# Everyone Has AI. What's the difference?

*(Formatted and edited in Claude. Shh. It's ok to use AI. Sometimes it's good.)*

**Conor Corcoran** (HI AI Design) · 19 October 2026 · 5-minute read
Tags: AI strategy, Branding, Startups, SaaS, Go-to-market

![AI-generated image of AI looking at itself](https://www.hiai-design.com/images/blog-masterpiece-ai-sameness.webp)

This article takes five minutes. Please read to the end. The last few lines are the point. Focus!

Ahh recruitment , the chefs kiss of AI enshittification ([Cory Doctorow](https://pluralistic.net/2023/01/21/potemkin-ai/)). Let’s pick a random company, ahh ,Workday, they will do, allegedly they have 30 million job seekers submit more than 432 million applications through it every year [5]. As I write this, on 15 October 2026, its Trustpilot page shows 1.1 out of 5 from 504 reviews, 98% of them one star [3]. Take Trustpilot with the usual pinch of salt, since it is self-selected and the profile is unclaimed, but read the themes: re-typing a CV you have just uploaded, parsing that gets it wrong, a new account for every employer [3]. Workday has itself logged a product defect where resume parsing on career sites stopped filling in fields (June 2025, since resolved) [6].

Credit where it's due. On 13 October Workday announced Candidate Profile: save your details once, reuse them across employers that use Workday, no new account each time, a one-time code instead of a password [4]. It says it can prefill up to 80% of standard fields [5]. It isn't due until early 2027 [4]. That fixes the account problem. Next step: a parser that works. A platform handling hundreds of millions of applications a year that can't reliably read a CV is not a story about AI solving problems. It's a story about AI as packaging.

## I run an AI consultancy, so this is aimed at me too

Two years ago, putting "AI" at the front of everything said you were early. Today it says nothing. Strip the word from our own homepage and ask what's left?

Agentic deployment and development, AI compliance and AI management system all in the bin.

What’s left? GTM/Sales/Marketing/Operations/Pricing/Budgeting/Branding/Project management, that’s my audit done. I still have skills and knowledge left to sell. What would you have if you did the same?

Buyers have noticed: 98% of B2B buyers at least sometimes feel competing vendors do the same thing, and 64% struggle to tell them apart from their websites and messaging [2]. Only 6% of the leaders surveyed called their own brand very distinctive [1].

So what am I selling? Let me tell you a story………

## Every start-up hub, the same story

What are they doing? Why are they here?

Wherever the best early-stage teams gather, they're building the same thing at the same time. Station F's Future 40 is full of AI agents for customer support, finance, productivity and sales [7], and the campus says roughly 80% of its startups build on AI [8]. One count of Y Combinator's Spring 2026 batch has 95% touching AI and 70% building agents [9]. Depending on how you count, between two-thirds [10] and 91% [11] of the Summer 2026 batch is AI-related, after 154 of 170 the summer before [12]. That's remarkable talent. It also means "we do AI" is no longer what a buyer remembers.

The rest of Europe feels the same oppressive sameness. Germany saw a record 3,053 startups founded in the first half of 2026, more than a third in AI [13]. AI took 55% of European venture capital, but 73% of that went to just 38 companies [15]. In Berlin, four deals made up 58% of the half-year's funding, and the investor quoted warned that smaller startups are being squeezed out [14]. The technology is strong everywhere. What's scarce is a clear story.

Funders in Asia are asking for one. Google and Accel reviewed more than 4,000 applications for an India AI program, and about 70% of the rejected ones were thin layers on existing software [16]. In Malaysia, AI-related startups are 60% to 70% of roughly 1,000 yearly applications to Cradle Fund, and one investor said they back sector expertise, not the AI label [17]. A Singapore commentator put it simply: in a crowded market, brand and story set the winners apart [18]. We need these new companies to start telling unique and clear stories with standout branding and a great use case.

## The snake eats its own tail

What is our USP? A great business question. The companies unique selling position, key is in the first word unique. Let’s boil the ocean and get granular on this. Or let’s use plain English, why do all the websites and logos look the same? What do we sell and what pain-point does it solve? Let’s use AI.

