#13 — ChatGPT 4 prompt engineering for founders
January 24, 2024•2 min read

Get actionable go-to-market insights delivered weekly. No fluff, no spam, just strategy that works.
Get actionable go-to-market insights delivered weekly. No fluff, no spam, just strategy that works.
Why it matters: GPT-4 offers startup founders powerful capabilities for building AI-powered products, with recent updates dramatically expanding what's possible through effective prompt engineering.
By the numbers:
- GPT-4 scores in the top 10% on simulated bar exams
- Context window of 128K tokens (equivalent to 300+ pages)
- Training data cutoff: April 2023
Key capabilities for founders
GPT-4 Turbo: The latest iteration brings improved instruction following, JSON mode, reproducible outputs, and parallel function calling - critical features for product integration.
Vision integration: The model now processes images alongside text, enabling applications from chart analysis to visual content moderation.
JSON structuring: Founders can force structured outputs by setting response_format
to { type: "json_object" }
, making API integration significantly easier.
Engineering best practices
Steering tone and style: System messages allow founders to define product voice once rather than in every prompt:
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="gpt-4-1106-preview",
messages=[
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Who won the world series in 2020?"}
]
)
Function calling: Define functions and let GPT-4 intelligently generate JSON for function calls - perfect for converting natural language into API calls or extracting structured data.
Reproducibility: Set the seed
parameter to any integer for deterministic outputs, critical for testing and debugging AI features.
The bottom line
GPT-4 excels at drafting documents, writing code, answering questions, and creating natural language interfaces - but founders should implement safeguards against hallucinations and reasoning errors, especially for high-stakes applications.
Go deeper: Combine GPT-4 with external knowledge sources and advanced prompt techniques like chain-of-thought prompting to maximize reliability.
Keep reading

#14 — SaaS: In a nutshell
Understand why SaaS businesses work and how to grow them.

#15 — Pricing SaaS
How to approach pricing and packaging a low-touch SaaS product.

#16 — How to create subscription tiers that work
Learn how to structure tiers that maximize revenue, align with customer value, and create clear upgrade paths for growth.