Helping startup operators move faster with automation

I train business professionals on AI and automation: workshops, online courses, live sessions, all the way down to custom company training programs

Alexandre Kantjas

Claude Chat, Claude Cowork, and Claude Code run on the same model. What's different is the harness: 

And most people have no idea what that means. 

The harness is everything that sits around the model. 
→ The context it's given 
→ The tools it can access 
→ The permissions it holds 
→ The memory it carries. 

Imagine Claude is like a racing driver. 
Same person. Same skills. Same reflexes. 

Put them in a kart, a road car, or a Formula 1 car - and you get completely different results. Not because the driver changed, but because the machine around them changed. 

This is what the harness does in this case.

Back to our Claudes: so what does each harness actually do?

Claude Chat
This is the conversational harness. It gives the model chat history, system instructions, file uploads, web search, and lightweight project memory. It’s built for thinking, drafting, analyzing, and exploring ideas. You ask. It responds. Minimal permissions. No deep system access.

Claude Cowork
This harness adds an operational layer. The model can access a Linux VM, a sandboxed terminal, files, documents, browser automation, and structured projects. It can run commands, manipulate data, create artifacts, and orchestrate tasks. It doesn’t just think with you. It executes workflows.

Claude Code
This is the developer harness. The model gets direct access to a local codebase, terminal, Git, GitHub, and developer tools. It can read, modify, and refactor real files. It can commit changes. It can work across an entire repository. This is where the model becomes an embedded engineering assistant.

Claude Chat helps you think.
Claude Cowork helps you operate.
Claude Code helps you build.

Same models. Different harness. Different capabilities.

I was still wasting 10 minutes after every sales call. Here's how I cut it to one command in Attio.

Every (good) Sales call ends the same way: switch to CRM, type in the deal name, the amount, close date, link the contact, the company...

It's manual. It's slow. And it's no longer needed in 2026. 

Here's what I do now instead:
1. I start from call recording in Attio.
2. I open Ask Attio.
3. I paste this prompt:

Create a deal:
- Deal name: {Company} | {Service sold}
- Deal amount: {Amount}
- Deal owner: me
- Close Date: in two weeks
- Link the external attendees to the deal
- Link the company to the deal
- Suggest Channel Type and Channel detail based on Transcript

Ask Attio will:
→ Reads the call transcript
→ Fetch the linked records information
→ Gather additional context

It fills in every field - and lets you approve or edit before saving.

No switching between tabs. No manual typing. Done.

I find it's doing a great job for defining deal attribution values. We just ask people during calls how they've heard about us. Ask Attio will fetch this and combine it with CRM data and determine the channel(s).

Pro tip: you can save this text as a Prompt in Attio (this is what I did here) so all you have to do really is just pressing a command.

And if you're not using Attio, well, here's to hoping your CRM wakes up and brings baseline AI capabilities.

I personally won't go back to manual entry.

Oh my goodness: Notion Custom Agents are finally here! Here's why this is a big deal:

Notion AI was already among the top products last year for me - before this release.

By far and large the easiest and most versatile embedded AI agent out there built directly into your workspace where you actually work.

Notion AI could:
→ Make modifications to pages
→ Create entire databases from scratch
→ Handle complex, multi-steps tasks

Compare that to what Gemini or Microsoft Copilot can currently do: day and night.

And today, Notion Custom Agents take this to an even higher level.

Now you can build (agentic) 𝘸𝘰𝘳𝘬𝘧𝘭𝘰𝘸𝘴 with these AI capabilities.

This opens up a world of possibilities, literally every process relying on Notion documents and databases can now be automated.

If you are on Notion, and not using Notion AI, now's the time.

P.S: I'll be covering this release and showcasing use cases next week. Will announce the webinar + link to register tomorrow. 

Stay tuned!

It's official: pleased to announce our upcoming conference: Berlin AI Day 2026. September 15th. Festsaal Kreuzberg.

We've spent the last two years building a community of operators and leaders who don't just talk about AI. They build with it. They deploy it. They own the results.

This September, we're taking it further.

A full-day conference for the people actually driving AI adoption inside their organizations - not the ones still writing strategy decks about it.

Speakers. Case studies. Real systems. Real results.

And if you've been following what we've built with buildersklub, you know the room will be worth it.

Super Early Bird tickets are live right now at €149 - the lowest price they'll ever be, until April 15th.

After that, they go up.