A Nature paper found that when AI models learn from content produced by other AI models, they lose the rare, unusual material first, their output narrows towards the average, and in the worst cases it degrades into nonsense [21]. That’s about training data, not marketing, but brands run the same loop. Brief an AI on your competitors’ pages, many of them already AI-written, and it hands back the average of them. They do the same with yours. Each round sands off a little more of what made anyone different, and the strange, specific, human details go first. Nobody is daring to create a masterpiece and try and be different. The human details will be what people remember. (See the article image, AI generated looking at AI, boom).

The death loop of originality and uniqueness breeds sameness and group think. I missed the memo on gilets and hoodies as acceptable office wear. Also, when did financial projections become bullshit bingo for metrics and future revenue? The fish rots from the head down, and if we look at the AI market, it's very smelly at the moment (please read Ed Zitron, “[Where Are All The Data Centers?](https://www.wheresyoured.at/where-are-all-the-data-centers/)” [23]). Losing 80 billion to turn over 20? My management accounting lecturer would have thrown the book at me. (Books are fun.) In the real world turnover for show, margin for dough. Get the spreadsheets out and start doing the costs. Do all these guys who are losing billions deserve our praise or scrutiny? What is success in business anymore? Most of them shouldn’t be allowed run a bath let alone a global billion dollar company.

AI doesn’t create more free time for you at work. All AI does is give you more time to do more work, for example you complete all your tasks, looks good, and you have used your favourite AI to help, spreadsheets, PowerPoint, Gamma all excellent with an AI booster. Then your boss, whose use of AI is inherently superior to yours, will ask their AI about your work and their AI said your work had to be more original. Kafkaesque: describes situations, environments, or systems that are nightmarishly complex, bizarre, illogical, and oppressive, closely mirroring the surreal fiction of an author by the same name. I’m sure he hoped his books wouldn’t come true, but it has. Look at your emails from 2019 and 2020 short and to the point not word jumble missives like today. Wading through AI slop, listening to AI slop bots, what are they doing? Why are they here? What are we doing about it?

## Brewing is easy, we can all homebrew, who fancies a pint?

Buy a homebrew kit, add water, yeast and sugar, put in warm dark place, burp co2 out very few days and you will have beer/wine/cider depending on the kit and ingredients used. Its officially a beverage with alcohol and everyone can do it, barriers to entry are low, we have the internet and you tube and its relatively easy to do. I do not see Guinness defecting in their pants with shock and terror that that someone can replicate their beer. Vibe coding is the same. I think homebrewers and vibe coders are the same person, they both need burping for Co2 and be taken out of their dark place. A homebrew beverage does not have not have the same quality, It’s not the same brand , it’s not the same provenance. Vibe coding is the same to Saas. You can make a website/agent etc. at home but will taste as good as the real thing?

Please stop saying SaaS is dead. When Edison, Tesla and Westinghouse wired the world, we didn't all become electricians. We became people who flip a switch and expect it to work. The "SaaS is dead" debate never resolves because each commentator means a different layer [22]. The easy moat is gone. Trust isn't, expertise isn't. I think companies that embrace AI, train their staff and enhance the work will be the winners. Laying off people creating a brain drain and 100s of potential competitors is not a smart play.

We still buy and trust the big brands more than we should. Back at the end of 90s when the internet was still young( and so was I) trust was the biggest problem. The internet didn’t explode until we could trade in a safe environment, payments, security, were the key aspects of online success. Holding a torch under your face in the dark in front of a mirror and repeating AI will kill us all and take our jobs is not helpful and very silly. (thanks Cory Doctorow, [google him](https://pluralistic.net/), he is great). The big AI labs can’t make money with their current business model. The LLM has no clothes. When the LLM gets dressed and we recognize the clothes its wearing, then we can start to relate to them. Naked lunatics saying the world is nigh are not people you hand money over to. Also where do they put the money?

## The wake-up call

Having AI in your stack is not a differentiator. A name, a product, a service that means something is. You need a story people can repeat, a product that does one job well, a price on the page, and a human who picks up the phone. The seven Ps still stand, and none of them says "add a chatbot."

Before you ship another feature, first ask: will someone pay for this?

## Seven questions to ask before you ship AI

If you build AI products, ask yourself seven questions.

