From scanned manuals to a working restoration

Restoring an Otari MX-5050 reel-to-reel recorder required turning specialist documentation into practical guidance. The website had to make unfamiliar documentation easy to find, understand and apply to a physical machine.

With limited formal training and study in electrical engineering, I used the site and the AI-assisted research behind it to complete a full recap and calibration without an expert guiding me through the work. Building the resource and learning the subject reinforced each other: organising a manual raised questions, researching a circuit improved the explanation, and the bench work gave that learning a concrete use.

The public result is otarimx5050.com, a technical reference and learning resource. It combines source documents, navigable digital editions and interactive explanations with a record of the restoration.

The knowledge problem

A scanned manual contains valuable information, but using it requires more than locating a PDF. Readers need to move between sections, inspect diagrams, understand unfamiliar terms and check an explanation against the original source.

The project therefore had two connected outcomes: build a resource that made the material usable, and use it to support the restoration. That kept software development tied to a real task.

What we built

Navigable digital editions. Manuals and brochures are organised into sections that readers can open directly. A reusable edition structure supports further documents without designing a separate publishing system for each one. Readers can browse manuals and brochures in the document library.

Diagrams with source comparisons. Where a vector redraw and original scan are available, the image viewer lets readers zoom, pan and switch between them. Readers can check the clearer drawing against the original scan.

Visible editorial annotations. Source notes distinguish what was printed from a correction or explanation. That matters when a source contains an error: silently changing it would blur the boundary between the historical document and the new editorial work.

Interactive learning tools. The Inside the Audio page includes a signal-path explorer and a capacitor demonstration. Readers can examine stages of the audio path and vary controls to explore an explanation. Site search helps them find relevant material across the resource.

These features turn a collection of documents into something a reader can navigate, question and use during a task.

How AI and human judgement worked together

AI assisted with research, explanations, content structuring and implementation. Claude and GPT models were used across our development work, with Superpowers supporting specifications, plans, bounded tasks and review.

None of the Otari site’s copy was edited directly in an editor. I reviewed the output and sent prompts back for specific revisions; the LLM applied the changes. Human editorial judgement operated through those decisions and revision requests.

Preserving a digital edition and developing a technical explanation are different editorial jobs. The first must retain the source’s meaning and wording. The second must help a reader understand a concept and needs checking against suitable evidence. Treating both as generic text generation would miss that distinction.

The physical work remained mine. The LLM helped develop and organise the knowledge; I made the practical decisions, worked on the machine and performed the calibration.

From documentation to a working restoration

Manuals, diagrams and interactive explanations gave me a way to move from unfamiliar material to practical decisions at the bench. I completed the recap and calibration while building a resource that makes the same material easier for others to use.

Original scans and source comparisons remain available beside the explanations, so readers can check the details relevant to their own machine.

The result connects the software to a practical outcome: better access to specialist knowledge helped me carry out work that was new to me.

Where this applies in an enterprise

The corresponding enterprise opportunity is to make specialist knowledge easier to find, understand and apply. Maintenance documentation, technical onboarding and engineering reference libraries often combine valuable source material with a demanding learning curve.

A first project could take one bounded collection and turn it into a task-oriented resource, with references back to the originals and a defined review process. The domain owner would set the acceptance criteria; the delivery work would make the knowledge accessible in the form its users need.

The Otari project demonstrates the connection between software, learning and a practical outcome in a new technical domain. Our enterprise AI delivery approach explains how we would establish the controls and verification for an organisational setting.