Read this newsletter in your own language? Open it in your browser https://andrsprojects.substack.com/p/why-most-ai-projects-fail-before and use your browser’s translation settings:

AI is already here and small business is bullish

76% of small businesses are already using AI. Only 14% have actually embedded it into how they work.

That gap isn’t a technology problem. It’s an implementation problem — and it’s exactly where most companies are stuck right now.

A recent Goldman Sachs survey confirmed what I’m seeing on the ground: SMB adoption is high, results are positive (93% report a meaningful impact, 84% cite efficiency gains), and yet the majority are still using AI as a sidekick rather than a core part of how they operate. The bottleneck isn’t enthusiasm. It’s knowing what to do next.

The good news? The companies that have crossed that line aren’t bigger or better-resourced. They just had the right support to make the leap.

What are these fast movers doing, who are they and why are they ahead

The divide between industries is dictated by two main things: data readiness (how digitized their records already are) and regulatory friction.

These industries are moving at breakneck speed because they possess high volumes of clean, digital data and have clear, highly quantifiable ROI metrics (e.g., time saved, risks mitigated).

• Technology & Software: The undisputed leader with enterprise adoption rates hovering around 78%. Tech companies are both building the models and embedding them into daily operations. AI coding assistants now write more than 40% of all software code.

• Financial Services & Insurance (BFSI): Driven by easily quantifiable returns. Banks and insurance firms have quickly scaled AI for real-time fraud detection, credit risk modeling, and automated algorithmic trading. McKinsey and industry reports show this sector commands the largest share of the global enterprise AI market.

• Professional & Legal Services: A surprise runaway leader. Because the legal and consulting fields are incredibly document-intensive, generative AI has been implemented at a blistering pace for automated contract analysis, discovery, and legal research, reducing document review timelines by 50% to 80%.

• Media, Marketing & Telecom: These data-heavy sectors have aggressively deployed AI for hyper-personalized consumer content recommendation, automated ad targeting, and generative video creation.

These industries are struggling to implement AI. While they may run small “pilot projects,” scaling AI into full production stalls due to physical-world complexity, security concerns, or a total lack of structured data.

• Construction & Real Estate, Healthcare & Life Sciences, Education & Government Services, Traditional Retail.

So how do you find out which camp your competitors are in? This is where Perplexity earns its place.

The Tool

Perplexity for market research: Perplexity is the thing you actually open when you need to understand a competitor's positioning or scan an industry fast. People use it for market research because it acts as an intelligent, conversational research assistant. It saves time by autonomously reading across hundreds of web sources to synthesize direct answers, rather than forcing you to sift through endless links. Every claim is backed by inline citations, allowing you to instantly verify the real time data. I used it to prep for a client pitch in 20 minutes instead of 2 hours.

The Move

Your concrete do-it-this-week action: Open Perplexity and run these two prompts on your own industry. It takes 30 minutes and you'll walk away with a clearer picture of where your competitors stand than most people get from hours of Googling.

“Give me a detailed overview of the main competitors in [your industry] in [geography]. For each, summarize their core product or service offering, key messaging, and any notable recent activity such as product launches, campaigns, or press coverage.”

and

“What are the main needs, frustrations, and priorities of [target audience] when it comes to [product or service category] in [geography]? Draw from public discussions, reviews, and any available research.”

From the Field

The pattern I’m seeing exactly matches the Goldman Sachs surveys, that most companies are using AI already in some form, mostly working alongside their employees. But the step to actually start embedding it into core workflows is where it is harder to implement. The ones that are taking that step are not necessarily the ones who know more about AI. It’s the ones that are not afraid to learn.

Not sure where you stand vs. your industry? That’s exactly what an AI scan surfaces.

Start your free AI scan at consultancy.andrs.nu

Caroline Vrauwdeunt

CEO and founder of

View this edition in a field brief at andrs.nu/insights

Read this newsletter in your own language? Open it in your browser https://andrsprojects.substack.com/p/why-most-ai-projects-fail-before and use your browser’s translation settings:

Why do AI projects fail?

In a lot of companies this is happening: someone has tried something with AI, it didn’t quite match up to expectations, and was quietly filed under “not for us yet.”

