My AI starter kit

My AI starter kit

Preread: How this started

At Payflip, we are increasingly embracing agentic coding, AI in our internal operations, and experimenting with AI enabled features in our products. Since blindly purchasing tools or winging it is not a decent strategy, we have formed a core 'AI taskforce' spanning across our sales, engineering, product, and support units.

This core team then identifies opportunities, brings them to the table, and evaluates our overall AI adoption within the company. Not all of our taskforce's members were per se experts in (Gen) AI concepts, but they were definitely enthusiastic about taking ownership around AI in the company and learning how Gen AI truly works.

Over the past few years, I have self-hosted LLMs, developed POCs using techniques like RAG and MCP, and experimented with models from various providers to find the differences. This experimenting has given me a waaaaaay better understanding of how these models function, how you can elevate them, and why certain outputs may not meet expectations, along with clear methods for improving that.

First thing that we started with in the taskforce is my AI starter kit. I compiled some key resources (articles, videos, whitepapers, ...) that really helped me understand, and put them into the AI starter kit. Since I believe items in the kit are both easy to understand & interesting, I wanted to share it here for anyone looking to develop a solid foundation in AI knowledge for either their team or personal learning.

Part 1: Core concepts

What is a System prompt? | PromptLayer
Understand system prompts in AI - special prompts that set overall context for models. Learn how they shape AI persona and behavior in interactions.
AI Agents vs. AI Assistants | IBM
Explore the differences, similarities, benefits and risks of AI assistants and AI agents.
Markdown Reference

Whitepapers

Prompt Engineering
Discover what actually works in AI. Join millions of builders, researchers, and labs evaluating agents, models, and frontier technology through crowdsourced benchmarks, competitions, and hackathons.
Startups technical guide: AI agents
Download our guide for startups on building AI agents with Vertex AI, Gemini, & ADK. Learn RAG, multimodal AI, autoML & more on Google Cloud.
DORA | State of AI-assisted Software Development 2025
DORA is a long running research program that seeks to understand the capabilities that drive software delivery and operations performance. DORA helps teams apply those capabilities, leading to better organizational performance.

Courses

The Rundown AI
AI for Marketers | Maven | Unlock your career growth
Live courses with real-world experts in AI, Product, Leadership, Design, Data & engineering, Marketing & sales, and Business & finance.
AI for Operators | Maven | Unlock your career growth
Live courses with real-world experts in AI, Product, Leadership, Design, Data & engineering, Marketing & sales, and Business & finance.

Newsletters & blogs

How I AI | Lenny’s Newsletter | Substack
Practical AI tips, tricks, and workflows from top operators, with screen sharing, prompts, and playbooks you can copy. Click to read Lenny’s Newsletter, a Substack publication with millions of subscribers.
Marcel Salathé | Substack
Passionate about technology’s potential, I explore its impact as an EPFL professor. I’m captivated by smart machines and enjoy using them to enrich my life. Love to travel, read, and be social.
Newsroom
Anthropic is an AI safety and research company that’s working to build reliable, interpretable, and steerable AI systems.
Cole Medin
Hi I’m Cole Medin and I’m your guy for learning how to make a huge impact with AI agents and AI coding assistants. AI is no doubt the future of not just software development but the whole world. And I’m on a mission to master it and share everything with you. Join me as I push the limits of what is possible with AI. I’ll be uploading videos weekly - at least every Wednesday at 7:00 PM CDT!

Reads on AI and ethics, society, environment, economy, security

AI is the new plastic
Plastic is in your car, your kitchen, the chair you’re sitting on right now. And just as our most ubiquitous material transformed the world in a matter of decades, AI is doing the same.
AI gets more ‘meh’ the more you use it, researchers find
: Most scientists now use the tech in their work, but still question its usefulness
A Better Way to Think About AI
AI can be used to automate tasks—and entire jobs. But it could also be designed to collaborate with humans. David Autor and James Manyika on why we should focus on the latter:
Fear is not advocacy
And you are going to be fine.
AI fatigue is real and nobody talks about it · Siddhant Khare
You’re using AI to be more productive. So why are you more exhausted than ever? The paradox every engineer needs to confront.
My AI Skeptic Friends Are All Nuts
My smartest friends have bananas arguments about LLM coding.
LLMs are a tool for thought
Explained: Generative AI’s environmental impact
MIT News explores the environmental and sustainability implications of generative AI technologies and applications.
Introducing the Anthropic Economic Index
Announcement of the new Anthropic Economic Index and description of the new data on AI use in occupations
AI Sovereignty: The Emperor Has No Clothes
Most countries should stop chasing illusions of AI sovereignty and focus on building and strengthening talent pipelines.
#agenticai | Stefan Wolpers
For everyone getting too excited about #AgenticAI; Get back into thinking mode in less than 3 minutes.
AI is a tool in the same way that a compiler is a tool. Current-day LLM-based "AI" cannot write a program any more than a compiler can write a program. It's good with the drudge work given precise… | Allen Holub | 109 comments
AI is a tool in the same way that a compiler is a tool. Current-day LLM-based “AI” cannot write a program any more than a compiler can write a program. It’s good with the drudge work given precise instructions in the same way that a compiler is good at the drudge work of writing assembly language when given precise instructions in a higher-level language. Neither is capable of writing a computer program, much less a product, however. It takes a human for that. | 109 comments on LinkedIn
S02E07: The 100x employee and the AI-native organization
2026 predictions, smallest value-adding unit, agentic companies, shadow orgs, introducing the trillion dollar question
Thoughts on slowing the fuck down
Thoughts on slowing the fuck down
AI, wat niemand ons vertelt - Uitgeverij Prometheus
Als AI nóg krachtiger wordt, hoe zal de toekomst er dan binnen vijftig jaar uitzien? Het is niet uitgesloten dat Trump en Poetin beslissen om slimme technologie te gebruiken voor kernwapens. Terroristen kunnen met AI virussen maken die niet te detecteren zijn en wanneer slimme technologie fouten maakt in de financiële sector is de schade … Lees verder