
AI strategy
Prototyping
Innovation
5 Must-Know Truths Before You Dive Into AI
May 13, 2025
SmplCo Team
The great AI gold rush is on. It has become an indispensable part of new tech, while those with existing products and services are rushing to integrate.
But there are many traps for the unwary. Luckily AI experts Jesse Lord and Ole Anders Andersen, from our partners Fabriq AI, are here to steer you around those traps.
So, before you take another step towards AI, take these lessons on board.
1. You Might Not Need AI (Yet)
There’s too much noise in the AI space.
Everyone from start-ups to big corporates is piling in without asking the most basic question: “Do we actually need this?”
Yes, AI will eventually be part of most businesses. But that doesn’t mean now is the time for you. Some businesses don’t need AI. Some don’t even need a laptop.
Use the same filter you’d apply to any investment. That means asking:
Will this move the needle for your goals, your customers, or your bottom line?
If it won’t, AI is not for you (right now at least).

The Fabriq team: Johannes Aasheim, Ole Anders Andersen & Jesse Lord
1. You Might Not Need AI (Yet)
There’s too much noise in the AI space.
Everyone from start-ups to big corporates is piling in without asking the most basic question: “Do we actually need this?”
Yes, AI will eventually be part of most businesses. But that doesn’t mean now is the time for you. Some businesses don’t need AI. Some don’t even need a laptop.
Use the same filter you’d apply to any investment. That means asking:
Will this move the needle for your goals, your customers, or your bottom line?
If it won’t, AI is not for you (right now at least).

The Fabriq team: Johannes Aasheim, Ole Anders Andersen & Jesse Lord
1. You Might Not Need AI (Yet)
There’s too much noise in the AI space.
Everyone from start-ups to big corporates is piling in without asking the most basic question: “Do we actually need this?”
Yes, AI will eventually be part of most businesses. But that doesn’t mean now is the time for you. Some businesses don’t need AI. Some don’t even need a laptop.
Use the same filter you’d apply to any investment. That means asking:
Will this move the needle for your goals, your customers, or your bottom line?
If it won’t, AI is not for you (right now at least).

The Fabriq team: Johannes Aasheim, Ole Anders Andersen & Jesse Lord
1. You Might Not Need AI (Yet)
There’s too much noise in the AI space.
Everyone from start-ups to big corporates is piling in without asking the most basic question: “Do we actually need this?”
Yes, AI will eventually be part of most businesses. But that doesn’t mean now is the time for you. Some businesses don’t need AI. Some don’t even need a laptop.
Use the same filter you’d apply to any investment. That means asking:
Will this move the needle for your goals, your customers, or your bottom line?
If it won’t, AI is not for you (right now at least).