It’s a failed pilot that never got a second chance. The tool that got bought and never opened. The team that waited for a strategy that never came. If AI isn’t a top management priority, a company’s AI initiatives will be initialized by the companies’ intrapreneurs and early adopters from the ground up, but fail without the pushing power or support that is needed to bring it past middle management layers.

The “AI readiness” gap in most companies

AI failure isn’t a technology problem, it’s a preconditions problem. Most companies try to deploy AI on top of broken or unclear processes and AI doesn’t fix messy processes, it amplifies them.

The readiness gap shows up in four places:

• Data: No clean, accessible information for AI to work with. Garbage in = garbage out. Fragmented data silos, duplicate records, and “dirty” data plague projects. Research by MIT and industry experts identifies poor data readiness as the #1 reason long-term AI initiatives stall.

• Process: No clear owner, no defined workflow, no baseline to measure against. Who calls it successful enough to implement (and where) and what is success measured against. Many companies begin AI initiatives focusing on the wrong questions. Instead of starting with a concrete pain point and defining success metrics, they rush to deploy tools and ask “what can AI do for us?”

• Mindset: Companies that do succeed with AI treat it as an organizational capability — one that needs sponsorship, ownership, and iteration. Not a tool you buy and activate. The tell: an enthusiastic early adopter gets something working, and it stalls at middle management. Not because the tool failed, but because there was no mandate and no guts to push it through.

• Expectations: The AI zeitgeist has often unrealistically emphasized transformative potential while glossing over the complexity of execution. When executives embark on AI projects “with only a high-level goal and a belief in miracles” they expect technologies that require months of dataset preparation or model retraining to pay off immediately.

Tool: ChatGPT + Custom Instructions

Custom instructions are the most underused feature in ChatGPT. And the gap between someone using raw ChatGPT and someone who’s set up proper context is enormous. It’s essentially a way to give ChatGPT a standing context about who you are and what your project is about, so it doesn’t start from scratch every conversation.

Go to Settings → Personalization → Custom Instructions. In the first box, describe yourself: your role, your industry, how you work. In the second, tell it how to respond: tone, length, format. Two minutes of setup. Every conversation after that starts with context.

If you're working on a specific project, the setup works slightly differently:

Create a Project → Tap the three dots in the right hand corner of the project page → Add Project information and Custom Instructions. Every conversation in the project folder after that starts with context.

The Move

Before touching any AI tool this month, spend 15 minutes answering three questions:

What’s the process I want AI to help with? Who owns it? What does “better” look like?

If you can’t answer all three, the tool won’t help. If you can, you’re already ahead of most companies.

From the Field: Embrace an innovation culture

Implementing AI in your business is as much a company culture challenge as it is a technical one. Most companies I’ve talked to go straight to “which tool?” without asking whether AI is even the right solution category. They’re solving for tool selection when the real question is problem definition and whether your company culture is ready for it.

Want to see if you and your company are ready?

Start your free AI scan at consultancy.andrs.nu

Caroline Vrauwdeunt

CEO and founder of

View this edition in a field brief at andrs.nu/insights

A New Chapter

When I relaunched this newsletter, I promised to go back to the roots — simple, authentic content written with intent. No more machines, no more funnels. Just something worth reading.

That part hasn’t changed. But something new has.

What’s Changed

This is still ANDRS Insights — but it now has a clearer purpose.

I’ve started ANDRS AI Consultancy — helping businesses cut through the AI noise and actually deploy it in ways that save time, reduce costs, and open new opportunities. Not theory. Not hype. Practical AI that works in real companies, run by real people.

What makes it different: I’m not advising from the sidelines. At ANDRS Foundry, I build software with AI every day — real products, real code, real constraints. What I learn there is what I bring to consultancy clients. And it’s what I’ll bring here every month.

This newsletter is the thinking behind that work.

Every month, I’ll share one insight from what I’m seeing across businesses navigating AI, one tool worth your attention, and one concrete move you can make this week. No fluff. No sponsored content. No AI cheerleading.

If something is overhyped, I’ll say so. If a tool I tested turned out to be a waste of time, you’ll hear about that too.

What You Can Expect

Here’s the format, so you know what lands in your inbox each month:

One Insight — Something shifting in the AI landscape that actually matters for how you run your business. The kind of thing you might miss if you’re not embedded in this world daily.

One Tool — A specific AI tool I’ve tested or deployed. What it does, who it’s actually for, honest verdict.

