The $300B Lie: "Our Business Is Doing Great" Is a Death Sentence

ServiceNow's CEO went on CNBC this week and said "our business is doing great." India's outsourcing industry is telling the BBC the same thing about its $300 billion labor arbitrage machine. If you've studied what happened to Kodak, Blockbuster, or Nokia, you recognize this sentence. It is the corporate equivalent of "the iceberg is far away and we're a big ship."

I run a tech company from Jönköping, Sweden. Not San Francisco. Not Bangalore. Jönköping. And from here, watching two of the largest pillars of global enterprise IT simultaneously insist everything is fine while the ground beneath them turns to liquid, I have one question for every CEO reading this: do you actually believe them, or are you just comfortable?

The Denial Pattern Is Always the Same

Let me map it out. ServiceNow sits on a massive installed base of workflow automation. Thousands of enterprises pay billions annually for what is, at its core, a ticketing and process orchestration layer. It works. It prints money. The CEO is not lying when he says the business is doing great today.

But here is what he is not saying: the entire category of "workflow automation via human-configured rules and forms" is being eaten from below by AI agents that can reason, act, and self-correct. The gap between a ServiceNow workflow and an autonomous agent pipeline is closing at a rate that should terrify any honest product strategist inside that building.

Then look at India's outsourcing sector. The BBC is asking the right question right now: can a $300 billion industry built entirely on the cost delta between a developer in Hyderabad and one in Stockholm survive when an AI agent development stack can do what a team of 15 offshore engineers does, in minutes, for pennies? The BBC piece lays out the scale. This is not a thought experiment. Infosys, Wipro, TCS. These are companies that employ millions. Their entire value proposition is "humans are cheaper here." That proposition has an expiration date, and the date just moved up.

Stanford economists who spent years telling us AI displacement was overhyped are now reversing course. The timeline, they say, has compressed dramatically. When the academics who were your best argument for "don't worry" start worrying, you should pay attention.

The Consultancy Shell Game

Meanwhile, watch what Accenture is doing, not what it is saying. Accenture just acquired Keepler, a data analytics firm. Tredence is expanding its Google Cloud AI practice. The big consultancies are quietly rebuilding themselves as AI-native shops. They are absorbing AI capability as fast as they can.

But what are they telling clients? "Take it slow." "You need a strategy first." "Let us run a six-month assessment."

This is not advice. This is a business model. Every month you spend "assessing" is a month you are paying them instead of building. The consultancies have a structural incentive to slow you down while they speed themselves up. If you are a CEO paying a Big Four firm to tell you how to adopt AI, understand that you are paying the fox to design the henhouse security system.

The real work of building custom AI solutions for your specific business does not require a 200-slide deck and a steering committee. It requires engineers who understand your domain, models that fit your data, and the willingness to ship something that works before it is perfect. That is what we do at HEIMLANDR. No slide decks. We build.

Sweden Is Sleepwalking. The SCB Data Proves It.

Now let me bring this home to Sweden, because this is where it gets personal.

SCB, Statistics Sweden, released its latest report on AI adoption in Swedish enterprises. The headline looks decent. Adoption is up. More companies report using AI. But read past the headline and the picture is grim. The adoption is cosmetic. Chatbots on customer service pages. Basic analytics dashboards. Pilot projects that never graduated to production.

Almost nobody is deploying autonomous agents into core business processes. Almost nobody is replacing actual workflows with AI-native architectures. Swedish firms are checking the "AI" box on their annual reports without changing how anything actually works.

Compare this to what is happening in the US and China. American startups are shipping AI automation for business processes at a pace that makes Swedish enterprise IT look like it is running on dial-up. Chinese manufacturers are deploying AI into production lines with a speed and ruthlessness that European regulators cannot even conceptualize. And here in Sweden, we are having fika and discussing whether we need an AI ethics board before we can deploy a support agent.

I love Sweden. I built my company here on purpose. The engineering talent is extraordinary. The infrastructure is solid. The trust between businesses and employees is real and valuable. But the institutional speed is killing us. By the time a Swedish enterprise finishes its AI strategy process, an American competitor will have shipped three iterations and a Chinese one will have automated the entire function.

Swedish policymakers are not helping. The EU AI Act is well-intentioned but it is regulation designed for a world that existed two years ago. It classifies risk based on categories that are already obsolete. It creates compliance overhead that disproportionately hits European builders while doing nothing to slow down the American hyperscalers or Chinese state-backed labs that are actually setting the pace. The EU's regulatory framework reads like it was written by people who have never shipped a product under pressure. Because it was.

Swedish enterprise leaders need to stop treating AI adoption as a risk management exercise and start treating it as a survival exercise. The risk is not that you deploy AI and something goes wrong. The risk is that you don't, and your competitors eat you alive.

Where This Goes: 2026 to 2030

Let me lay out what I think happens in the next two to five years, because this is what should be driving your decisions today.