1. Product: What one job does it do? AI makes building cheap, so the job is the product, not the model. 450 models and counting are available on open router with one API key.
2. Price: Can a buyer see the cost without a call or a mail to the supplier? Will they be pissed off at the end of the month? They say AI lets you price on outcomes, like cases resolved, instead of seats. How do you know the outcome? Clients don’t like sales people, companies won’t tell you what it cost per outcome if they have to. The company wants the best product and the best price and then go away please until I need you or you have a promo for me or a cool update that I need. A lawyer won't charge less per hour if they used AI. It’s hard. How to price token use? People like predictability of outcome and stable pricing. Sell people a product not a cloud and promise.
3. Place: Where does my buyer already work? AI should arrive inside their tools and language, not as another login. How many tabs and apps do you have open at one time in the day now? Does the person even want AI? Do we need an AI phone bot that points you back to the same number on your own website when you call for an insurance query? (my bank has that service)
4. Promotion: What can only we say? AI can write a thousand versions of your message, AI doesn’t love you product, you do and so should the client and so should your team. Be authentic. Provenance is key in FMCG. People pay a fortune for Bio labels, AOC for wine. They want to show quality and attach a premium price for it. Will we pay more for human made?
5. People: Where does a human step in? AI does the repetitive work. A person owns the judgement and the phone call. Do the humans even want the AI? Check out Microsoft copilot. It got put everywhere and no one wanted it. A U2 album on a iPod.
6. Process: What happens after go-live? AI is never a one-off project. It needs monitoring and an owner. Training, simplicity, traceability, non-hallucination. Is this an enhancement or a drag? Is my AI app as useful as an ashtray on a motorbike test? (should be a real benchmark)
7. Physical evidence: What proves it works? AI claims are cheap. Numbers, case studies and audits aren’t.

Fail any of the seven and you, the AI company, will be sending your customer/candidate/supplier an auto-generated email that starts not with their name but with the ubiquitous "Dear M", all because your AI didn't extract their name properly. (This happens.)

For HI AI Design, (Human Intelligence and Artificial Intelligence) that means plain-language promises with a number attached, prices a buyer can find without booking a call, and humans you can contact. If we can't explain what makes us different without the word AI, we don't have a business, we have a template. The teams building the next generation of products deserve a story as good as their code. That's the work we do. Let me help you write your next chapter.

If you read this far, thank you. Let's start trying to be unique again. Please share any Kafka AI stories you have with me. Drop me a mail at contact@hiai-design.com.

Visit the Atelier: https://www.hiaiworks.com/ · Talk to us: https://www.hiai-design.com/#contact

## Sources

1. B2B SaaS branding is stuck: 2025 survey findings (Wynter)
2. B2B Vendor Differentiation Survey (Wynter)
3. Workday reviews (Trustpilot), viewed 15 October 2026
4. New Workday Candidate Profile lets job seekers save their information once (Workday newsroom, 13 October 2026)
5. Workday Candidate Profile Reinvents How People Apply to Jobs (Workday blog)
6. Resolved: Resume parsing on the career site is not functional (Workday alert, as mirrored by a university customer)
7. STATION F Future 40, 2025 edition
8. Station F profile (ActuIA)
9. YC Spring 2026 batch trends
10. Y Combinator batches by outcome (Dealroom)
11. YC Summer 2026 analysis (ChainCatcher)
12. YC Summer 2025 AI concentration (CO/AI)
13. Germany sets new six-month start-up record (GTAI)
14. Berlin retakes startup funding crown
15. European VC for AI reaches $23bn in H1 (AI Magazine)
16. Google, Accel India accelerator picks 5 startups, none "AI wrappers" (TechCrunch)
17. AI startups crowd Cradle Fund applications (The Edge Malaysia)
18. Singapore's homegrown AI startups (People Matters)
19. Why AI-generated websites all look the same (Managed Code)
20. Why AI standardises the visual identity of start-ups (Startupbusiness)
21. AI models collapse when trained on recursively generated data (Nature, via Oxford OATML)
22. Structuring the "SaaS is Dead" Debate (SmartScope)
23. Where Are All The Data Centers? (Ed Zitron, Where's Your Ed At)