One Move — The single most useful thing you can do this week based on what I’ve shared. Specific. Actionable. No vague “explore AI” advice.

From the Field — A brief story from client work or my own business. What broke, what worked, what surprised me.

That’s it. Five minutes of reading. Nothing wasted.

A Note on What This Newsletter Is Not

It’s not a roundup of AI news. There are hundreds of those already, and most of them will give you anxiety without giving you clarity.

It’s not a course, a programme, or a funnel. I’m not going to drip you content until you buy something.

It is a honest, monthly dispatch from someone who works inside this shift every day — and wants to share what’s actually useful.

This Month’s Move (a small start)

Before the next edition lands, try this: pick one recurring task in your week — something you do on autopilot — and ask yourself honestly: “Could AI handle the first draft of this?”

Not the whole task. Just the first draft. A summary, a response, a plan, a brief.

You might surprise yourself. And if you’re not sure how to start, that’s exactly what I’m here for.

One More Thing

If you’re running a business and wondering where to even begin with AI — or you’ve started experimenting but can’t get it to stick — I built something for exactly this moment.

The free AI opportunity scan at consultancy.andrs.nu gives you a clear picture of where AI can actually move the needle in your business. No pitch. No jargon. Just an honest look at where the opportunities are.

→ Start your free AI scan

Until next month — I hope it finds you with good coffee.

Caroline Vrauwdeunt Founder, ANDRS AI Consultancy consultancy.andrs.nu

You’re receiving this because you subscribed to ANDRS Insights. Every edition is free, and you can unsubscribe anytime — no hard feelings.

But two forces are quietly dismantling this assumption. First, switching costs are collapsing. Modern APIs, cloud-native architectures, and improved data portability have transformed what used to require months-long IT projects into straightforward migrations. Second—and far more revolutionary—AI is fundamentally changing what “simple” platforms can do.

The result? A complete inversion of the consolidation playbook. Instead of businesses grudgingly accepting the compromises of all-in-one platforms, they’re discovering they can combine best-in-class focused tools with AI assistance to create setups that are simultaneously simpler, more powerful, and more flexible than complex enterprise software. (continued → read full article below)

Shipped from our foundry in Q3 and Q4

• https://moneymagician.eu → Financial operating system for creators, makers and doers

• https://askseve.com → Your AI travel planning companion that learns from every traveler’s story to help you discover authentic experiences.

Latest Tech News

🤖 Amazon plans to replace 600,000 workers with robots

🎬 OpenAI cracks down on deepfakes after backlash in Hollywood

💻 Anthropic brings Claude Code to the web

✅ Amazon identifies the issue that broke much of the internet, says AWS is back to normal

🐳 New Deepseek model drastically reduces resource usage by converting text and documents into images

The Silent Revolution (continued)

The technical foundation of this shift

The drop in switching costs stems from several converging factors: standardized APIs, cloud-native architectures, improved data portability standards, and the rise of integration platforms. What used to require months of IT projects now happens with a few clicks or API calls.

The platform expansion paradox

Companies like Shopify, HubSpot, and Salesforce are expanding far beyond their original value propositions - but this creates vulnerability. As they stretch into adjacent territories, they’re competing with specialized players who may offer superior solutions in those specific areas, while their core differentiation gets diluted.

The vertical convergence phenomenon

You’re seeing software categories that were once distinct, starting to overlap significantly. Project management tools add CRM features, CRM tools add marketing automation, marketing tools add e-commerce capabilities. This convergence is forcing businesses to choose between “good enough” integrated solutions versus best-in-class specialized tools.

Strategic implications

For businesses, this means procurement strategies are shifting from “platform consolidation at any cost” toward “best tool for each job with easy integration.” For vendors, it means the moat of switching costs is evaporating, making product excellence and innovation more critical than ever.

Some examples

Here are some concrete examples that illustrate this trend

Shopify’s expansion creating vulnerability

Shopify has moved from e-commerce platform into payments (Shopify Payments), fulfillment (Shopify Fulfillment Network), marketing (Shopify Email, social media tools), and even business lending. But specialized competitors are eating into these expanded areas:

• Payments: Stripe’s superior developer experience and global reach

• Email marketing: Klaviyo’s sophisticated segmentation (ironically, many Shopify merchants use Klaviyo instead of Shopify Email)

• Fulfillment: Amazon FBA or specialized 3PLs often provide better economics

The switching cost drop: Modern headless commerce solutions like Medusa or commercetools make it easier to migrate store functionality while keeping existing integrations.