Grab yours in the comments below 🎟️

#BerlinAIDay2026 #jointheklub

Notion Custom Agents are a game changer. Here are 99 Custom Agents across 11 categories you can implement in 5 minutes:

There are use cases for every function:
 → Productivity
 → Management
 → Marketing
 → Sales
 → Operations
 → Legal
 → People
 → Finance
 → Product
 → Support
 → Customer Success

A well-crafted prompt can one-shot a Notion agent. Wire up the right data, add a trigger, and it runs on its own. 

So I documented 99 of them.

Each template gives you:
→ The exact prompt to create the agent
→ The databases it should interact with
→ The connectors to set up

More importantly, this should give you ideas for agents that can automate your own processes. 

Try them and let me know!

---
P.S: Hi Notion! Hope you enjoy this resource I made. Feel free to share 🫶

99 Notion Custom Agents

Next week we're dropping the biggest resource we've ever made: AI Starter, a 4-hour course for businesses that want to leverage AI.

This is for anyone who keeps hearing about AI but doesn't know where to actually begin:

Not developers. Not data scientists. Not automation nerds.

Business operators. 

People running teams, processes, and companies who know AI matters but haven't found a clear entry point yet.

We built it because most AI content assumes you already know what you're doing. AI Starter assumes nothing. It gives you the foundations, the tools, and the first concrete steps - in one place.


Here's what's you'll learn:

→ Lesson 1: AI-First Operations
How to rethink the way you work when AI is at the core. Not theory. An actual operating model.

→ Lesson 2: The AI Tools Landscape
There are hundreds of AI tools. This lesson cut through the noise and shows you what's actually worth your time.

→ Lesson 3: AI-First Operations Patterns
Teaches you 15 patterns of automations, AI workflows, AI agents and internal tools that you can implement to automate your processes

4 hours. Entirely free. 

🔗 Link to sign up in the comments.

4,000 people laid off today at Block "because of AI". 

Announced less than 12 hours ago by Jack Dorsey on X.

"we're not making this decision because we're in trouble. our business is strong. gross profit continues to grow, we continue to serve more and more customers, and profitability is improving. but something has changed. we're already seeing that the intelligence tools we’re creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company. and that's accelerating rapidly."

Whether companies support them in this transition or not, this year everyone will need to learn to work proficiently with AI.

Update: Jack posted that Block overhired over the recent years and this is also part of the reason behind the adjustment he is making.

Thinking context dumping alone is the key to better AI results is lazy thinking.

And I get the idea is appealing: give AI more context and the outputs will get better.

Just dump everything:
Your notes.
Your documents.
Your meeting transcripts.
Your entire knowledge base.

And the AI will magically figure things out.

This is lazy thinking -  and you'll soon find out what happens when you do.

Turns out: context alone won't solve all your problems.

Once you've figured out context, you will find that something else becomes the bottleneck.

You.

You will be bombarded with decisions to make:
What outcome are we optimizing for?
What is “good enough”?
When do we automate vs. review?
Which outputs do we ignore?
Which signals actually matter?

If those decisions aren’t clear, adding more context just multiplies noise.

And so the real work in the AI era is building decision frameworks.

That means unpacking how work actually happens:
 – what good outputs look like
 – what tradeoffs matter
 – where humans should intervene
 – how systems should behave when things are uncertain

This is slower.

More complex.

And far less talked about.

But the teams that win with AI won’t be the ones with the most context.

They’ll be the ones with the clearest thinking.

Edit: swapped context engineering for context dumping after I got a very good comment hinting my post was actually implying context engineering best practices

Yesterday, Notion finally dropped Custom Agents. Next week, I'm running a live session to break it all down:

I've been waiting for this one. It's the kind of update that actually changes how you use the tool - not a UI tweak, not a new template. An entirely new way to build AI that lives inside your workspace and does real work.

I'll cover:
→ What Custom Agents actually are (and what they're not)
→ The building blocks: instructions, tools, permissions, triggers
→ How to pick a workflow worth automating
→ A full live build, end-to-end

No technical background needed.

60 minutes. 

Q&A at the end.

Free. 

Sign up here → https://luma.com/cf4vezot

OpenClaw is the hottest thing under the sun right now. So: hype or revolution?

Between the "AGI is here" crowd and the "security nightmare" warnings, it's hard to know what's actually true.

Jan's been testing OpenClaw over the last weeks. 

His verdict: it genuinely changes how you work. 

But only if you approach it the right way.

Tomorrow, he's breaking down everything you need to know in our next 9x live session.