The Fabriq team: Johannes Aasheim, Ole Anders Andersen & Jesse Lord
2. Ask the Only Question That Matters
If you’re considering AI, start here:
“Where and how will this create real value – for us, for our customers, and for the business long-term?”
This isn’t about shiny tech. It’s about outcomes.
The hype cycle’s over. What’s left are lessons from people who wasted time and money chasing novelty. Don’t be them.
Build a quick prototype. Test if AI actually delivers what you need.
Make sure you’re solving real problems and that the complexity of the project is justified by clear, measurable value — not just a vague sense that “we should be doing AI.”
2. Ask the Only Question That Matters
If you’re considering AI, start here:
“Where and how will this create real value – for us, for our customers, and for the business long-term?”
This isn’t about shiny tech. It’s about outcomes.
The hype cycle’s over. What’s left are lessons from people who wasted time and money chasing novelty. Don’t be them.
Build a quick prototype. Test if AI actually delivers what you need.
Make sure you’re solving real problems and that the complexity of the project is justified by clear, measurable value — not just a vague sense that “we should be doing AI.”
2. Ask the Only Question That Matters
If you’re considering AI, start here:
“Where and how will this create real value – for us, for our customers, and for the business long-term?”
This isn’t about shiny tech. It’s about outcomes.
The hype cycle’s over. What’s left are lessons from people who wasted time and money chasing novelty. Don’t be them.
Build a quick prototype. Test if AI actually delivers what you need.
Make sure you’re solving real problems and that the complexity of the project is justified by clear, measurable value — not just a vague sense that “we should be doing AI.”
2. Ask the Only Question That Matters
If you’re considering AI, start here:
“Where and how will this create real value – for us, for our customers, and for the business long-term?”
This isn’t about shiny tech. It’s about outcomes.
The hype cycle’s over. What’s left are lessons from people who wasted time and money chasing novelty. Don’t be them.
Build a quick prototype. Test if AI actually delivers what you need.
Make sure you’re solving real problems and that the complexity of the project is justified by clear, measurable value — not just a vague sense that “we should be doing AI.”
3. The other 'only question that matters'
Once you’ve answered the question, ‘Can we get value out of AI?’ – and found the answer is ‘yes’ – it’s time for a reality check.
Unless you’re really lucky, you’ll only have so many people and so much resource at your disposal.
If you rush into solving everything at once, you’ll create a monster that will cost you endless amounts of time, money, and stress.
At best your supporters will lose interest. At worst you’ll make them very angry.
Instead, map the process, starting with what AI solutions balance the lowest complexity with the highest value for shareholders and stakeholders.
How one Fabriq client found value in AI
Lie of the land
We’d recommend doing a feasibility study that covers:
A landscape review, which looks at stakeholder needs, current assets / approaches, and front-end requirements
A use case definition, prioritising business goals and documenting the use cases for different AI methods and models
Systems & integration mapping, asking what current systems exist and how connections are going to be made
Out of this you’ll get a draft platform design that will let you prioritise your focus, understand the technical implications of that, and scope the work.
Doing a feasibility study is particularly important if you’re in a larger organisation with multiple stakeholders and is essential to keeping the scope grounded, the priorities clear, and any "shiny-object syndrome" your senior execs may suffer, in check.
3. The other 'only question that matters'
Once you’ve answered the question, ‘Can we get value out of AI?’ – and found the answer is ‘yes’ – it’s time for a reality check.
Unless you’re really lucky, you’ll only have so many people and so much resource at your disposal.
If you rush into solving everything at once, you’ll create a monster that will cost you endless amounts of time, money, and stress.
At best your supporters will lose interest. At worst you’ll make them very angry.
Instead, map the process, starting with what AI solutions balance the lowest complexity with the highest value for shareholders and stakeholders.
How one Fabriq client found value in AI
Lie of the land
We’d recommend doing a feasibility study that covers:
A landscape review, which looks at stakeholder needs, current assets / approaches, and front-end requirements
A use case definition, prioritising business goals and documenting the use cases for different AI methods and models
Systems & integration mapping, asking what current systems exist and how connections are going to be made
Out of this you’ll get a draft platform design that will let you prioritise your focus, understand the technical implications of that, and scope the work.
Doing a feasibility study is particularly important if you’re in a larger organisation with multiple stakeholders and is essential to keeping the scope grounded, the priorities clear, and any "shiny-object syndrome" your senior execs may suffer, in check.
3. The other 'only question that matters'
Once you’ve answered the question, ‘Can we get value out of AI?’ – and found the answer is ‘yes’ – it’s time for a reality check.
Unless you’re really lucky, you’ll only have so many people and so much resource at your disposal.
If you rush into solving everything at once, you’ll create a monster that will cost you endless amounts of time, money, and stress.
At best your supporters will lose interest. At worst you’ll make them very angry.
Instead, map the process, starting with what AI solutions balance the lowest complexity with the highest value for shareholders and stakeholders.
How one Fabriq client found value in AI
Lie of the land
We’d recommend doing a feasibility study that covers:
A landscape review, which looks at stakeholder needs, current assets / approaches, and front-end requirements
A use case definition, prioritising business goals and documenting the use cases for different AI methods and models
Systems & integration mapping, asking what current systems exist and how connections are going to be made
Out of this you’ll get a draft platform design that will let you prioritise your focus, understand the technical implications of that, and scope the work.
Doing a feasibility study is particularly important if you’re in a larger organisation with multiple stakeholders and is essential to keeping the scope grounded, the priorities clear, and any "shiny-object syndrome" your senior execs may suffer, in check.
3. The other 'only question that matters'
Once you’ve answered the question, ‘Can we get value out of AI?’ – and found the answer is ‘yes’ – it’s time for a reality check.
Unless you’re really lucky, you’ll only have so many people and so much resource at your disposal.
If you rush into solving everything at once, you’ll create a monster that will cost you endless amounts of time, money, and stress.
At best your supporters will lose interest. At worst you’ll make them very angry.
Instead, map the process, starting with what AI solutions balance the lowest complexity with the highest value for shareholders and stakeholders.
How one Fabriq client found value in AI
Lie of the land
We’d recommend doing a feasibility study that covers:
A landscape review, which looks at stakeholder needs, current assets / approaches, and front-end requirements
A use case definition, prioritising business goals and documenting the use cases for different AI methods and models
Systems & integration mapping, asking what current systems exist and how connections are going to be made
Out of this you’ll get a draft platform design that will let you prioritise your focus, understand the technical implications of that, and scope the work.
Doing a feasibility study is particularly important if you’re in a larger organisation with multiple stakeholders and is essential to keeping the scope grounded, the priorities clear, and any "shiny-object syndrome" your senior execs may suffer, in check.
4. This AIn't a quick or easy fix
Open AI has a lot to answer for here. Yes, it is very impressive, but it has made everyone think that ‘AI is easy’.
The company’s Chat GPT model has made certain tasks trivial – often ones that were difficult to do well before.
This model type - Generative AI - has amazing applications. Staff at call centres, for example, used to spend so much time summarising what was said on their calls. All that can now be easily automated using AI, providing loads of value to those businesses.
But, in most cases, the AI solution you’ll need will not just come out-of-the-box if you’re going beyond the most basic tasks.
Even Open AI’s market-leading – and extremely user-friendly – product won’t give you what you need to go to market with a truly valuable solution of your own.

