More intelligence. Less artificial.

AI: “Here’s a finished answer, delivered with total confidence.”

Where does this come from?

Fredrik Malmström
Fredrik Malmström, GothenburgSenior Digital Strategist & Creative DirectorConsultant and freelancer, mostly within marketing and communication, with broad experience from startups, scaleups, agencies and large international corporations in RecTech, MarTech, MedTech, automotive and software.

What I do — the short version

A bit of a designer, strategist, product person and marketer. I help organisations grow, change and tell their story, and I use AI where it frees up time for people.

Part artist, part scientist

I’ve always had two sides. One is idealistic and human-centric: real conversations, people who feel lonely, trust that can’t be faked. The other is entrepreneurial and creative: building, testing, launching and making the numbers move. Good marketing needs both.

Algorithms → contentTransaction → relationshipArtificial → authentic

GHaaS: Growth Hacking as a Service

GHaaS is my model for growth. It’s a method, not a company: what matters is how it’s used. It connects what the data says about the past with a vision of the future, and balances strategy, quantity and quality.

The GHaaS model: three overlapping circles for strategy, quantity and quality, with insight, tactics, production and distribution in the four corners STRATEGY QUANTITY QUALITY DATA PAST VISION FUTURE 1. Insight 2. Tactics 3. Production 4. Distribution
How GHaaS works
  1. Insight. Read the data: what has happened, and where your customers actually are.
  2. Tactics. Turn insight into a strategy. Focus on a few initiatives that matter.
  3. Production. Quality: content with a story worth caring about.
  4. Distribution. Quantity: the right channels, enough reach, then follow up and optimise.

Every campaign runs the same loop: package, plan, produce, publish, optimise and evaluate.

Since 2012 I’ve built three marketing departments from scratch and launched hundreds of digital campaigns. At Humly, an app for substitute staffing, we went from about 50 candidates a month at €50 per acquisition in early 2018 to more than 600 a month at €3 in mid 2019.

Better to scatter and pour, than guard a single lonely core.

Tired of AI, and I use it every day (both true)

I’ve written more than a thousand prompts since 2024. I’ve built apps without knowing how to code, edited film with Runway and shipped with GitHub and Vercel. I’m pretty tired of AI, and pretty dependent on it. Both things are true.

More about me

I work with growth hacking, change management and concept development, and I’ve been building SaaS products since 2012. That’s more than 15 years in growth companies, across communication, content and digital strategy. Since 2023 I’ve moved between large and small organisations as a consultant, including as marketing strategist at Kyrkans SOS, a Swedish support line. The common thread is people gathering around something and getting a message out.

Four more things I’ve learned
  • An answer is not a source.A language model sounds just as sure when it’s wrong. Figures, names and quotes get checked against the source.
  • Some conversations shouldn’t be solved.Sometimes you want to be helped, sometimes heard. A text box can give an answer, but it can’t stay with you.
  • Story before message.Ninety percent story, ten percent message. What moves the heart opens the mind to what you want to say.
  • Unpolished is allowed.I document a lot, partly because my memory is bad. The craft is in rhythm, timing and telling what actually happened.
Technology can shorten the path from thought to action, but it can’t decide what’s worth caring about.
Use AI for what frees up time. Spend that time on something that matters to people.

Synthetic soul care

Use AI in moderation, and for the right things. Let it handle the reports, applications and bookkeeping. Be careful when it starts to stand in for care.

A reply that looks like care

When AI is used as emotional support, you get what I call synthetic soul care: an answer that resembles care, but with no one behind it who cares. The problem is that it can feel human. We’re so used to chatting that a text box feels like a conversation, whoever or whatever is answering. In the chats at Kyrkans SOS, one of the most common questions is: “Are you a robot?”

One night an ad for a depression app showed up in my feed. Its AI chat asked if I’d had thoughts of harming myself. I answered yes, to test it, and was referred to an emergency number. Maybe correct. But that’s where the conversation ended. Nobody asked a follow-up question. Nobody stayed.

Lowering the threshold, with a human on the other side

For many people, calling a support line is a big step. That’s why Kyrkans SOS built a chat where you still get human support, at a lower threshold than a phone call. Now we’re looking at ways to lower that threshold even further, and to build knowledge about synthetic soul care and involuntary loneliness. The question isn’t whether people should be able to write instead of call. It’s whether there’s a person on the other side.

What a screen doesn’t do

Few of us stop to think about what it means to spend time with a screen. It only answers when you write. It never misses you, never notices that you’ve gone quiet, never gets in touch. It can feel like company, but it doesn’t break loneliness. It risks deepening it.

Do you want to be helped, hugged or heard? AI can help. It can’t hug, and it can’t really hear.

The question comes from Charles Duhigg’s talk on better conversations, where he describes it as something teachers learn to ask.

An answer is not a source

This autumn, buses in Stockholm and Gothenburg carry the question “Would you rather ask Claude than a friend?” The sender, A Non Thinking Generation, presents itself as a non-profit working for a less thinking society. It’s satire from Läromedelsförfattarna, the Swedish textbook authors’ association. When I searched for the name, Google’s AI overview described it as a real non-profit, with a mission that doesn’t even match the campaign. A harmless example, but the mechanism is the same everywhere: an AI answer sounds just as sure when it’s wrong, and it doesn’t get irony.

Trust works the same way. In a study by Bynder, 52 percent of consumers said they disengage from content they suspect is AI-generated. Polaroid made a point of it when Meta’s AI suggested replacement ads, generated in a minute, for a campaign their studio had spent months on. The trust isn’t in the words. It’s in knowing a person stands behind them.

Need someone to talk to in Sweden? SOS-linjen: 0771-800 650. In an emergency, call 112.

What I’ve worked on

Get in touch

Got an idea that needs some pushback, an organisation growing too fast or a message that needs to get out? Write a few lines and I’ll get back to you.