Context Switch

Practical AI systems, open research, and useful build work.

Context Switch is the build-house layer inside Mark Hay: a place for practical systems, public-good research, and product experiments that turn messy information into useful, reviewable outputs.

The hierarchy stays deliberate: Mark Hay is the public identity, Context Switch is the build-house label, and each product or research output keeps its own name and purpose.

Read the research track Browse selected builds

A build-house layer by Mark Hay

Current focus

Reducing admin burden with practical AI

Cross-industry evidence-to-action work covering small-business automation patterns, AI governance for non-technical teams, and reusable implementation templates.

Why it exists

A build-house layer for useful systems work

Context Switch gives practical AI, automation, and research outputs a coherent home without turning the whole site into an AI consultancy brochure.

The role is simple: build useful things, document the judgement behind them, and publish reusable methods where sharing creates more value than hiding the work. It supports the broader markhay.net trust layer rather than replacing it.

Proof tracks

Three lanes, one operating thesis

Context Switch groups selected work by what it proves, not by how shiny it looks.

Public-good research

Free evidence-to-action packs that turn source-backed research into practical methods, templates, and implementation guidance.

Open track →

Practical systems

Automation patterns, admin-burden reduction, and workflow tools for teams that need usefulness before theatre.

Open track →

Selected builds

Focused products, utilities, and experiments that show the product-minded lane without becoming a second portfolio index.

Open track →

Operating principles

No fake authority. No novelty theatre.

The work should be useful, reviewable, and honest about its limits.

Useful before impressive
Evidence before confident claims
Human review where decisions matter
Open outputs where public value is stronger than secrecy
Product names stay primary; Context Switch stays the build-house layer
No fake authority, especially in specialist or regulated domains

Live AI data proofs

Proof that turns into a useful next step

Each asset has a first-party route, a buyer question, a conversion path, and a visible limit. The point is commercial clarity, not vague AI theatre.

Public proof feed

A small live registry of buyer-facing proof

The cards below are backed by a no-store JSON feed. That keeps the visible page, analytics events, dashboards, and future agent checks aligned around the same proof contract rather than treating the page as copy that drifts.

Source-backed research

Admin-burden evidence-to-action pack

Interactive diagnostic

Conversion Critic public-page scan

Adoption workflow

AI Support workflow layer

Assets: 3
View event: proof_asset_viewed
CTA event: proof_asset_cta_clicked
Feed: Open JSON

Live feed check

Checking

Generated: Awaiting timestamp
Feed id: Awaiting feed
Source route: /context-switch
Feed assets: 3 expected
Cache policy: no-store

Proof assetCredibility check

Admin-burden evidence-to-action pack

Turns source-backed admin-burden research into reusable implementation guidance for practical AI decisions.

Buyer question

Can this work turn research into a practical decision aid rather than generic AI advice?

I do not want another AI opinion piece; I need a route from evidence to a usable operating decision.

Live signal: Public research route with evidence, methods, and next-step templates.
Conversion path: Research CTA shows demand for evidence-led practical AI guidance.
Next validation: Track whether research readers continue to a practical tool, selected build, or contact route.

Visible limit

Research guidance only; implementation choices still need local process review and human sign-off.

Open research pack

Tool trialPre-discovery scan

Conversion Critic public-page scan

Uses live public-page extraction to turn visible copy into buyer-friction findings and rewrite priorities.

Buyer question

Can a live public page be turned into specific buyer-friction findings without a long discovery call?

I am not ready to book a review until I can see whether the scan produces concrete, relevant findings.

Live signal: Working route fetches a public URL and analyses extracted page copy.
Conversion path: Tool CTA shows whether visitors want a first-pass scan before a manual review.
Next validation: Compare scan starts with manual review or contact intent from visitors who used the tool first.

Visible limit

Automated first pass; paid reviews remain manual where commercial judgement matters.

Try page scan

Workshop pathGuided adoption

AI Support workflow layer

Shows assistant selection, reusable instructions, critique, and vault-style workflow support in one public surface.

Buyer question

Can practical AI support be packaged around workflow habits instead of tool-name comparison?

I need help adopting AI safely in real work, not a generic comparison of ChatGPT, Copilot, Claude, and Gemini.

Live signal: Live guidance routes for Copilot, ChatGPT, Claude, and Gemini adoption patterns.
Conversion path: Support CTA shows interest in moving from general AI curiosity to guided adoption.
Next validation: Measure whether assistant-specific guide views lead to workshop, support, or contact intent.

Visible limit

Training and governance support; it does not replace private policy, security, or legal review.

View support workflow

Selected outputs

A curated proof strip, not a second project catalogue

These examples show the type of work that fits the build-house thesis. The full browse surface remains Projects.

Public research

Evidence-to-action research

A public-good research programme focused on reducing admin burden, small-business automation patterns, and AI governance for non-technical teams.

View output →

Live product

UK Shortlists

A live buying-guide product with category architecture, structured shortlist logic, and disciplined publishing automation.

View output →

Product concept

AI Support

A practical support layer for Copilot, ChatGPT, Claude and Gemini adoption, prompt patterns, critique, and reusable instruction building.

View output →

Tool suite

E-Commerce Utility Calculators

A compact calculator suite that turns messy commercial arithmetic into fast decision support.

View output →

Hierarchy

Clear ownership, no brand clutter

Mark Hay

Public identity, trust layer, and main point of contact across the wider body of work.

Context Switch

Build-house layer for practical AI systems, open research, and selected systems work.

Products and research outputs

The actual proof: named tools, case studies, templates, reports, and public outputs.

Start with the public research track

The first programme focuses on reducing admin burden with practical AI, automation patterns, and governance that non-technical teams can actually use.

Open research track