knowledge architecture · energy transition

Research Lab

BWAboutWorkResearch LabOff DutyKnowledge Lab

Tracking the knowledge layer as it becomes infrastructure.

The Research Lab is where I track a transition I think is already underway:

Documentation is becoming knowledge infrastructure.
The titles haven't settled.The tooling categories haven't settled.The architecture hasn't settled.

But the signals are appearing everywhere.

I track those signals across hiring, tooling, organisational design, AI architecture and the energy transition, then publish the underlying data, methodology and working assumptions.

No black box. No analyst theatre.

Just sourced evidence, explicit uncertainty and research you can inspect.

LATEST EDITION · AUG 2026
Explore the research ↓GitHub ↗

01 / THE ALTITUDE MAP

Same function. Different name.

RESEARCH EDITION · KNOWLEDGE WORK · PUBLISHED · AUG 2026

What happens when you stop comparing job titles and start comparing the work underneath them?

Walmart calls it Principal, Data Architect.GitLab calls it UX Architect.Anthropic calls it Technical Documentation and Content Engineer.

Different companies. Different titles. Increasingly similar altitude.

The Altitude Map tracks documentation, content design and knowledge architecture from practitioner through Principal, Distinguished and Director-level work.

The emerging signal:

"Knowledge Architect" may not be a speculative future role. It may simply be the name that hasn't caught up with work already happening at scale.

The map currently tracks 24+ organisations across AI-native technology, developer platforms, European enterprise software and financial services.

THE SIGNAL
Titles fragment. Functions converge.

02 / THE KNOWLEDGE HIRING RADAR

Follow the hiring. Find the architecture.

DATASET / SIGNAL RESEARCH · KNOWLEDGE WORK · PUBLISHED · AUG 2026

57 sourced hiring signals · 4 linked research repositories

Job descriptions reveal things organisation charts often don't.

What capabilities are companies actually hiring for?Where does documentation end and Knowledge Architecture begin?Where are RAG, ontology, MCP, AI evaluation and knowledge engineering appearing inside roles that aren't called any of those things?

The Knowledge Hiring Radar tracks those signals across industries and maps them by organisational maturity and strategic direction.

A companion radar narrows the same question to the energy transition.

The point isn't to predict the next fashionable job title.

It's to watch where the underlying function is becoming necessary.

THE SIGNAL
The job title is metadata. The requirements reveal the architecture.

03 / THE STACK ATLAS

Who is building the knowledge stack?

RESEARCH EDITION · KNOWLEDGE WORK · PUBLISHED · AUG 2026

24 vendors.7 categories.8 capabilities.One unexpectedly consistent pattern.

I mapped the companies building docs-as-code platforms, enterprise knowledge bases, AI-native documentation systems, code-aware tooling, headless CMSs, localization infrastructure and API tooling.

Then I looked at who founded them.

Every identified founder across all 24 companies was an engineer.

Zero exceptions in the dataset.

Not one was a documentarian building the documentation system they had always wanted.

That doesn't prove engineers build bad documentation tools.

It raises a more interesting question:

What happens when an entire tooling category is predominantly designed by people adjacent to the job rather than people doing the job?

And then there's the blank quadrant.

Gartner now has a Magic Quadrant for Customer Service Knowledge Management Systems.

There still isn't an equivalent analyst category for the developer-facing documentation and knowledge infrastructure market.

The market exists.The category hasn't caught up.
THE SIGNAL
The tools exist. The category doesn't. Yet.

04 / SIX SURVEYS. ONE WORD.

Context.

SYNTHESIS RESEARCH · KNOWLEDGE WORK · PUBLISHED · AUG 2026

This may be the signal I'm most interested in.

I cross-referenced six independent practitioner datasets:

GitBookPostmanUX Content CollectiveState of the GraphZendeskIntercom

They come from different professional communities.

They aren't citing one another.

Yet all six independently converge on the same missing layer:

Context.

API teams need it.Content designers need it.Knowledge-graph practitioners model it.Customer-service platforms are trying to recover it.AI systems increasingly depend upon it.

The interesting finding isn't that six reports use the same word.

It's that six largely separate disciplines appear to be describing different symptoms of the same architectural problem.

THE SIGNAL
Independent convergence is stronger evidence than repetition.

05 / THE SIGNAL STACK

Not all evidence proves the same thing.

FRAMEWORK / ONTOLOGY · KNOWLEDGE WORK · ACTIVE · AUG 2026

One of the problems with trend research is that observations quickly become conclusions.

So I use a small classification ontology to separate them.

TITLE SIGNAL

What does an organisation call the work?

REQUIREMENT SIGNAL

What capability is it actually hiring for?

TOOLING SIGNAL

What infrastructure is being built around it?

SCALE SIGNAL

Where is the behaviour appearing repeatedly?

SYNTHESIS SIGNAL

What emerges when independent evidence begins converging?

The Signal Stack isn't a maturity model.

It's a way of asking:

What does this piece of evidence actually allow me to claim?

FIELDWORK / ENERGY

When physical infrastructure becomes software-defined, knowledge becomes part of the infrastructure.

My professional work sits inside the energy transition, so I'm applying the same research method there.

VPPsDERsEMSsGrid connectionsFlexibility marketsAPIsRegulation

Different systems increasingly need to exchange not merely data, but meaning and context.

The Grid Knowledge Graph is an early worked example: mapping the fragmented regulatory and organisational relationships around the UK grid-connection ecosystem as connected knowledge rather than another collection of documents.

THE QUESTION
What knowledge architecture does an increasingly distributed, software-defined and agentic energy system require underneath it?

METHOD

Public evidence. Public methodology.

This isn't Gartner.

I don't have vendor briefings, NDA datasets, private revenue disclosures or a pre-publication dispute process.

So I don't pretend I do.

The research uses publicly available evidence with explicit inclusion criteria, sourcing-confidence tiers and methodology notes.

Where a number is estimated, it's labelled.Where evidence is adjacent rather than direct, it's labelled.Where I personally use a tool being analysed, that's disclosed.

And where the data is wrong:

Open an issue or PR with a source.

The correction mechanism is part of the research.

SOURCE → CLASSIFY → CONNECT → TEST → PUBLISH → CORRECT

A working thesis.

Across hiring, tooling, AI and the energy transition, I keep encountering different versions of the same problem.

Organisations have spent decades creating information.

AI has made the structure underneath that information impossible to ignore.

The next question isn't simply:

How do we create better documentation?

It's:

What knowledge architecture allows humans and machines to share context reliably?

That's the signal I'm following.

Binoy Watts
Research Lab
Research · Data · Knowledge Graphs · Field Notes
← back home