Edward Frank Morris

I’m Edward Morris, and no. I don’t have a Computer Science degree. I never did a stint at Google. And I certainly wasn’t blessed with seed funding from some Stanford dorm room. What I do have, though, is a journalist’s nose for nonsense, a knack for systems thinking, and a mildly unhealthy obsession with Prompt Engineering.

In the last 18 months, I’ve somehow ended up:

- Featured on the front page of the Financial Times
- Blasted 80 feet high in Times Square (NASDAQ screen, not just loitering)
- Published almost monthly in Forbes with a penchant for calling out AI waffle
- Top 10 UK AI Influencer (Favikon, 2025)
- LinkedIn Top Voice for Prompt Engineering. The platform’s first.

And recently, featured in Gulf News, because apparently even the Middle East wanted in!

I've also had the pleasure of collaborating with some incredible brands and AI tools, such as:

- Tencent
- Notion
- Lovable
- Attio
- Superblocks

And many many others.

I’ve worked with teams inside OpenAI, Microsoft, Mercedes-Benz, and Google, and consulted on everything from legal AI Agents to prompt defence systems. I’ve built tools that save thousands of hours across Pharma, HR, law, education, and charities. Because when AI is done right, it's not about replacing people. It’s about freeing them to do better work.

I've also held live sessions with Taras Shevchenko University in Ukraine, alongside lectures at Oxford, University of Santo Tomas in Chile, and the American University in the Emirates. I’ve judged AI competitions, advised Boards, spoken at major conferences, and when time allows, still find joy in helping not-for-profits discover what AI can do without needing a Silicon Valley translator.

But none of this makes much sense without where it started. I grew up in extreme poverty in Cagayan de Oro, Philippines. Dodging typhoons and malnutrition. Later, it was a UK council flat, a carer’s allowance, and my mum’s disability income keeping the lights on.

No silver spoon. Just stubbornness, storytelling, and a refusal to accept the limits of where I began.

So if you're here looking for another polished tech bro with a hoodie and a pitch deck, you’ve taken a wrong turn. But if you're after someone who’s lived the stakes, learned the system, and now builds things that actually matter? We should probably talk.

Email me at Edward@Enigmatica.co.uk or just drop me a message.

Prompt Engineer & AI Evangelist

I create AI tools (and text games) for others to use using Prompt Engineer that works with ChatGPT, GPT 3.5, GPT 4, Claude, PaLM 2. Ranging from simple copywriting AI applications all the way to Prompt Injection AI Tools.

I was invited to an official United Nations meeting. 

Commemorating 80 years of the UN.

Which was definitely a "stop and take it in" moment for me.

And the room... was extraordinary.

I had the honour of being joined by:

→ Annalena Baerbock, President of the General Assembly
→ António Guterres, Secretary-General of the United Nations
→ Lord Hermer, Attorney General
→ HRH The Duchess of Edinburgh, Princess Sophie

Along with leaders, innovators, and changemakers who are genuinely trying to make the world better. Not in theory, but in practice.

We discussed peace.
We discussed conflict.
We discussed the future.
And yes. We discussed AI.

Upon reflecting, in this global race to build the most advanced AI systems we forget something crucial.

That AI’s biggest bottleneck isn’t the technology. It’s us.

Every organisation trying to adopt AI learns this quickly.

Upskilling. Adoption. Workflow change.

None of it is a technical issue first. It’s a human one.

Even when AI joins the creative process, it can only rise to the level of the person using it. The AI cannot tackle what it doesn't know, and where does it get that sense of direction? That's right. Us.

AI in the right hands creates opportunity.

It bridges gaps.
It lets talent shine that would otherwise be overlooked.
It frees people to do the work that matters while it handles the work that doesn’t.

There is so much untapped potential ahead.

And standing in that room, surrounded by people who shape nations and futures, I was reminded that AI isn’t here to replace humanity.

It’s here to elevate it.

What a privilege.

I build AI frameworks, lead strategy, and teach AI to anyone from Fortune 500s to universities. My face has been on NASDAQ, FT, and Forbes. My jokes have not. Yet.

No alternative text description for this image

There's a quiet revolution happening in who gets to build software.

