INVESTMENTS · TECH ASSETS

Six technologies, two connected families

MediaTech invests in technologies, not companies. Our current portfolio is one body of work built on a single idea: take material that arrives in many separate streams, understand what each item is about, and present it in one governed, ordered place.

MediaTech does not acquire or invest in companies; it invests only in technologies (tech assets) and sells only technologies (tech assets).

PART A

Aggregated social feeds

Three capabilities that produce and monetize one merged, topic-categorized feed drawn from the reader's own social platforms and news providers.

See the three capabilities ↓
PART B

AI-assisted development

Three capabilities that raise the quality and lower the cost of producing software with AI assistance — generation, governance, and verification as one operation.

See the three capabilities ↓
PART A · AGGREGATED SOCIAL FEEDS

One feed instead of four applications

A reader connects the accounts they already use — Facebook, Instagram, YouTube and X, plus any news providers. Items are collected, normalized into one card format, de-duplicated, and ordered into a single timeline. Every card keeps its source badge.

A1

Merged, topic-categorized feed & social magazines

Every item is assigned topics from a published list — football, fashion, news, technology — so one feed can be cut by subject. Any followed topic can be published as a social magazine: a periodic issue with a cover, a running order, and a fixed size, built from every connected source.

Merge & de-duplicate Topic channels Magazine issues
A2

Profile-driven feed generation

Feeds assemble themselves from a reader profile: declared attributes held as ranges, stated preferences that always win, and observed signals — scroll, dwell, read and view time — converted into per-topic interest weights the reader can see, reset, or switch off entirely.

Reader profile Interest weighting Automatic generation
A3

Content-to-advertising pairing

The same topic labels that organize the feed place advertising in it. Campaigns are matched by topic, interest weight, and declared attributes; every advert is labeled as sponsored, never changes editorial ranking, and reporting is aggregate only — no individual reader is ever disclosed.

Topic matching Placement rules Aggregate reporting
FLOW
Sources mergedTopics assignedMagazines publishedProfile builtFeed generatedAdvertising paired
PART B · AI-ASSISTED DEVELOPMENT

Generation, detection, correction and verification as one operation

Three separable capabilities that compound: fragmentation raises the quality of what is generated, the joined detection loop catches what remains the moment it is written, and governance makes both repeatable across every developer and team.

B1

Vibe-coding joined to bug detection & fixing

The developer describes a change once, in plain language. The code that comes back has already been scanned, corrected and verified — describe, generate, detect, fix, verify, accept, in one continuous loop. Nothing is merged without explicit developer acceptance.

One instruction Faults fixed in-context Verified before review
B2

Prompt fragmentation framework

Long instructions are served unevenly — early requirements honored, later ones lost. The framework splits a long prompt into ordered, individually executable steps, each checked against its own requirement before the next begins. Failure is localized; only the failing step repeats.

Ordered steps Checked per step Reusable fragments
B3

Prompt library & prompt governance

Proven prompts are collected, versioned and owned. Every prompt is held against a versioned governance rulebook — naming, approved libraries, secrets, personal data, licensing — and every use is checked: passed and logged, corrected with the correction shown, or blocked and referred. Nothing passes silently.

Versioned prompts Rulebooks Full traceability
WHY THESE SIX

Each capability stands alone — and strengthens the others

In Part A, topics are the unit everything is built on: magazines are topics on a schedule, interest is weighted per topic, and advertising is paired to topics. In Part B, defects reaching later stages fall and effort per change is redistributed — less rework, less waiting, less variation, less re-explaining. And judgement stays human: nothing merges without a person accepting it.

Readers get one governed feed instead of four applications — with the profile fully visible, resettable, and deletable

Advertisers buy topics, not people — reporting is aggregate and no individual reader is ever disclosed

Developers ship faster because faults are corrected the minute they are written, not after review or release

Every generated result records the prompt, prompt version and rulebook version behind it — governance by design

Interested in these technologies?

Also see: The AI thesis · Approach
Get in touch