𝐎𝐩𝐞𝐧𝐂𝐥𝐚𝐰 𝐟𝐨𝐫 𝐁𝐞𝐠𝐢𝐧𝐧𝐞𝐫𝐬
👥 𝐇𝐨𝐬𝐭𝐞𝐝 𝐛𝐲: Jan Meinecke, Co-Founder at 9x
🕒 𝐓𝐢𝐦𝐞: Tuesday, February 25, 2026, 00:00 CET / Monday, February 24, 18:00 EST
📍 𝐋𝐨𝐜𝐚𝐭𝐢𝐨𝐧: Online

𝐖𝐡𝐚𝐭 𝐲𝐨𝐮'𝐥𝐥 𝐥𝐞𝐚𝐫𝐧:
→ What OpenClaw is and how it's different from ChatGPT or Claude
→ Security considerations and how to approach them sensibly
→ When OpenClaw makes sense and when it doesn't
→ The four capabilities that make it genuinely useful
→ How to set it up step by step

𝐖𝐡𝐨 𝐭𝐡𝐢𝐬 𝐢𝐬 𝐟𝐨𝐫:

Anyone curious about OpenClaw but unsure where to start. Whether you've been following the hype or you're ready to set it up, this 90-minute session will get you there.

No technical background required.

𝐍𝐨𝐭𝐞: Jan is in Australia, so this session is outside our usual timeslot. If you're in Europe, the timing won't work live. Register anyway to get the recording and watch when it suits you.

See you there. Link to register: https://luma.com/fpv874pt

Nick Roco

Hi, I'm Nick, from Morningside AI.

We're a completely done-for-you AI workflow automation agency.

Where we help you:

→ Transform your business into an operational machine
→ Turn AI into a tool, not just a toy
→ Ultimately save time, headaches, and money

We do this by working with you 1-on-1 to find the biggest bottlenecks in your business today, and help you solve them with AI and automation.

Here’s how we do it:

1. Audit your current workflows.
2. Identify bottlenecks, inefficiencies, and repetitive tasks.
3. Verify if they can be outsourced to AI or automation.
4. Break down the task into steps.
5. Build custom AI automation tools to streamline.
6. Launch.

Simple as that.

What you get when you work with us:

→ Built-for-you AI and automation solutions
→ A full top-to-bottom AI Automation Business Review
→ Expert guidance on integrating AI into your business for maximum impact

Past clients have received:

→ An efficient AI System for customer support + lead generation.
→ Completely automated client onboarding system.
→ Infinite AI-enabled content creation.
→ Strategies focused on tangible outcomes (time savings, reduced overhead, and increased revenue).
→ Much more sleep.

Here's what 2 of our clients have had to say:

"Nick has taken out a lot of those pain points and a lot of those tedious day-to-day tasks that would feel very overwhelming... I can always count on him when it comes to anything AI and really automating anything" - Ryan, Marketing Manager @ Aragon

"We've been working together with Nick trying to figure out how we can integrate automation in making our workflow much more efficient... Not only did he help us figure out what to do with our workflow to make it more efficient, he also helped us develop what we have to make it better" - Syra, Marketing Project Manager @ Adcrate

----------

If any of this resonates, let's have a genuine conversation about your most annoying challenges.

Reach out to me directly, and let's explore how AI and automation can be a part of your success story.

I respond promptly and am here to help.

Or go here for a Free 16-page E-Guide on Ways to Use AI in your Business: https://tally.so/r/wMElKg

See ya soon.

Member

This is the exact ROI calculator we use to calculate ROI for AI automation projects.

We've used it to identify millions of dollars ($$$) in value across 53+ engagements in 11 industries.

And now, we're giving it away.

It includes:

✅ The exact A + B + C formula - including the 2 parts most people never calculate

✅ 3 real world ROI examples (RFP generation, speed-to-lead, meeting prep) with actual numbers filled in

✅ 3 blank tabs ready to go - plug in your own numbers and get a dollar figure in minutes

✅ Step-by-step instructions so you can do it yourself

Most ROI calculations stop at time saved. This one goes 3 dimensions deeper.

To get it:

1. Connect with me
2. Comment "ROI" below
3. If you repost, I’ll send it to you first

Happy calculating.

---

P.S. Works whether you're a business owner building the internal case OR a service provider selling AI to clients.

Head of AI Consulting @ Morningside AI | Sharing real AI transformation strategies from the front lines | Launched 53 AI projects for clients (and counting)

Just wrapped our first Morningside AI webinar. Man that was fun.