The Outsourcing Collapse (2026-2028)

India's IT outsourcing sector does not die overnight. But it hollows out. The commodity work, testing, basic development, L1/L2 support, system administration, disappears into AI agent pipelines. The smart Indian firms (and there are many smart people in those organizations) pivot to AI-native services. The slow ones become shells. Every Swedish enterprise that depends on offshore IT labor for core operations needs a Plan B right now.

The Platform Reckoning (2027-2029)

ServiceNow, Salesforce, SAP. These platforms survive but their pricing power erodes massively. When an AI agent can orchestrate a workflow without a $150/user/month platform license, the negotiating leverage shifts to the customer. Expect brutal repricing. If you are locked into long-term enterprise contracts with these vendors, start planning your exit architecture today. Think modular. Think agent-first. Think about SaaS architectures that you own and control.

The AGI Proximity Effect (2028-2030)

We do not need AGI to arrive for AGI proximity to reshape markets. As models get closer to general reasoning capability, every incremental improvement makes another category of human knowledge work automatable. The question is not "when will AGI arrive?" The question is "how capable does AI need to be before my current business model breaks?" For most enterprises, the answer is: less capable than it already is. The bottleneck is not the technology. It is organizational willingness to deploy it.

The Regulatory Gap Becomes a Chasm

EU regulation will lag reality by at least three years. This means European builders operate in a strange liminal zone: technically constrained by rules written for yesterday's AI, practically competing against American and Chinese companies that face no such constraints. Swedish companies that wait for regulatory clarity before acting will find that clarity arrives just in time to be irrelevant.

What to Look At

If you are a builder, a CTO, or a founder trying to cut through the noise, here are the things I would spend time on this week:

1. Langflow (AI Agent Pipelines)

Langflow has crossed 147k stars on GitHub and for good reason. It is one of the best open tools for building and deploying AI-powered agents and workflows. If you are still hand-rolling your agent orchestration, look at this. It will save you weeks. This is what practical AI agent development looks like in 2026, not vendor slide decks, but open source tools that let you build and iterate fast.

2. Claude Code (Agentic Coding in Your Terminal)

Claude Code from Anthropic is at 117k stars and it represents where software development is heading. An agentic coding tool that lives in your terminal, understands your codebase, handles git workflows through natural language. If your engineering team is not using tools like this, they are working with one hand tied behind their back. This is not about replacing developers. It is about making every developer 3x more effective.

3. AutoGPT (The Agent Framework That Started It All)

AutoGPT sits at 183k stars. It has evolved significantly from its early chaotic days. The mission of accessible AI for everyone, with real tools you can build on, matters. If you are evaluating agent frameworks, this belongs in your comparison set alongside Langflow and the commercial options.

4. The SCB Report Itself

If you run a Swedish company, read the actual SCB data on AI adoption. Not the summary. Not the press release. The data. Look at where your industry sits. Then ask yourself honestly: are you ahead of that curve or behind it? If behind, you have work to do. Starting today.

What You Should Actually Do

I will be concrete. If you are a CEO or CTO reading this on a Sunday morning, here is what I would do this week:

Monday: Ask your team to list every process that currently depends on offshore labor or a SaaS platform to function. Not the strategic ones. The boring ones. Ticket routing. Data entry. Report generation. Test case creation. That is your automation target list.

Tuesday: Pick the three highest-volume, lowest-complexity processes from that list. Assign one engineer to each with a mandate: build a working AI agent prototype in two weeks. Not a strategy. Not a POC document. Working software.

Wednesday: Call your ServiceNow/Salesforce/SAP account manager and ask for your contract renewal terms. Then ask your team what it would cost to replace that functionality with an agent-first architecture you own. You need both numbers to make real decisions.

Thursday: Look at your consultancy spend. How much are you paying for "AI strategy" that has not resulted in production deployments? Cut it. Redirect that budget to builders.

Friday: Read the news from this week again. ServiceNow saying everything is fine. India's outsourcing sector saying it will adapt. Accenture buying AI companies while selling you patience. Ask yourself: who benefits from me moving slowly?

The answer is never you.

The Honest Bit

I write these posts because I believe the biggest risk in tech right now is not moving too fast. It is the comfortable lie that the current structure holds. "Our business is doing great" is a factual statement about the present that tells you absolutely nothing about the future. Kodak's business was doing great in 2005. Nokia's business was doing great in 2006. The present is the worst predictor of the future when the substrate is shifting.

From Jönköping, I watch Swedish companies that have every advantage, brilliant engineers, stable markets, deep domain expertise, lose the AI race to companies half their size because of institutional inertia. It makes me angry. It does not have to be this way. Sweden can build. We have the talent. We have the infrastructure. What we need is the courage to stop asking permission and start shipping.

If you are building something real, I want to talk to you. If you are waiting for someone to tell you it is safe, I cannot help you. Nobody can. Because by the time it feels safe, it will be too late.

Fredrik Brunnberg is the CEO of HEIMLANDR.IO, building AI and software solutions from Jönköping, Sweden. This is the daily HEIMLANDR briefing. If you found this valuable, share it with someone who builds things.