In seven months, Emergent went from 0 to 5M+ builders and $50M ARR. After Google's AI Futures Fund, Khosla Ventures and SoftBank just led a $70M Series B
.
That's validation. But the real story is what builders are creating.

A non-technical founder built a construction procurement platform in 80 hours. Cost? Under $1,500. Result? Raised $150K and delivers 40% cost savings to users.

This isn't theoretical. People are shipping real products that generate revenue and solve real problems.

Emergent handles the full software lifecycle through AI: backend, frontend, database, authentication, payments, deployment. What you build is production-ready from day one.

The technical barrier that kept most people from building is dropping fast.

Worth exploring if you've got ideas sitting in notes that you've never acted on.

#emergent #emergentai

My first ever role in AI was Prompt Engineering.

It's still one of the heaviest parts of my offerings and portfolio.

I've worked with some of the largest companies in the world with that title and that responsibility. My Prompts went from simple such as "Write a LinkedIn post" all the way to Prompting into existence a synthetic personality that can perform actions on its own. Autonomously.

But I live under no illusions. 

Prompt Engineering and Context Engineering will eventually be things of the past. Probably within 2 or 3 years, depending on how rapidly AI develops.

Will you still need to command AI in a very specific way? Absolutely.
Will you still need to oversee AI Agents in a particular way? For sure.
Will businesses still require Automation? Yes.

This will likely not change.

What will change however is the depth and ease of which we do it. 

For instance, to get an Orchestration to work well with Microsoft Copilot, you need to Prompt meticulously and make sure it behaves correctly, especially if it works with other Agents that also have key tasks.

It'll likely be the case in the future that the AI Agent of the future will simply build the tool it needs. I demonstrated something similar to this in 2023 with a Multi-Chain of Thought project.

An AI that can summon/create any role or tool, with context that fits the existing context, and logic to fit around the end goal.

All while improving itself constantly and trimming the unnecessary. 

That's the future we are likely heading towards.

If not this year than next.

I should also say this. If you're upskilling currently in these AI areas? Don't stop. There's a huge shortage out there of competent AI users that don't regurgitate the same nonsense whatever some faceless influencer peddled on Instagram, TikTok, or even LinkedIn.

There are fantastic courses out there from Google, OpenAI, and others. As well as that, you live in the magic of being able to summon and create anything with just a few sentences.

The biggest risk you can take currently is not adapting.

The second biggest risk is thinking you don't have to.

Thoughts?

I once watched a senior exec at a massive company type this into Microsoft Copilot:

“generate 10 point report”

That’s it.

No topic. No context. No clue.

He stared at the screen like it owed him money. 

And when nothing happened, he looked shocked. As if he'd just asked a toaster to make him a steak and it had the audacity to only toast.

Meanwhile, over at a tiny charity with a budget smaller than a packet of crisps, they were using ChatGPT to rewrite grant applications, summarise legal docs, and auto-generate stakeholder emails. 

It was like watching someone drive a Lamborghini properly, while the bloke from before had just reversed a Prius into a pond.

Do I blame the Exec? No. He wasn't trained or upskilled properly.

Same AI. Completely different outcomes.

Now here’s the painful truth. Especially if you're a decision maker.

AI is not the problem. 

Humans are.

We keep repeating this mantra, “AI is just a tool”, "Garbage In, Garbage Out", as if that somehow helps. It doesn’t. 

A chainsaw is also a tool. But you wouldn’t hand one to someone who thinks health and safety is a type of salad.

I’ve spent years inside companies, even Governments, teaching people how to actually use this stuff. And what I’ve learned is this:

Most people use AI the same way your dad uses the TV remote. Mashing buttons, sighing heavily, and eventually just giving up.

It doesn’t matter how fancy your prompts are. Or how many frameworks you’ve memorised. Or whether your GPT has a custom name and a little hat.

If you don’t know what good looks like, you’re done.

Here’s what happens in the wild:
Bad user + bad AI = apocalyptic rubbish
Good user + bad AI = slightly more elegant rubbish
Bad user + good AI = still rubbish, but now in bold
Good user + great AI = finally, something usable

It’s not “garbage in, garbage out”.
It’s “clueless in, nonsense out, everyone confused”.