Topic was: our 3-phase process that we use to identify, prioritize, and prototype opportunities.

AND we gave away our AI ROI calculator template - the one we use to help identify millions of dollars in value from AI for our clients.

If you missed it, no biggie - you can sign up below, and we'll send you the recording.

https://lnkd.in/gVMKxXbp

And if you don't want to miss the next one, shoot me a DM and I'll automatically add you to the list.

Every company is buying AI agents. Almost nobody is teaching their people how to work with them.

I honestly think we're watching the same gap that appeared with spreadsheets in the '90s. The tool showed up. Training didn't. And for years, most workers used Excel like a glorified calculator.

The same thing is happening with AI agents right now.

I'm calling this skill “Agentic Fluency” - the ability to delegate to, manage, and multithread across AI agents as part of your daily work.

This is NOT just prompt engineering. It's not honestly not even necessarily technical. It's operational.

Let me explain:

Consider the CSM.

Think about what a strong CSM does during renewal season. They're pulling account history, drafting outreach, prepping risk assessments, updating the CRM - all in parallel. Agentic Fluency means knowing how to hand 50-80% of that to agents, while the CSM focuses on the judgment calls that actually save the account.

Same for an AE doing account research. The agent handles the data pull, the competitive landscape, the first-pass account plan. The (human) AE shapes strategy and owns the relationship.

The MAJOR SHIFT HERE is ensuring that your CSMs and AEs all have the wherewithal and the skillset to pull this off.

And for the AI maxi’s: Yes, the tooling will keep getting easier. Interfaces will abstract away much of the complexity - you won't need to know what folder you're in or which tools are connected. The software will eventually handle all that.

(Exhibit A: Clawdbot)

But the core skill? Managing multiple agents, knowing what to delegate, when to intervene, how to quality-check output across parallel workstreams?

That doesn't get abstracted. It compounds.

The companies building this fluency now will operate at a fundamentally different speed in 18 months. The ones waiting for the tools to "get easy enough" will be playing catch-up - again lol.

Agentic Fluency is the spreadsheet literacy of the AI era. The question isn't whether your team needs it… it's whether they are building the skills fast enough.

I don't subscribe to the whole "AI is replacing SaaS" narrative right now.

There are a ton of reasons why (and honestly, smarter people than me have written about this at length already).

However… Claude Code (and vibe coding generally) can get you INCREDIBLY far for a very large amount of software applications.

Let me explain.

Last night, with Claude Code, I vibe coded an entire feature prioritization survey using MaxDiff methodology.

Quick context: MaxDiff is a specific survey approach designed to force trade-offs from respondents. Instead of "rate these 1-5" (where everyone gives everything a 4 and you learn nothing), it makes people choose between subsets - revealing what they ACTUALLY value.

It's one of the best ways to build a software roadmap based on real user signal.

Issue is - this isn't something you can just whip up with a Google Form or Tally. The survey needs to randomize option sets, present balanced combinations, and the analysis on the back end is fairly complex.

I remember in my past life paying Qualtrics or other niche survey companies $10-15K to access this kind of functionality.

Last night I built the whole thing in about 4 hours.

1. Requirements doc written in the first hour
2. Code written by hour two
3. Tested completely by hour three
4. Deployed via Railway by hour four

It looks exactly how I want it. Custom UI, custom flow, custom everything. It’s… beautiful.

So here's my actual take on "AI eating software" -

It's not coming for your systems of record. Not your CRM, not your ERP, not the core platforms that run your business.

Where it IS eating software is these niche, tangential applications that you need spun up. The tools that sit adjacent to your core work. The ones where you used to have two options: pay a premium vendor or just go without.

THAT is the version of "AI replacing SaaS" I'm here for.

Being able to go from idea → requirements → code → deployed in a single evening is wild.

And honestly, it's these kinds of use cases that make me most excited about where this is all heading.

---

P.S. I'm Nick. I post about what we're learning on the front lines of AI transformation - including when AI isn't the answer.

If you're interested, follow along.

taplio

The Battle

Alexandre Kantjas

Nick Roco

Followers

Posts

Engagements

Likes

Performance ScoreAlexandre Kantjas's LinkedIn Performance

Performance ScoreNick Roco's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Alexandre Kantjas

Nick Roco

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Alexandre Kantjas's LinkedIn Performance

Performance ScoreNick Roco's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreAlexandre Kantjas's LinkedIn Performance