Alex Northstar

I’ve been doing this for almost 3 years, made 6 figures, what you should know: - SMALL businesses will NOT pay you 2k to be their AI guy - you CAN extract 2-3k value with a bunch of automations - you want to be aiming for 3-10M/year businesses - you need clarity with business owner and clear ROI - they do not care about AI - some of their employees will be resistant to the change - do NOT AUTOMATE SALES - they just need more leads and speed to lead Ask me more questions. Gurus yapping on podcasts are delusional.

PwC is spending $100M to teach people to THINK. 3 things YOU should do: They are spending more than $100 million to teach 85,000 people how to think. Starting in December, every partner and employee across PwC US, Mexico and their Acceleration Centers goes through a Critical Thinking and Judgment certificate. Completion target: June 2027. It starts with a diagnostic of how each person thinks. Then a framework, masterclasses, and personalized simulations. One example: a tax professional gets an AI-generated analysis of a complex tax position. It looks convincing. But some of the supporting information is incomplete. The job is to question the sources, consider other readings, and work out what's missing before signing off. Another: review an AI-generated assessment of a company's internal controls, find the deficiencies, and challenge the tool's conclusions. they're training people to catch bad AI For 2 years, most companies I worked with treated AI adoption as an access problem. So they buy licenses, run a prompting workshop and measure usage. I've trained more than 10,000 people on AI at SMBs and Fortune 500 in the last 3 years and I have to agree with pwc… this stuff is needed. When writing a report took a week, you thought about it along the way. When it takes 40 seconds, the thinking has to be a deliberate step, or it doesn't happen. PwC also said that they started their AI learning push with entry-level hires, because those jobs change first, and they pair a human skill with every technical skill. The junior work that used to teach judgment is exactly the work AI now does. If you don't replace that training on purpose, you raise a generation of reviewers who never learned what wrong looks like. What a founder can learn from this and what I’m doing now for my clients: 1. Run a weekly "find the error" drill. Take one real AI output from your business. A proposal, a report, an analysis. Plant one mistake in it, or find one that's already there. Give it to the team. Whoever finds it explains how they spotted it. Fifteen minutes. That's a simulation program. 2. Add 3 questions before anything AI-made goes out. What would have to be true for this to be wrong? Where did this number or claim come from? Could I defend every line of this to a client, out loud? 3. Make the reviewer a named person. Not "the team checks it." One name per output. When everyone is responsible for checking, nobody does. Every company now has access to the same models. What's left is whether your people can tell a good answer from a convincing one. PwC is betting $100 million that this is the skill that separates firms in the next 5 years. I think they're right, it’s something I already added in my AI productivity trainings this year. If your team is already using AI for proposals, research, reports, or client deliverables, DM me “CHECK”. I’ll show you a simple way to set up an AI quality-control process for your team.

“AI will take all jobs!” it’s the messaging I keep hearing… but you have a choice I’ve been an “AI consultant” for 3 years, and my business is NOT tech, it’s taking care of people, I’ll explain “AI will take all jobs!” is not completely wrong but it sounds disempowering, like people don’t have a chance. This makes no sense to me. I used to wash frying pans for $3/hour and there was 1 thing that actually saved my life: the power of education, of being curious, learning new things, upskilling and being useful to other people the tech folks keep forgetting that they’re ultimately selling to people, to humans, not to AI agents… those are just tools. So as AI consultant your job shouldn’t be to shove this technology down peoples throats. It should be to facilitate an improvement of their lives when I do my workshops for small businesses and also midmarket companies the whole idea is giving their people the opportunity to be smarter, better, and faster so that they’re the best in their fields. It makes much more sense to me. If you master AI agents you could potentially just achieve much more. You have another skill in your toolkit. And our Copilot + Claude trainings feedback shows that people become at least 20% more productive. I’m just thinking out loud on my walk with the dogs… Am I wrong? Shouldn’t AI consultants help people put AI to work instead of just selling licences?

NEW Fable 5.2 and Mythos 5.2! they are a step above Opus 5.5 in almost all the capabilities no one will talk about this but it's MASSIVE: AI controlling robots better and better. we are ... accelerating!

AI is like someone who read 1000 books about love. But only a person can experience love. After working for 4 years with it, that’s how I would describe it.

why?

Germany adds tax on AI productivity So I’m flying to Canada For real. If smart people work hard and smart and only get taxed more for it in EU, it’s a lose lose business with garbage incentives. France adds tax on internet traffic, you watch a Netflix show, you pay €3.13 Italy adds exit tax to squeeze more if you don’t like 40% business taxes. So I pay €70k to leave instead of buying my mom a house. EU produces 0 AI models yet squeezes the hardest taxes out of me. Yeah just no thanks. Let me know if you’re in Vancouver :)

so with Fable 5.5 you can now make Arcane season 3

no way.

I am leaving Europe if they tax me for Internet traffic and AI usage.

this is VERY valuable:

New Gemini Carbon! "significantly improves on usability in domains like coding" can someone confirm this? better than Opus 5.5?

this is just INCREDIBLE

How to get value from AI: - know what you want - know what makes money - communicate clearly

If I could have Opus 5.5* in the backend of Grok Bots, it would be the best AI assistant hands down, the end. *I mean the original good Opus 5.5, not the quantised version they released today, rip

Current Claude situation:

The Internet is getting hacked today, 12 rules to stay AI safe:

The internet is under attack today here's what to check in your company before Monday: Every AI tool your team connected this year holds a key to your business. like ChatGPT workspace, Zapier flows, the agent that drafts emails, the assistant that reads your drive. each one is a door and most companies have no idea how many doors they've opened. What changed AI can now find weak spots in software and break in on its own, for about $50 a break-in. This week brought a PHP flaw that touches most websites, a cloud platform warning that customers' secret keys may have been accessed, and US banks finding AI agents inside customer accounts that they can't identify. Attackers only need one way in and small companies like yours are the easiest targets because nobody is watching. Your real risk is not some genius hacker. It's the API key someone pasted into a chat in March, the agent with permission to send email from the CEO's account, and the integration an ex-contractor set up and nobody removed. I've trained 10,000+ people on AI and tested 500+ tools in the last 3 years. The pattern I often see is that teams connect fast and never audit! this is what we now do in month 1 of working together here's my 30-minute check - List every AI tool connected to email, the drive, the CRM, or payments. - ask your team, because they've connected more than you think. - Rotate every API key that was ever pasted into a chat, a doc, or a Slack message. - Remove any access nobody has used in 30 days, including old contractors' and tools you stopped paying for. - Require a human yes before any agent sends, pays, or deletes. Read and draft is fine. Act needs approval. - Turn on two-step login NOW for every admin account, starting with email. - Ask whoever runs your website whether it's patched for the PHP flaw. Get a yes or a date to fix this week What to tell your team today "Quick security check: this week, please don't paste passwords or API keys into any AI tool, and send me a list of every AI app you've connected to work accounts. If anyone asks you for a login code by phone, text, or email, it's a scam. Hang up and tell me." before panicking: AI still makes your team faster, so don't rip it out. Just treat every connection like you'd treat a new hire with keys to the office. Know who has access, to what, and why. Send this to whoever set up the AI tools in your company, DM me if you need help or the AI safety audit.