If you want AI to actually work in your company, do this:

- Get a clear brief from a user. Not “make it good”. Actual instructions.
- Let someone who knows what they’re doing write the prompt.
- Both look at the output. Be honest. Is it good? Or is it digital beige?
- Reiterate it. Fix it. Rinse and repeat.

AI isn’t magic. 

It’s not some all-knowing oracle in the cloud. Yet.

It’s a glorified autocomplete that panics under pressure and makes stuff up when it’s unsure. 

Basically, it’s like a teenager on their first day at work.

But in the right hands?

It’s like giving a flamethrower to a Michelin-starred chef. Dangerous, precise, and somehow everything comes out perfectly crispy.

And in the wrong hands?

Well... you get a 10-point report that just says “10 points”.

Oh and give this a share. You never know who might need this. ♻

And if you want more posts like this? Or just the humour? Drop me a follow.

Remember January?

That was back when you could still name every major AI model off the top of your head.

Now?

Since then, we’ve had:

1. DeepSeek R1
2. o3-mini
3. Claude Sonnet 3.7
4. Gemini 2.0 Flash
5. Grok 3
6. Gemini 2.5 Pro (Experimental)
7. GPT-4.1
8. o3
9. o4-mini
10. Gemini 2.5 Flash (Preview)
11. Claude Opus 4
12. Claude Sonnet 4
13. LLaMA 4
14. o3-pro
15. Gemini 2.5 Pro (Stable)
16. Gemini 2.5 Flash
17. Grok 4

...and GPT-5 is basically circling the runway.

That’s more foundation models than I have socks.
And I own a deeply worrying number of socks.

So what do you do when the landscape looks like a GPU fueled game of Whack-A-Mole?

Three things👇

1. You can’t track it all
Unless your job title is CAIO, Chief Prompt Engineer, Prophet & Wizard, you’ll fall behind.
So stop trying to know everything. Instead:
→ Pick your lane.
→ Focus on use case, not headline scores.
→ Build with what solves your problem today. Not what looks impressive on Hacker News.

2. Benchmarks age like milk
That “top model” from March?
Obsolete by Easter.
If you're building anything that matters:
→ Retest often
→ Bake flexibility into your stack
→ Don’t marry a model. Date around

3. Your edge isn’t knowledge. It’s adaptability
The game isn’t “who read the most whitepapers.”
It’s:
→ Who experiments fastest
→ Who ships faster than the landscape shifts
→ Who learns in public and adjusts without ego

This is both evolution and whiplash. All at once.

And the winners?

Are fast and nimble.

If you’re still operating like it’s 2023...

You’re already behind.

♻️ Share this with someone still quoting ChatGPT-3 like it’s cutting edge.

And for more like this, follow Edward Frank Morris right here.

I built a real SaaS business in about 27 minutes.

Which is annoying, because it normally takes months and several emotional breakdowns.
I had an idea.

A simple subscription tool for freelancers.
In the old world, this would involve:

- Meetings
- Developers saying “it’s almost ready”
- Stripe setup nightmares
- And a launch date that slowly drifts into the next decade

Instead, I used Atoms

Atoms isn’t a chatbot.
It’s a Vibe Business Team in a browser.

You type an idea.

It builds the business.

First, DeepResearch.
It analysed competitors, pricing, and demand.
Proper thinking. None of this “here are 10 ideas” nonsense.

Then I switched on Atoms Backend.
Auth. Database. Stripe. Deployment.
One toggle. Done.

After that, Race Mode.
Multiple AI teams built the product at the same time.
I picked the best one. Like Top Gear, but instead of cars, it’s backend logic.

Finally, the SEO Agent.
Because a business that can’t be found isn’t a business. It’s a hobby.

End result?

Idea → Live app → Stripe connected → Revenue-ready

No dev team.
No configuration.
No begging anyone to “just push it to prod.”

You don’t code with Atoms.
You decide. It does the rest.

Try it before your competitors do:

https://atoms.dev
#VibeBusinessEra #BuiltwithAtoms

AI Slop is a boring term.

It's overused. Trite. More robotic than actual AI usage.

Yes, it's controversial but no one else apart from maybe one person I'm following wants to say it, so I'll say it.

