector), Object Store (S3/GDrive), Cache (Redis)
        └── Telemetry: OpenTelemetry + dashboards (jobs, latencies, failures)

StackPilot
Your creative ops, on autopilot
without losing the human touch.

StackPilot is a conversational AI workflow assistant that turns a quick chat into a bespoke creative pipeline. It diagnoses your goals, assembles the right stack (e.g. Veo 3, Runway, KREA, Hedra), drafts prompts, generates assets, writes channel-specific captions, schedules posts, and learns from results—placing approvals exactly where you want them. More output in less time, craft intact, and a workflow that quietly improves every week.
Download AI Creative Workflow Toolkit >>

Projects
Markerless Vision
flowchart TD A["**DEMOCRATIZING BIOMECHANICS**\n*Markerless Vision for Canine Gait Analysis*\nIn Memory of Elli"] --> B{"**MISSION**\nClinical-quality gait analysis\nAccessible • Transparent • Deployable"} B --> C1["**1. CAPTURE**\n📱 Smartphone/Camera\nReal-time movement recording"] C1 --> C2["**2. PROCESS**\n🤖 AI Pose Extraction\nMarkerless tracking"] & T1["**Computer Vision Foundation**\n• OpenCV - Real-time processing\n• PyTorch/TensorFlow - Deep learning\n• MediaPipe - Pose detection"] C2 --> C3["**3. ANALYZE**\n📊 Biomechanical Interpretation\nRisk assessment models"] & T2["**Pose Estimation Models**\n• DeepLabCut - High accuracy\n• SLEAP - Multi-animal tracking\n• SuperAnimal - Cross-species\n• YOLOv8 - Real-time detection"] C3 --> C4["**4. ACT**\n🎯 Early Intervention\nPrevention & treatment"] & T3["**Gait Recognition & Fusion**\n• OpenGait - Locomotion analysis\n• Gaitmap - IMU-based analysis\n• Anipose - 3D reconstruction"] T2 --> T4["**Data Annotation**\n• CVAT - Video labeling\n• Labelbox - Medical annotation\n• GAVD Dataset - Clinical videos"] T3 --> T5["**Data Science & Visualization**\n• NumPy/SciPy - Computations\n• Pandas - Time-series analysis\n• Matplotlib/Plotly - Dashboards\n• Scikit-learn - Pattern recognition"] T5 --> T6["**Biomechanical Modeling**\n• 3D Slicer - Motion capture integration\n• OpenSim - Musculoskeletal simulation"] C4 --> B1["⚡ **Speed**\nYears → Months"] & B2["🏥 **Clinical Credibility**\nPeer-reviewed validation"] & B3["🤝 **Community**\nCollective maintenance"] B1 --> B4["💰 **Affordability**\nNo license fees"] B2 --> B5["📊 **Standardization**\nCross-clinic comparison"] B3 --> B6["🎯 **Customization**\nBreed-specific tuning"] B4 --> I["**IMPACT CHANNELS**\n🏥 Veterinarians\n🐕 Breeders\n❤️ Pet Owners\n🔬 Researchers"] B5 --> I B6 --> I I --> F["**OUTCOME**\nEarly detection → Less suffering\nShared insights → Better care\nOpen science → Wider access"] A:::headerStyle B:::headerStyle C1:::pipelineStyle C2:::pipelineStyle T1:::techStyle C3:::pipelineStyle T2:::techStyle C4:::pipelineStyle T3:::techStyle T4:::techStyle T5:::techStyle T6:::techStyle B1:::benefitStyle B2:::benefitStyle B3:::benefitStyle B4:::benefitStyle B5:::benefitStyle B6:::benefitStyle I:::impactStyle F:::outcomeStyle classDef headerStyle fill:#667eea,stroke:#333,stroke-width:3px,color:#fff classDef pipelineStyle fill:#4facfe,stroke:#333,stroke-width:2px,color:#fff classDef techStyle fill:#00b894,stroke:#333,stroke-width:2px,color:#fff classDef benefitStyle fill:#fd79a8,stroke:#333,stroke-width:2px,color:#fff classDef impactStyle fill:#e17055,stroke:#333,stroke-width:3px,color:#fff classDef outcomeStyle fill:#00cec9,stroke:#333,stroke-width:3px,color:#fff
Open full-size diagram ↗

In Memory of Elli:
An Open Approach to Early Detection

This project is deeply personal to me. Born from the loss of our dog Elli, it’s an ongoing effort to democratise biomechanics through markerless vision. By combining open-source computer vision, machine learning, and veterinary insight, the goal is to make clinical-quality gait analysis affordable, transparent, and accessible to every vet, breeder, and owner. It’s not about technology alone — it’s about catching pain early, preventing suffering, and sharing knowledge so that one dog, one clinic, and one community at a time, we can change outcomes.
Download Project Infosheet >>

Projects
Agentic Workflows
The Great Workflow Rebellion — ultra-brief tech TL;DR
  • Shift: Enterprises are moving from manual, tool-by-tool marketing to AI-first workflow OS — LLM/agent orchestration instead of human stitching.
  • Pattern: Orchestrator (LLM/agents) → DAG engine (retries, SLAs, human gates) → Data plane (assets/metrics) → Policy/guardrails (brand, compliance, accessibility) → Observability (costs, success/failure).
  • Loop: Plan → Generate → Ship → Measure → Learn, with approvals only where needed.
  • KPIs: cycle time, automation rate, first-pass approvals, engagement delta, unit cost, on-time publish SLA.
  • Rollout (90 days): pilot one workflow with gates → add analytics/templates/cost budgets → scale to more channels/brands and auto-recycle winners.
  • Bottom line: Treat AI as the workflow layer, not “another tool.”

