Introduction
The Lab45 AI platform, launched by Wipro's innovation lab in June 2024, gives Wipro employees and clients access to multiple large language models through a single SaaS platform, with no-code apps for HR, sales, marketing and operations and API access for developers. Each app built on the platform needs a default model. Defaulting to the highest-quality frontier model gives better answers but can cost 10–30 times more per request than a smaller model. Defaulting to the cheapest capable model controls cost at Wipro's scale (about 2.3 lakh employees) but risks lower quality, which could undermine trust and adoption. The question asks which default the platform should choose. I'll clarify, identify the trade-off, understand the product, form a hypothesis, choose metrics, design an experiment, plan analysis, set a decision framework and recommend.
"Default" is a powerful product lever because most users never change it. The right default should be good for the typical user and task, with clear ways to upgrade when needed.
Step 1
Clarifying Questions (3 minutes)
- Which apps? Employee-facing apps: document summarisation, email drafting, knowledge Q&A, proposal writing, code explanation.
- Usage scale? Assume 80,000 weekly active users and about 3 million requests a month.
- Cost structure? Wipro pays model providers per token; the platform team has an annual budget.
- Quality bar? Varies: casual drafting tolerates small errors; client proposals and code require high quality.
- Who decides? The platform product team sets defaults; app builders can override within policy.
Practice similar questions
Subscribe to access the full answer
.png){kind=logo}