Time and time again I see videos similar to the below that are genuinely beautiful. A watched is moved by it, feels for it, and can see the beauty in it. 

But the moment you say "This was animated by AI", suddenly it's as if all the things they felt previously were entirely false. 

They show disgust. Contempt. Perhaps even a bit of anger.

But you felt what you felt. You know what you felt.

So why believe the lie?

Why not simply accept that yes, there are things that are poorly made with AI.

But there are also wonderful things. Beautiful things.

Not just limited to art but so much more.

Switching LLM providers is easy. Keeping your gateway from breaking down under real production load is the actual problem.

I've seen LiteLLM work for teams at smaller scales, but as workloads increase, the Python runtime starts showing its limits: P99 latency hitting 90+ seconds at ~500 RPS and memory usage climbing over time. Bifrost, written ground up in Go, with just 11μs overhead at 5,000 RPS is 50x faster than LiteLLM (link to benchmarks in comments)

Bifrost takes a different approach:

🔹 Cluster mode for high availability deployment with automatic failover and load balancing
🔹Adaptive load balancing to dynamically distribute traffic based on latency, error rate, TPM limits, and fatal errors (5xx)
🔹Extensible with your own custom plugins 
🔹 Code Mode that cuts tokens by 50%+ and latency by 40-50% 
🔹 MCP Gateway with security-first design and stateless execution 
🔹 Built-in governance with virtual keys, budget controls, and SSO 
🔹 Prometheus metrics and OpenTelemetry built in, not bolted on

Migration from LiteLLM is a single parameter change:

Point your SDK to Bifrost's URL and you're done. Everything else including your model names, message structures, provider setup; stays exactly the same.

Same SDK. Same providers. You get semantic caching, proper observability, and tool execution with explicit approval flows.

If you’re looking to scale your AI agents reliably, check out Bifrost: https://lnkd.in/dHzH2M4d

What most people are missing about Ant Group’s health AI assistant – AQ
It’s not just healthcare innovation. It’s a glimpse of the AI leapfrog effect.

Some sectors that were slow to digitise aren’t simply catching up and instead they’re skipping legacy systems entirely.

Healthcare is a defining example.

Everywhere, AI is becoming part of healthcare’s support system. OpenAI introduced ChatGPT Health, and Anthropic is expanding its focus on healthcare and life sciences. AQ’s difference lies in focus, moving care upstream into prevention and everyday health management.

OpenAI has introduced ChatGPT Health, and Anthropic is expanding into healthcare and life sciences.

But China’s experience with AQ shows how quickly AI can move from experimentation to real application when a large, motivated user base and supportive policy align.

Today with AQ:
- 30 million monthly active users
- 5,000 connected hospitals
- 500+ top Chinese physicians with AI digital twins

AQ isn’t replacing doctors and it isn’t diagnosing. It’s becoming the first layer of care in a preventive sense. Helping individuals recognize issues early, interpret symptoms responsibly, and connect with the right medical channels faster.

The impact is structural:
- Low cost or free AI consultations lower financial barriers and encourage earlier care‑seeking.
- Doctors offload repetitive tasks, reducing burnout and focusing on complex cases.
- Patients receive accessible, contextual health guidance that improves self‑care and relieves pressure on hospitals.

So while it looks like efficiency on the surface, it’s really about access, empowerment, and scale.

It shows what happens when intelligence becomes part of public infrastructure. Shaping how people engage with healthcare from the ground up.

The transformation isn’t about AI replacing medical judgment. 

It’s about how preventive, AI enabled care can redefine participation in health itself.

A recent survey shows that >70% of Chinese doctors would recommend "AI doctor services" for everyday health questions and routine care, signaling strong professional support for AI-assisted healthcare.

Among these tools, Ant Group’s AQ was mentioned most frequently by physicians. It now serves 30M monthly active users and handles 10M+ health queries each day, making it China’s leading AI health platform.

Doctors value these services for their ability to handle large volumes of patient inquiries without time or location constraints, drawing on extensive medical knowledge to provide patient, reassuring responses, particularly for routine concerns.

graphical user interface, text, application, chat or text message

A lot of people will lose their jobs soon.

Because when the CEO of McKinsey says he now counts AI agents as “people” the firm employs you know the ground is already shifting.

