All AI videos

AI video

I Spent Two Years Building the Median — Hariprasad Sudharshan

Hariprasad Sudharshan argues that shared models and coding tools pull teams toward similar products, then proposes a novelty budget and three diligence questions.

AI Engineer · 2026 featured video

Watch on YouTube

What this video shows

Sudharshan distinguishes deterministic software, rented model capability, the harness around a model, and the product that reaches customers. When many builders use similar models and tools for a familiar problem, they can independently produce similar solutions. He calls that common starting point the median. Fast implementation can still create valuable internal tools and learning projects, but it does not establish a product advantage.

His practical response is a novelty budget: identify one distinctive capability before building and review it as deliberately as cost or schedule. He asks leaders to check who already solved the problem, what the team has that a competitor lacks, and how the product will reach users. These questions support product diligence, but the talk does not measure which forms of novelty produce retention or revenue. Learnetto recommends testing a specific advantage with users before expanding the build.

Check the event's original session description for the speaker's stated argument and scope. Inspect the speaker's public GitHub profile for projects linked from the video description.

What you will learn

  • Separate model capability, harness quality, product value, and distribution when assessing an AI idea.
  • Write one testable source of differentiation before implementation starts.
  • Use fast AI-assisted building freely for internal tools and learning while applying a higher evidence bar to a commercial product.
  • Ask how a competitor could reproduce the feature with the same public models and tools.

How to apply this safely

  1. Describe the user's job and list the closest existing ways to complete it.
  2. Name one advantage based on data, workflow access, trust, distribution, or execution that the team can test.
  3. Build the smallest experiment that measures whether target users notice and value that difference.
  4. Stop, revise, or narrow the product when the test shows that implementation speed is the only advantage.

Important limitations

  • The median and novelty budget are the speaker's product concepts, rather than results from a controlled study of startups or product outcomes.
  • A familiar implementation can still create value through service, integration, price, trust, or distribution. Novel technology is one possible advantage among several.

Sources to check

Continue learning on Learnetto