Maybe he was using a mainstream LLM. That’s why we build finance infrastructure by fine-tuning open-source LLMs so that data doesn’t go outside the system. It’s very important for SEC-regulated companies to secure the data.
BREAKING: A Morgan Stanley banker accidentally leaked an internal document listing over 100 deals its Asia investment banking division is currently pitching and working on.
Recently had a meeting with a potential client for a SaaS project.
They are raising funds for the SaaS and want to start in January 2027.
They were asking me for the stack recommendations in the proposal.
But I declined saying this: “In this AI era, the market is moving very fast, and most of the stack relevant today can be obsolete when you start.”
We have been building SaaS for many years.
Previously, if we gave any stack recommendations, they used to be relevant even after 2-3 years.
But those days are gone.
Either you take a decision fast and build fast or become obsolete.
Did you guys experience something like this yet?
We sent our representative to an AI meetup event in London.
Most of them were AI agencies.
They are charging way more for a simple build.
We are undercharging compared to them.
Even most of them started new AI agencies and are doing really great just because of their physical presence in London.
I feel like:
- We need to send more representatives into the UK & US market
- We need to partner with companies present in the market physically.
What do you think?
@Codie_Sanchez Going from 0 to 1 is the difficult task. First you have to go from 0 to 1, then 1 to 10, and then 10 to 100. If you figure out the 0 to 1, then 1 to 10 and 100 become easier.
Yeah everyone can do a prompting on the AI tools like ChatGPT and Claude but everyone cannot build a working product or system using a basic prompt. AI is the enhancer: you give good input, you get great output. It only enhances what you have. If you don't have the domain knowledge of any industry and don't know about the workflow, you cannot build a good system. This is why we are still in the game and building AI systems for businesses.
@PrivatEquityGuy@OneManLBO Highly agree. Warm introduction is the best way to get the M&A deal but having a targeted system which you can follow regularly helps a lot. We are doing exactly that. We are building AI systems for M&A businesses and helping them do the targeted outreach.
Yes the legal help is the important part in the M&A process. You are working with many M&A businesses and build systems for them but through AI all the legal activities cannot be checked. You need a human in the loop and a good legal process to find out the fraudulent activity in the M&A process.
Fully agree, Luke. Recently, I talked with a potential client, and in the proposal he asked for the stack recommendations for his AI builds. He wants to start the project in January next year, and I told him that most of the stack will be obsolete at that time. We will focus on mapping out your process, finding out which problems can be automated with AI, and implementing that through the stack relevant at that time.
I fully agree with you on reusing the 80% and saving a lot of time in our agency. We target a specific niche, build out for that niche, and target other people in the niche, and just do that 20% customization and hand over.
It's never been easier to compare options than it is today.
AI search does it for people in seconds.
If you're not showing up there, you're not in the game.
Maintain these 3 things:
- Make yourself visible with AI SEO
- Use AI to filter unqualified leads
- Give a reply to qualified leads within 5 minutes
And, you will be ahead of 90% of business owners.
Thinking about implementing AI in your business?
You need to understand:
- Buying Vs Building
- Mainstream LLMs Vs Cheap LLMs
- Workflow Automation Vs Agent
Most of the businesses don’t know these differences.
And, they often make mistakes implementing AI.
@dan__rosenthal You're on point. Learned many new strategies from this post. We build GTM systems for companies, and this is what is working for our clients.
Calling all the M&A businesses out there.
What’s the most time-consuming task you are doing manually right now?
Recently, we are doing a lot of AI systems built for M&A businesses.
And I see most of the businesses doing the operations the old way.
A CEO turned down my AI implementation offer.
He had engineers. He had budget. But no idea which workflow can be automated.
He runs a community app for builders.
From the outside, hiring us to build looked like the obvious next step.
Inside, his real problem was older.
Nobody had mapped how work actually moves through his company.
So we did the boring work:
- Walked his workflow end to end
- Marked every spot people move information around
- Handed him a step-by-step plan
His team built it. Not us.
Most agencies call that a lost deal.
But, I consider it the right move.
AI consultancy matters more than implementation.
Don't rush to build.
Most teams already have implementers.
What they don't have is clarity.
If you're serious about AI, start with a real assessment by an expert.
It's the cheapest part. And it decides if the expensive part works.
Buy the map before you buy the machine.
Comment "MAP" and I'll send you the audit questions I use with every client.
China is going ahead in this AI game.
Kimi K3 is suppressing Claude's Opus 4.8 in many benchmarks.
Basically the power of chips, good model, GPU compute, all are powered by electricity.
China has everything.
China has figured out the way to get good output through less compute.
Imagine what they can do with more compute power.
China already produces more electricity than the US, India, and Europe combined.
They are increasing their electricity by 4x.
In the future China will be the leader in the AI space.
They are already competing with frontier models with less compute.
With more electricity, more GPU power, more compute power, they will produce more great models than the US models.
They're already doing really great in robotics.
The next digital AI and robotics is in the hands of China.
Yes, US can ban Chinese model's use in US businesses but it cannot suppress the use of Chinese models outside of US.
In coming years, AI will be more cost efficient for everyone.