Field Guide AI

Helping professionals who don’t write code understand and use AI to become 100x more valuable.

[Field Guide AI @FieldGuideAI]

I’ve spent the last 5 years grinding in tech. I’ve seen the good, the bad, and the ugly, but nothing to this scale.

As a fellow Gen Z, I entered the workforce as AI tools were starting to proliferate. I’ve watched people of all ages react in two ways:

• Some treat AI like magic
• Others treat it like a threat

These two gut reactions are dead wrong.

AI will 10x your productivity and it will create even more jobs (contrary to popular opinion).

Unfortunately, most of what’s online is just as unhelpful (listed in order):

1. garbage prompt dumps

2. useless hype threads

3. screenshots with no context

4. Unnecessary fear posts.

This style of content is designed to optimize for one metric - engagement.

It’s not helpful at all. Honestly, it’s a joke and a waste of time for those with a job to do and a family to provide for.

That’s not what this account is for.

@FieldGuideAI exists because the future of work is going to look very different.

And most professionals are left with fluff or complicated engineering lectures. There’s almost nothing in the middle.

I CARE about the middle. The middle is where 99% of the world operates in, including myself.

I’m not here to make you an engineer (I am not one). I am here to help you stay confident and competent as these tools change under our feet.

If you want less noise and are curious to learn more about AI, then follow along.

Fair warning. I will be unapologetic in my approach.

[Field Guide AI @FieldGuideAI]

@AbrahamFriesse @AnkanXplorer Which one is your favorite?

[Field Guide AI @FieldGuideAI]

@lozatmamo It truly is a black box, makes no sense haha

[Field Guide AI @FieldGuideAI]

@paulbeneteauu @AnkanXplorer Great question, right now I have been building multiple different Grok Bots to help me streamline daily workflows (filter emails, draft emails, news digest, meeting minutes, etc). And my team is heavy users of Cursor (but that was an enterprise decision).

[Field Guide AI @FieldGuideAI]

@mayuri_3015 I personally am a heavy user of Azure, but to be honest, Google Cloud is a great, great product.

[Field Guide AI @FieldGuideAI]

@MansiCodez Yes, software engineers will always be necessary because often the code output isn’t clean and needs a lot of rework (for now). It will get better over time and then cracked SWEs will become FDEs/Solution Architects that help customers.

[Field Guide AI @FieldGuideAI]

🚨FIELD NOTE 01: Your model can’t spell because it doesn’t see letters.

Before it even reaches the model, your text gets chopped into bit sized tokens.

For example, “strawberry” might arrive as:

str + aw + berry.

Three opaque token IDs.

Asking the model to count the letter “R” is like asking you to count the letters in a phone number you only heard once.

It doesn’t work like that and this also explains the rest.

→ bad arithmetic on long numbers

→ weird behavior on rare names

→ why “ hello” and “hello” produce different outputs

This is not a reasoning failure by the model. It’s simply a vision problem.

[Field Guide AI @FieldGuideAI]

@anah_sahh @X New account, but helping professionals who don’t code understand AI, use it, and become 100x more valuable. Excited to meet you all.

[Field Guide AI @FieldGuideAI]

@YashHustle_22 Selling. Distribution is the most important moat as AI collapses coding/product development. That’s just me though. I am sure others have different views, but distribution is the hardest aspect of building a business.

[Field Guide AI @FieldGuideAI]

@AnkanXplorer It’s freaking great. I am honestly so impressed with how much they have caught up. And Cursor is fantastic.

[Field Guide AI @FieldGuideAI]

@DavidChat18396 People wined when railroads were being built. That didn’t change a thing. It birthed new industries. And that is what AI will do.

[Field Guide AI @FieldGuideAI]

Everyone needs to embrace AI regardless of their emotions. It’s not going away. You either adapt or become obsolete.

[Joshua @CreeCoder]

People either love AI or absolutely hate it.

[Field Guide AI @FieldGuideAI]

@iamAlexTurnbull But it’s never been easier to develop a “world class product.” And it’s never been harder to distribute the product. Distribution is becoming much more of a challenge for startups then product development.

[Field Guide AI @FieldGuideAI]

SaaS products need to become agentic systems of records. SAP, Salesforce, PeopleSoft, etc. Now, I am not sure what the pricing model will look like, but these most established software companies still have a sizable moat in my opinion.

[Yasser @yasser_elsaid_]

SaaS is not dead but SaaS dashboards are dead

[Field Guide AI @FieldGuideAI]

@sickdotdev Some jobs are also cheaper for humans to do than exquisite AI systems. The cost of human labor will be weighed against token consumption. I’m already starting to see this calculus play out across my organization and customers.

[Field Guide AI @FieldGuideAI]

I would tell them this and only this: have fun playing with each tool. Let them explore their curiosity and use the tools in real life for whatever the reason may be. After they are interested and familiar with AI, then you can begin to help them navigate the complexities of the labor market and higher education.

[Field Guide AI @FieldGuideAI]

@aditikbuilds This is exactly what I am trying to do. It’s going to take a mountain of effort and time, but I am squarely focused on this side quest.

[Field Guide AI @FieldGuideAI]

Context is the most important lever in AI. And EVERYONE neglects it.

These systems (ChatGPT, Claude, Grok) are large-scale prediction engines, not search engines.

Once you understand this core concept, one thing becomes abundantly clear.

The quality of the output is dictated by the contextual quality of the input.

Vague Prompt → Vague Output

Detailed Context → Useful Output

Here’s what this looks like in real life from an account executive (sales) perspective:

Weak version

“Write a follow-up email after a sales call.”

Stronger version

“Here’s the call summary and the prospect’s main objections: [paste notes].
Draft a short, professional follow-up email that acknowledges their concern about implementation timeline, reference the pilot we discussed, and propose two concrete next steps. Keep the tone direct and low pressure.”

The second version always produces something more valuable and usable.

This doesn’t just apply to sales roles. It applies across ALL roles:

• Sales: past call notes, CRM history, specific objections

• Marketing: brand voice examples, previous high-performing copy, audience details

• Customer Success: ticket history, customer goals, recent interactions

• Delivery / Project leads: scope documents, risk notes, stakeholder preferences

For most grunt daily work, you don’t need fancy graphs, loops, or complicated prompts.

I often see clients and my peers writing novels into the prompt. Yes, details and context matter, but what’s more important is relevance. Only provide the AI system with relevant context for the specific use case.

Also, stop treating the model like it already knows your situation. It doesn’t (yet). And don’t be afraid to start a new chat. If you keep using the same chat thread, the AI model will reference previous conversations which will add a layer of confusion.

The bottom line is this. Feed the #AI model relevant raw material. Then apply judgment to what comes back. You, the human in the loop, know what is right and wrong.

This single habit is what separates people who get mediocre results from people who get consistently useful ones.

The prompt pyramid below is one of my favorite references:

[Field Guide AI @FieldGuideAI]

@NickpxJ Problem framing. Once you have a problem clearly identified, you can then begin to craft tailored prompts.

Problem > Prompt

[Field Guide AI @FieldGuideAI]

@wilczyn Appreciate it man! Thank you for the support 🫡