Right now the firm has:

→ 40,000 humans
→ 20,000 AI agents

And their stated goal?

Parity in 18 months.

This is a hiring strategy. Unfolding in real time.

And here’s the uncomfortable truth that most of us already know.

Companies won’t replace people with AI because it’s futuristic. They’ll replace them because it’s efficient, predictable, and infinitely scalable.

But here’s what won’t be replaced:

→ Critical thinking
→ Leadership
→ Creativity
→ Judgment
→ The ability to turn insight into action

AI may take tasks. But people who know how to direct, challenge, and leverage AI? They’re about to become the most valuable talent in the market.

The question isn’t
“Will AI take my job?”

It’s
“Will someone who knows how to use AI take my job?”

What do you think? Is this progress, or a warning sign?

So you’ve had an app idea.

(And this time… you actually built it.)

Mine’s called: Do it or pay £5.

It’s a productivity app with teeth.

Set a task. Set a deadline.

And if you don’t do it?

You pay the price. Literally.

£5 goes to your punishment of choice:

→ Your ex’s holiday fund
→ A political party you can’t stand
→ Your rival football team’s youth academy
→ …or whatever keeps you up at night

Why I made it is pretty simple honestly.

We’ve all missed deadlines. We’ve all forgotten things. Pushed things aside and hit the snooze button. The reason? Because there’s no repercussions. There’s no punishment apart from the consequences of missing a deadline.

But what if the snooze button actively funded your ex’s vacation to Mykonos?

Now you have far less incentive to snooze.

Which is why I made the “Do it or pay £5” app.

Brutal? Maybe.
Effective? Oh, absolutely.

On a real note?  Coming up with a fun idea is the easy part.

Actually building it? That’s where most people stall.

I’ve been there:

4,000 open tabs.
No idea how to connect a database.
Reading documentation that feels like ancient scrolls.

That’s where Lovable came in.

They made the backend so smooth, I could stay focused on the bit that mattered: the idea.

✅ Working backend (no setup, no config, no tears)
✅ Auth, storage, database. Already done.
✅ Built-in AI, no messy keys or token juggling
✅ Instant cloud deploy


In the left, you can see the Prompt I gave Lovable, Nothing elaborate. It even has an error.

And yet it made it within minutes.

It felt like cheating.

So if you’ve got an idea, even a weird one, don’t let the tech stop you.

Lovable handles the backend.

Lovable handles the AI.

You handle the fun part.

AI photography just got interesting.

We're seeing a shift from "AI that replaces" to "AI that teaches."

WayShot  has launched the worlds first photography guidance app. 

Instead of generating fake images, it acts as a real-time creative instructor:
✓ Compositional guidance using computer vision 
✓ Voice-directed coaching on pose and lighting 
✓ One-tap revamp that preserves authenticity 
✓ Salvaging poorly framed shots with intelligent correction

Why this matters for AI development? This is contextual, real time AI assistance done right. It's not removing human agency. It's expanding human capability.

Compare this to most consumer AI apps that either:
- Generate synthetic content (bypassing skill entirely)
- Apply heavy filters (masking poor technique)
- Distorts facial features (the photo no longer looks like you)

WayShot sits in the sweet spot. Lowering barriers while building competence. For anyone building needing photogaphy help? WayShot is the app to try out.

For anyone building AI products: this is how you create value without creating dependency.

Real time coaching. Skill transfer. Natural results.

That's the future of consumer AI.

Worth studying if you're in the space.

If you want to try it and see the app yourself, here's your chance. A glimpse at the future to come: https://lnkd.in/efZERQ5u

A lot of people have lost their LinkedIn Top Voice badge.

I still have mine. For now.

There are many theories about why it's happening. Some of them might hold water but for the most part it's all speculation. With that said though, quite a lot of people earned their Top Voice badge through one method.

Using AI entirely.

Often, I talk about AI in a positive light because I believe in it. But you don't have to go very far to see:

- AI Generated Posts
- AI Generated Comments
- AI Automations for Engagement

Although I haven't contributed and looked in months, I remember the Community Expert Pages being flooded with AI responses. Just so that users could have "Expert in (Insert Topic Here)".