How AI dev meetings really look!
Interesting times
Chat GPT will give you around 60% of what you need easily. But the 60 - 80% stage starts getting difficult and expensive. Want to go 80 - 100%? Strap in and hold on tight.
In this Gen-AI example, the first 60% is easy because you’re probably trawling the internet for publicly available information.
But when you start trying to access information and data that is not accessible to the web, things become, ahem, interesting. (Not least because you’re moving into using more than one model.)
Then, if you want AI models to access your own processes or behave in accordance with your requirements (where real value accrues), well, that’s when things get really interesting.
Like so many other situations in life, creating something truly valuable takes time and money and hard work. Integrating and using AI is no different.
4. This AIn't a quick or easy fix
Open AI has a lot to answer for here. Yes, it is very impressive, but it has made everyone think that ‘AI is easy’.
The company’s Chat GPT model has made certain tasks trivial – often ones that were difficult to do well before.
This model type - Generative AI - has amazing applications. Staff at call centres, for example, used to spend so much time summarising what was said on their calls. All that can now be easily automated using AI, providing loads of value to those businesses.
But, in most cases, the AI solution you’ll need will not just come out-of-the-box if you’re going beyond the most basic tasks.
Even Open AI’s market-leading – and extremely user-friendly – product won’t give you what you need to go to market with a truly valuable solution of your own.

How AI dev meetings really look!
Interesting times
Chat GPT will give you around 60% of what you need easily. But the 60 - 80% stage starts getting difficult and expensive. Want to go 80 - 100%? Strap in and hold on tight.
In this Gen-AI example, the first 60% is easy because you’re probably trawling the internet for publicly available information.
But when you start trying to access information and data that is not accessible to the web, things become, ahem, interesting. (Not least because you’re moving into using more than one model.)
Then, if you want AI models to access your own processes or behave in accordance with your requirements (where real value accrues), well, that’s when things get really interesting.
Like so many other situations in life, creating something truly valuable takes time and money and hard work. Integrating and using AI is no different.
4. This AIn't a quick or easy fix
Open AI has a lot to answer for here. Yes, it is very impressive, but it has made everyone think that ‘AI is easy’.
The company’s Chat GPT model has made certain tasks trivial – often ones that were difficult to do well before.
This model type - Generative AI - has amazing applications. Staff at call centres, for example, used to spend so much time summarising what was said on their calls. All that can now be easily automated using AI, providing loads of value to those businesses.
But, in most cases, the AI solution you’ll need will not just come out-of-the-box if you’re going beyond the most basic tasks.
Even Open AI’s market-leading – and extremely user-friendly – product won’t give you what you need to go to market with a truly valuable solution of your own.

How AI dev meetings really look!
Interesting times
Chat GPT will give you around 60% of what you need easily. But the 60 - 80% stage starts getting difficult and expensive. Want to go 80 - 100%? Strap in and hold on tight.
In this Gen-AI example, the first 60% is easy because you’re probably trawling the internet for publicly available information.
But when you start trying to access information and data that is not accessible to the web, things become, ahem, interesting. (Not least because you’re moving into using more than one model.)
Then, if you want AI models to access your own processes or behave in accordance with your requirements (where real value accrues), well, that’s when things get really interesting.
Like so many other situations in life, creating something truly valuable takes time and money and hard work. Integrating and using AI is no different.
4. This AIn't a quick or easy fix
Open AI has a lot to answer for here. Yes, it is very impressive, but it has made everyone think that ‘AI is easy’.
The company’s Chat GPT model has made certain tasks trivial – often ones that were difficult to do well before.
This model type - Generative AI - has amazing applications. Staff at call centres, for example, used to spend so much time summarising what was said on their calls. All that can now be easily automated using AI, providing loads of value to those businesses.
But, in most cases, the AI solution you’ll need will not just come out-of-the-box if you’re going beyond the most basic tasks.
Even Open AI’s market-leading – and extremely user-friendly – product won’t give you what you need to go to market with a truly valuable solution of your own.

