I build and operate agentic AI systems.
I help businesses turn knowledge-heavy work into AI-assisted products and workflows. I can build the operation from the beginning, run and improve an existing system, work alongside an internal team, or prepare the system for handover.
Typical deliverables include research workflows, publishing systems, specialist knowledge products, document operations, and customer-facing applications. My work covers the surrounding software and operating process as well as model calls: monitoring, failure recovery, evaluation, cost control, change management, and human approval.
Map the work, choose the simplest useful architecture, build the agents and supporting software, connect data and tools, and prepare the workflow for real users.
Monitor runs, manage model and tool changes, investigate failures, control cost, maintain quality, and keep human review and release work moving.
Use logs, failures, review results, and cost data to improve the workflow. For handover, document the system and train its owners.
I publish case studies and engineering notes when the material can be shared. They include failures, costs, evaluation methods, and implementation decisions. Confidential client material stays private, and experiments are labelled with their limitations.
Projects that fit this service.
I work with founders, specialists, and teams that have domain knowledge and an existing workflow to improve. Typical projects include research and analysis, editorial and publishing systems, specialist knowledge products, document-heavy operations, internal tools, and customer applications grounded in controlled source material.
The service is useful when work crosses several tools, depends on changing sources, contains repeated handoffs, requires human approval, handles sensitive context, or needs an audit trail. These requirements usually need software and operating controls around the model.
I recommend standard automation, search, or a single model call when it can meet the acceptance criteria. The architecture should remain as small as the work allows.
How a project runs.
- Scope the problem. We define the users, current process, desired output, source of truth, constraints, and measurable acceptance criteria.
- Design the workflow. I map inputs, tools, permissions, model calls, human review points, failure modes, privacy requirements, and the cost of a wrong result.
- Build a working version. I implement the smallest complete system using the simplest suitable mix of agents, deterministic software, APIs, data stores, and interfaces.
- Test it on representative work. We record completeness, quality, speed, cost, safety, recovery, and usability across real cases.
- Operate, improve, or hand over. I can run and monitor the system as a managed operation, work alongside your team, or train an internal owner and transfer it with clear documentation and controls.
A project can end after setup or continue into managed operation. The scope states how results are judged, which decisions stay human, who owns each action, and what the team receives at handover.
Private inputs and project details stay outside public case studies unless publication is explicitly agreed.
219 source notes, one book, and fourteen localized editions.
The client problem. A client brought 219 notes collected while building a house. They needed those fragmented observations to become a coherent, illustrated book for an analytical first-time home builder. Every requested comparison, checklist, example, visual, and personal experience had to survive the edit and remain traceable to its source.
What I built. I designed a content system around stable source IDs, structured content cards, chapter briefs, model-assisted drafting, fact and completeness checks, image tracking, and PDF inspection. When the brief expanded, I added resumable localization, separate native-language copy-editing, language-specific storefront generation, Stripe Checkout, checksums, and immutable edition records.
The production result. The system produced a 227-page Croatian source edition, fourteen localized books and storefronts, and zero unmapped notes at release. New notes can be processed without rebuilding unchanged work, and purchase-time edition metadata remains available for audit while buyers receive the active book for their language.
Stable IDs carried every author note through chapter planning, manuscript, audit, and release.
Source hashes, checkpoints, retries, and review gates kept long-running language work controlled.
The system made omissions visible before fluent prose or a polished PDF could hide them.
The controls make omissions visible, limit what a model may change, preserve human editorial review, recover interrupted jobs, record cost, and give each release a verifiable history.
Read the full Build Your Home case study for the architecture, model benchmark, controls, failures, costs, and release logic, or inspect the live product at buildyourho.me/.
Two products I built and shipped.
These are self-directed products. Cartoonio coordinates a multi-stage generative media workflow; Symmetry Studio performs image processing locally in the browser.
What it does: turns a written story into reusable characters, planned scenes, generated voices, motion, and a finished animated video. What it demonstrates: orchestration across several generative-media stages while maintaining a consistent project and editable result. Use it at cartoonio.net.
What it does: turns a source image into repeatable designs across all seventeen wallpaper symmetry groups. What it demonstrates: responsive worker-based image processing that keeps source images on the user’s device. Open it at operator0x.com/symmetry/.
Engineering notes and experiments.
The articles document workflow structure, model and architecture choices, failures, cost, evaluation, and release controls. Each one states whether it covers client work, a live product, or an experiment.
PolyMind is a local prototype that sends one brief to two configured AI perspectives and then synthesizes their responses. The published run cost $0.09 and produced useful criticism of this website. It lacks the baseline and human ratings required to compare the method with a single agent.
Loopy & Clippy is an unreleased local video editor for arranging scenes, camera moves, timed text, sound, and FFmpeg renders. Its preview and export still diverge on some projects, so it remains an internal experiment.
New case studies and engineering notes are listed at operator0x.com/feed.xml.
Send me the workflow you want to improve.
Explain how the work is handled today, who uses or approves the result, and any data, privacy, budget, or timing constraints. A model or framework choice is unnecessary at this stage.
I will reply with a proposed starting scope or explain why standard automation, search, or a smaller model workflow would be sufficient.
Send the brief to contact@digital-distributed.hr.
Start with the workflow, its users, and the acceptance criteria.
OPERATOR_0x / AGENTIC AI SERVICES GUIDE / 2026
FREE TO READ, COPY, AND SHARE WITH CREDIT.
SERVICES, CASE STUDIES, AND LEARNINGS AT OPERATOR0X.COM.