The concept was simple. LinkedIn would post an article, the article would ask questions, and users who were "experts" would give their expert input. But what happened? 

People copy and pasted the question into ChatGPT. Prompted it. Got a response. Copy and pasted it directly into the article. Earn a badge.

Zero actual thought from the user themselves. All AI.

This happens everyday. People posting opinions entirely generated by AI, comments that are the same, and more.

Opinions that are ChatGPT's. Not theirs.

So although none of us really know why LinkedIn are taking away Top Voice badges? I would bet you that it has a lot to do with authenticity.

The ones who can prove who they are and what they do on LinkedIn and outside of it keep the badge. 

The ones who faked it thinking they can make it? Lost it.

What's your theory?

No agonizingly long caption about AI. Thoughts?

I've heard and come across a lot of AI criticism in my time.

Some are valid.
A majority of it is vitriol and hate.

Ironically the same arguments that have been refuted time and time again. Whether it's around water consumption, general "AI slop" arguments, or whatever else.

There is definitely one thing that a lot of people miss with AI though.

Depth.

For the most part, when we talk about AI or Gen. AI, the automatic assumption is low effort content, work, or similar. Think an email with two emdashes and something cliche about tapestries, dances, or whatever ChatGPT cooks up.

It's either that or poorly generated AI pictures where a person has 7 fingers instead of 5.

But it's so much more than that.

It's the tool that teaches any topic at any level.
It's the tool that's used for freeing people of inhumane tasks.
It's the tool that charities are using to find and acquire funding.
It's the tool that helps figure out illnesses and potential treatment.
It's the tool that businesses use to help free time for people to do great things. 
It's the tool that people are using to understand threatening letters, legal threats, bills, etc.

And in developing countries?

It's the tool that bridges communication gaps. Skill gaps. Knowledge gaps. Capability gaps. Leveling the playing field.

There's far more to AI than just what the internet shows us and what "slop" makes it onto our social media channels. 

I said it once, and I'll say it again. There is no nobility in suffering and doing things that hard way, purely for the sake of it.

I am tired of the luddite stance. 

If you can't see the potential of this type of technology outside of "all it does is make slop!" or whatever else then that is a clear limitation on you rather than the AI tool itself. That is a problem involving your vision. Your ambition. 

You are your own bottleneck.

AI ad companies like Icon tried to put humans out of the equation.

Became irrelevant in 6 months.

Airpost did the opposite and quietly became the creative engine for growth teams, spending $250k+/month on paid social.

They are growing faster than ever:

- $1M ARR in <6 months
- 0% churn (literally unheard of)

And there’s a clear reason: Brands love them for their results.

> DoorDash saw ~40% better CAC vs their internal ads

> Cloaked ran $1M+ through Airpost — with a single ad scaling to $228,025.85

You know you can trust a team with your money if they have managed $350M+ ad spend for Walmart, DoorDash, and Coinbase.

Here’s really why Airpost works:

Everyone else built “AI tools.” Airpost built a system.

At the core is a structured creative taxonomy that understands how ads work at a strategic, narrative, and execution level.

AI handles the heavy lifting. Expert strategists refine every script and every angle.

The result?

10–100+ high-performance video ads per week, tailored to your brand, optimized for velocity, and designed to win.

Not generic. Not DIY.

Enterprise-grade creative output with the speed of AI and the judgment of people who’ve managed hundreds of millions in spend.

In the Andromeda era, growth teams need to take 5× more swings to outperform last year.

People alone can’t do that. AI alone can’t do that.

People + AI — done right — can.

That’s Airpost.

Built for performance. Powered by experts. Accelerated by AI.

Do you agree that AI alone was never going to fix creative performance?

taplio

Edward Morris's Linkedin Analytics

Edward Frank Morris

Performance ScoreEdward Frank Morris's LinkedIn Performance

What is Edward talking about?

Edward Frank Morris's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Edward Morris's Linkedin Analytics

Edward Frank Morris

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Edward Frank Morris's LinkedIn Performance

What is Edward talking about?

Edward Frank Morris's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Edward Morris's Linkedin Analytics

Performance ScoreEdward Frank Morris's LinkedIn Performance

Edward Frank Morris's Best Posts (last 30 days)