How AI dev meetings really look!
Interesting times
Chat GPT will give you around 60% of what you need easily. But the 60 - 80% stage starts getting difficult and expensive. Want to go 80 - 100%? Strap in and hold on tight.
In this Gen-AI example, the first 60% is easy because you’re probably trawling the internet for publicly available information.
But when you start trying to access information and data that is not accessible to the web, things become, ahem, interesting. (Not least because you’re moving into using more than one model.)
Then, if you want AI models to access your own processes or behave in accordance with your requirements (where real value accrues), well, that’s when things get really interesting.
Like so many other situations in life, creating something truly valuable takes time and money and hard work. Integrating and using AI is no different.
5. Centralise now, or pay later
This one’s for enterprises – and any start-up aiming to work with them.
AI efforts need a centre of gravity. Without it, teams chase the tool of the week, duplicate work, and drift away from business priorities.
One month it’s OpenAI. The next it’s Claude. Then it's DeepSeek, Mistral, or something no one’s heard of yet.
This can be understandable; the pace of change in the large language model (LLM) space is relentless. Betting everything on one vendor or model can leave you locked into tech that’s obsolete six months later.
You don’t need central control over every experiment. But you do need a clear place where strategy is set, standards are enforced, and efforts are tracked.
That’s how you make sure AI work stays aligned with actual business value – not just the latest hype cycle.
Without that, you're building on sand.
5. Centralise now, or pay later
This one’s for enterprises – and any start-up aiming to work with them.
AI efforts need a centre of gravity. Without it, teams chase the tool of the week, duplicate work, and drift away from business priorities.
One month it’s OpenAI. The next it’s Claude. Then it's DeepSeek, Mistral, or something no one’s heard of yet.
This can be understandable; the pace of change in the large language model (LLM) space is relentless. Betting everything on one vendor or model can leave you locked into tech that’s obsolete six months later.
You don’t need central control over every experiment. But you do need a clear place where strategy is set, standards are enforced, and efforts are tracked.
That’s how you make sure AI work stays aligned with actual business value – not just the latest hype cycle.
Without that, you're building on sand.
5. Centralise now, or pay later
This one’s for enterprises – and any start-up aiming to work with them.
AI efforts need a centre of gravity. Without it, teams chase the tool of the week, duplicate work, and drift away from business priorities.
One month it’s OpenAI. The next it’s Claude. Then it's DeepSeek, Mistral, or something no one’s heard of yet.
This can be understandable; the pace of change in the large language model (LLM) space is relentless. Betting everything on one vendor or model can leave you locked into tech that’s obsolete six months later.
You don’t need central control over every experiment. But you do need a clear place where strategy is set, standards are enforced, and efforts are tracked.
That’s how you make sure AI work stays aligned with actual business value – not just the latest hype cycle.
Without that, you're building on sand.
5. Centralise now, or pay later
This one’s for enterprises – and any start-up aiming to work with them.
AI efforts need a centre of gravity. Without it, teams chase the tool of the week, duplicate work, and drift away from business priorities.
One month it’s OpenAI. The next it’s Claude. Then it's DeepSeek, Mistral, or something no one’s heard of yet.
This can be understandable; the pace of change in the large language model (LLM) space is relentless. Betting everything on one vendor or model can leave you locked into tech that’s obsolete six months later.
You don’t need central control over every experiment. But you do need a clear place where strategy is set, standards are enforced, and efforts are tracked.
That’s how you make sure AI work stays aligned with actual business value – not just the latest hype cycle.
Without that, you're building on sand.
Next steps
If you want to find out how to nail your AI strategy, take advantage of a free, zero obligation, consultation with our experts.
Either click here or drop our Managing Partner Andreas a line directly.
You can also contact the Fabriq team directly, here. They'll help you find the best fit between your ambition, budget and emerging tech.
Next steps
If you want to find out how to nail your AI strategy, take advantage of a free, zero obligation, consultation with our experts.
Either click here or drop our Managing Partner Andreas a line directly.
You can also contact the Fabriq team directly, here. They'll help you find the best fit between your ambition, budget and emerging tech.
Next steps
If you want to find out how to nail your AI strategy, take advantage of a free, zero obligation, consultation with our experts.
Either click here or drop our Managing Partner Andreas a line directly.
You can also contact the Fabriq team directly, here. They'll help you find the best fit between your ambition, budget and emerging tech.
Next steps
If you want to find out how to nail your AI strategy, take advantage of a free, zero obligation, consultation with our experts.
Either click here or drop our Managing Partner Andreas a line directly.
You can also contact the Fabriq team directly, here. They'll help you find the best fit between your ambition, budget and emerging tech.