Estimation question
Estimate the LLM cost per Lindy agent run and its impact on pricing.
- Lindy
- Estimation
- Hard
Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.
Start a mock interview on this question · Mock interview from a job description
What this question tests
Tests structured estimation of unit economics: tying LLM inference cost to a per run cost and its effect on pricing.
How to approach it
- Break down a typical agent run into its LLM calls: assume an average run uses 3 model calls, each with roughly 2,000 input tokens and 500 output tokens.
- Apply a blended token price: assume 3 dollars per million input tokens and 15 dollars per million output tokens for a mid tier model.
- Calculate cost per run: 3 calls times 2,000 input tokens is 6,000 input tokens, costing about 0.018 dollars, plus 1,500 output tokens costing about 0.0225 dollars, totaling roughly 0.04 dollars per run.
- Scale to a typical user: assume an active user runs agents 200 times a month, giving about 8 dollars a month in LLM cost per active user.
- Compare to pricing: if Lindy charges 30 to 50 dollars a month per user, LLM cost is roughly 16 to 27 percent of revenue, leaving room for infrastructure and margin.
- State clearly that token counts, model pricing, and run frequency are assumptions you would confirm against Lindy's actual usage data.
What a strong answer includes
- Builds the estimate bottom up from token counts and per token pricing rather than guessing a blended cost per run directly.
- Shows how the estimate scales from a single run to a monthly active user cost, connecting directly to the pricing question asked.
- Explicitly ties the cost estimate back to pricing strategy, showing LLM cost as a percentage of revenue rather than leaving it as an isolated number.
- Flags heavy users as the real margin risk, since a small segment running agents far above average could erode margin faster than the blended average suggests.
Common mistakes
- Giving a cost per run without showing the token level math behind it.
- Not connecting the cost estimate back to what it means for pricing, leaving the strategic question unanswered.
- Ignoring that usage is likely skewed, with a minority of power users driving disproportionate LLM cost.
Likely follow-up questions
- How would usage based pricing for heavy users change this margin picture?
- What would you do if a cheaper model dropped accuracy on complex agent tasks?
- How would you monitor cost per user to catch margin erosion early?
More estimation questions
- YouTube Red is a premium service without ads. Assume that 1.5% of initial YouTube user base signs up for the service. What is the lifetime revenue Google generates from those users?Google · Estimation · Hard
- How would you design a service like Instagram? Estimate server and storage requirement for peak traffic.Google · Estimation · Hard
- How much storage is required to store all of the images on Google Maps?Google · Estimation · Hard
- A company has launched a new drug that eliminates the need for sleep. How would you price this drug?Google · Estimation · Hard
- How much storage space is required to host all the images of Google Street View?Google · Estimation · Hard
- Imagine you created a new type of product that would replace the mobile phone. How would you determine how many units to manufacture?Google · Estimation · Hard
More questions from Lindy
Learn the skill behind it
Chapters of the AI PM course that teach what this question tests.
- Chapter 2: Data fluency: SQL, logs, and reading the truth yourself
- Chapter 9: Prove it paid off: outcomes, economics, and pricing
- Chapter 14: Get the job: the AI PM interview loop