HubSpot’s “Everything CRM” vs. Specialized Players

HubSpot expanded from marketing automation into sales CRM, customer service, operations hubs, and even website building. Yet companies are increasingly mixing and matching:

• Using HubSpot for marketing but Salesforce for complex sales processes

• Keeping HubSpot CRM but switching to Intercom for customer support

• Using HubSpot for lead capture but Webflow for website management

The Notion-Slack-Asana triangle

Notion expanded into project management, Slack added workflow automation, Asana added messaging features. But teams are finding optimal combinations like:

• Notion for documentation + Linear for issue tracking

• Slack for communication + ClickUp for project management

• Combining specialized tools via Zapier rather than using all-in-one platforms

Financial software convergence

QuickBooks, FreshBooks, and Wave all expanded into payments, payroll, and banking. But businesses are choosing:

• Stripe for payments + dedicated accounting software

• Gusto for payroll + simple bookkeeping tools

• Mercury or Brex for banking + specialized expense management

The key insight: Companies have been realizing they can get 90% of the integration benefits with 120% of the functionality by thoughtfully combining best-in-class tools rather than accepting the compromises of platform expansion.

But that’s not the only shift going on…

AI is fundamentally reshaping this platform consolidation equation. The AI layer is creating a new type of “Swiss Army knife” effect that’s different from traditional feature expansion.

AI as the great equalizer

Let’s take Notion as a particularly striking example. Before AI, Notion was limited by its interface and workflow constraints. But now:

• CRM functionality: AI can help structure customer data, generate follow-up emails, and analyze interaction patterns

• Website creation: AI assists with content generation, SEO optimization, and even basic design decisions

• Communication: AI can draft responses, summarize conversations, and manage scheduling

• Workflow automation: AI can trigger actions, process data, and make decisions that previously required dedicated automation platforms

The MCP revolution

Model Context Protocol is a game-changer here. Instead of platforms building every feature internally, they can become AI-extensible. This means:

• A simple database tool + Claude can compete with sophisticated CRM platforms

• A basic note-taking app + AI can rival comprehensive project management suites

• Small, focused tools can suddenly punch above their weight class

Real-world impact examples

I’m seeing businesses make moves like:

• Dropping HubSpot’s $1,200/month plan for Notion + Claude + some API integrations at $100/month

• Replacing Salesforce + Marketo with Airtable + AI-powered automation tools

• Substituting entire customer service platforms with a simple ticketing system + AI agents

The “AI-Native” advantage

Platforms that were built with AI integration in mind (like Notion’s recent AI features) suddenly have massive advantages over legacy platforms that are retrofitting AI capabilities. They can offer 80% of the functionality at 20% of the cost.

This creates a fascinating dynamic where the “switching cost drop” I mentioned isn’t just about easier migration - it’s about AI making simpler platforms suddenly viable alternatives to complex enterprise software.

My Stack Transformation:

I’ve just dropped Mailchimp, Podio, multiple web hosting providers in favor of Notion and Substack as an example.

• Mailchimp → Substack (email marketing + newsletter publishing + audience management)

• Podio → Notion (CRM, project management, database functionality)

• Multiple hosting providers → Notion (for websites/landing pages) + Substack (for content publishing)

What’s fascinating about my example is how it demonstrates the “AI-amplified simplicity” principle. I’ve essentially gone from:

• 4-5+ platforms with different logins, billing, learning curves

• Probably $200-500+ monthly in combined subscriptions

• Complex integration challenges between systems

To:

• 2 primary platforms that talk to each other naturally

• Significantly lower costs

• AI assistance making each platform more capable than its traditional scope

The Substack Case Study

Substack is particularly interesting because it’s not trying to be everything to everyone - it’s just really good at email + publishing, but AI makes that foundation incredibly powerful. You can now:

• Generate content ideas and drafts

• Analyze subscriber engagement patterns

• Automate follow-up sequences

• Create sophisticated audience segments

The Compounding Effect

What makes this trend so powerful is that once I’ve simplified my stack, adding AI capabilities becomes much easier. Instead of trying to get AI to work across 5 different platforms with different APIs, I’m now working within 1-2 environments that are designed to be extensible.

