What is one specific practice you use to balance the need for customization with the benefits of scalable, reusable AI solutions for different customer segments?
Most answers focus on embedding company-specific knowledge or preferences into scalable AI systems, though approaches range from strict standardization to building separate tools or fine-tuning models for customization.
8 answers
Answers


Jake Stauch
Founder and CEO · Serval
Never Build Custom Features
Jake Stauch says his team never builds custom features or bespoke product capabilities, because everything customers ask for is something every other customer will eventually want, though it may affect prioritization.
No Priors: AI, Machine Learning, Tech, & Startups · Watch on YouTube

Mike Cannon-Brookes
CEO and Co-Founder · Atlassian
Model Routing Per Task
Mike Cannon-Brookes says managing technological cost is rooted in intelligent model routing on a per-task basis to optimize cost, applying the right model at the right time.
The answer starts at 26:58.
The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch · Listen on Apple Podcasts

Harry Stebbings
Founder · 20vc
Fine-Tune Open-Weight Models
Harry Stebbings says his team fine-tunes specialized open-weight models for particular purposes, which he says give great price performance for customers, though he doesn't claim these are their own models.
20VC with Harry Stebbings · Watch on YouTube

Marty Cagan
Partner · Silicon Valley Product Group
Configure LLM with Context
Marty Cagan recommends configuring your favorite large language model with your company's strategic context—product vision, strategy, and the problem you're solving—so it can act as a personal product coach without paying for extra applications.
Mind the Product · Watch on YouTube

Kieran Klaassen
GM of Cora | EIR · Every Inc.
Teach AI Your Taste
Kieran Klaassen says people should never outsource their taste, and that you have to teach AI what your taste is and why what you create is right for your customers and prospects.
The answer starts at 10:43.
Marketing Against The Grain · Listen on Apple Podcasts

Garry Tan
Founder · Garry'S List
Skillify Repeated Tasks
Garry Tan recommends skillifying every repeated task, saying that asking for something a second time is the failure. He points to his own finance team folding 100 spreadsheets into one internal tool as an example.
The AI Corner · The AI Corner · Read

Dan Martell
Founder & CEO · Martell Ventures
Build Separate Tools
Dan Martell says he built two separate tools with his team so users get an out-of-the-box solution without having to configure and manage the security and complexity themselves.
Dan Martell · Watch on YouTube

Lin Qiao
CEO and cofounder · Fireworks Ai
Bake Judgment Into Models
Lin Qiao says companies should bake their judgment, taste and customer understanding into the intelligence they build on top of, rather than relying on an off-the-shelf API, as an approach to building a durable business.
Sequoia Capital · Watch on YouTube
More Generative AI questions
View all →- What design choice improves AI agent privacy?5 answers
- What architectural choices enable reliable agentic AI systems?8 answers
- What responsibilities should AI creators have for their models?12 answers
- How does your team create feedback loops for AI models?12 answers
- What architecture optimizes compute for AI workloads?13 answers



