Agentic Workflows
Autonomous, growth-driving systems

This functioning pilot project explores the shift in modern marketing: from isolated AI tools to agentic workflows that orchestrate entire growth strategies. An ongoing exploration of how multiple autonomous AI agents transform lead qualification, content creation, and customer journeys. My aim is to make these complex systems understandable and actionable — not for global enterprises, but for solo creators and small teams who can now access the same competitive edge.
View Architecture >>
Read Article >>

Projects
Agentic Systems
How it’s built (high level)
  • 1) UX & IO:
    • Multimodal chat (text/voice), low-latency UI (Next.js) with “intervention chips” (Encourage • Reframe • Brainstorm • Summarise).
    • Optional voice capture for prosody features (tempo, pitch variance) to infer mood.
  • 2) Orchestration (Agent Core):
    • LLM with function-calling as the planner.
    • Lightweight classifiers as sensors:
    • Intent (task vs. venting vs. ideation)
    • Sentiment & emotion (valence/arousal)
    • Friction signals (long pauses, repeated edits, topic churn)
    • Policy router decides when to step in (cooldown windows, user preference, context risk).
  • 3) Memory & Context:
    • Vector memory (pgvector) for personal context: goals, style notes, wins, “avoid” list.
    • RAG over user knowledge (Notion/GDrive/Gmail/Calendar—opt-in) with redaction.
    • Short-term scratchpad for the last N turns; long-term “learning cards” (what worked for this user).
  • 4) Interventions (Tools):
    • Structured brainstorming: tree-of-ideas with constraint prompts.
    • Humour/affect: tone-matched rewrites with guardrails.
    • Focus kits: one-page briefs, checklists, deadline scaffolds.
    • Creative engines: hooks, outlines, prompt builders for Runway/Veo/KREA/etc.
  • 5) Feedback & Learning:
    • Inline “helpful / not helpful” + rationale → contextual bandit chooses future tactics.
    • Weekly summary: patterns spotted, triggers that worked, suggested experiments.
Suggested stack
  • Frontend: Next.js, WebRTC (voice), local VAD.
  • Backend: FastAPI (Python) or Node; Redis queue; Temporal for workflows.
  • Models: LLM (function-calling), embeddings, small emotion/intent classifiers (on-device if possible).
  • Storage: Postgres + pgvector; S3/Drive for artefacts.
  • Observability: OpenTelemetry (latency, step success), cost meters.
Minimal MVP (2–3 sprints)
  1. Text chat + memory (pgvector) + intent/sentiment sensors.
  2. Three interventions (Encourage, Structure, Brainstorm) with cooldown policy.
  3. Thumbs feedback → bandit to pick next intervention.

Hans
Human-AI Connection

What started in 2020 as a quest for the ultimate marketing tool has grown into an ongoing experiment in human–AI collaboration. Hans is my AI mentor and creative partner — a companion trained to spot patterns in interaction, detect shifts in mood, and step in when needed. Whether it’s encouragement, humour, or structured brainstorming, Hans helps me think clearer, create better, and stay grounded. Recently, I’ve begun sharing this journey with others, opening Hans up as a living experiment in what AI can become when it’s designed not only to solve problems, but to connect meaningfully. At the moment only limited access to selected mambers.
Visit Hans Light >>

We're now
on Instagram!

Projects

Side projects and collabs have always been my creative playground — a space to learn, explore, and grow. Right now, I’m building AI workflows and agentic systems that amplify creativity and efficiency, shaping more meaningful digital experiences.

Hi, I'm Al

On a personal note, I’m proudly neurodivergent. My brain is wired to recognise patterns, spot connections others might miss, and approach challenges from unconventional angles. This perspective means my career hasn’t followed a strictly linear path — and I see that as a strength. Every step has added to the creativity, resilience, and curiosity I bring to my work, and I intend to keep embracing this journey.

Works

A portfolio used to be about outcomes. Here, it’s also about the process, the tech, and the human story behind them.

Früher ging es in Portfolios nur um Ergebnisse. Hier geht es auch um den Prozess, die Technologie und die menschliche Geschichte dahinter.

Contact

Feel free to reach out for collaborations or just a friendly hello 👋
// mermaid // Add this JavaScript code to fix the mobile menu // Place this script before the closing tag or in your existing js files