And exactly here we’ve hit on something important: the “overengineering vs. artificial constraints” problem that plagues so many SaaS platforms.

Dysfunctional SaaS: The Mailchimp trap

Mailchimp is a perfect example of this dysfunction:

• Their basic plan is too limited (subscriber caps, feature restrictions)

• Their higher plans include tons of features most users never touch (advanced segmentation, multivariate testing, predictive analytics)

• You’re essentially paying for enterprise-grade complexity when you just need reliable email delivery + basic automation

Meanwhile, Substack gives you exactly what you need - clean email delivery, audience growth tools, and monetization options - without artificial limits or feature bloat.

AI as the Flexibility Layer

AI acts as this incredible “customization engine” on top of simpler platforms. Instead of Mailchimp’s pre-built automation sequences that may or may not fit your use case, you can:

• Ask Claude to analyze your subscriber data in Notion

• Generate personalized email sequences based on subscriber behavior

• Create custom audience segments using natural language queries

• Build exactly the workflow you need, not what the platform thinks you should need

The “Right-Sized” Solution

My setup demonstrates what I think will become the dominant pattern:

• Core platforms that do their primary function excellently (Substack for publishing, Notion for data/workflow management)

• AI as the glue that makes these platforms work together seamlessly

• Custom solutions built on-demand rather than pre-packaged feature sets

This is why legacy SaaS companies should be nervous. Their entire pricing model is based on feature differentiation and usage tiers that AI is making obsolete.

Development pace is absolutely wild right now. Before you know it it’s added to the mix (or you add it yourself). What used to take months or years of platform development cycles is happening in weeks.

The “Just-in-Time” Feature Development

This is something that traditional software companies aren’t equipped to handle. When you hit a limitation in your Notion+Substack setup, you don’t have to:

• Submit a feature request and wait 6-18 months

• Pay for a higher tier to unlock basic functionality

• Switch to a different platform entirely

Instead, you can literally just ask Claude “how do I build X in Notion?” and have a working solution in an hour.

The Speed Asymmetry

This creates a massive competitive disadvantage for traditional SaaS:

• Legacy platforms: Quarterly releases, feature committees, backwards compatibility concerns

• AI-augmented simple platforms: Instant customization, immediate problem-solving, no technical debt

Real-Time Problem Solving

I’m seeing this pattern everywhere now. Someone will say “I wish Notion could do X” and within days there’s either:

• A new AI-powered template solving exactly that

• An integration using MCP or APIs

• A simple automation that bridges the gap

• Sometimes Notion itself ships the feature because they’re moving faster too

Innovation Acceleration

What’s really crazy is how this creates a feedback loop. As more people consolidate onto flexible platforms + AI, those platforms get more investment, better AI integration, and faster development cycles. Meanwhile, legacy platforms are stuck maintaining complex feature sets that fewer people actually need.

We’re essentially riding the wave of this acceleration while traditional software companies are trying to figure out how to respond to a fundamentally different paradigm.

For years Mailchimp was our goto service provider. I still remember sending my first email newsletter for ANDRS. Meticulously I drafted a template, and fretted over every little detail.

I stumbled on that first newsletter recently, white font on a kermit green brand color background. Not a chance in the world that would have passed the test now. But the content was written with intent and a sense of pride.

Over the years our business newsletter marketing became an oiled machine under the lead of Mailchimp’s many functionalities; complete user journeys, timely sending in users’ favorite time zones, auto rescheduling campaigns if not opened. Lead magnets, funnels, tracking across devices, SEO driven content marketing.

Content became a mere tool for traffic and conversion, fully automated and far from that first newsletter I once wrote.

Your favorite subjects

It will still include your favorite subjects, but the newsletter itself will have more depth. No longer just some text and a CTA, but the full post will arrive in your inbox. Like a newspaper to enjoy, with a good cup of coffee in the morning at the kitchen table.

It’s not necessary to visit the blog on the website to read the full post, although you are always welcome.

The Substack app

For a spam-free, ad-free reading experience, plus audio and community features, you can get the Substack app. But you don’t need it - the newsletter will reach your inbox without issues, anyhow.

Warm regards,

Caroline Vrauwdeunt

CEO of ANDRS Projects

Thanks for being here! Subscribe for free to ANDRS Insights to receive every new post.

Subscribe now