Two years building 'the median': why AI demos don't make durable products
AI Engineer · youtube · 2026-10-15
AI Engineer talk by Hariprasad Sudharshan: after two years of solo building, he found others had independently built nearly identical calculators — the same models, tools, and familiar problems converge everyone onto the same solution, which he calls 'the median'.
Key points:
- Separate model capability from harness and from the product the customer receives; a working demo isn't a durable advantage.
- The 'build first' reflex — opening a coding agent instead of searching for existing solutions — is a trap.
- Novelty budget: name the one distinctive thing before starting, put it in the roadmap, and review it like other spending.
- Three questions for leaders evaluating demos: Who has already solved this? What does this team have that competitors don't? Where will the app reach users?
- A later calculator iteration examines running open weights in-enterprise vs buying inference, including utilization and infra between weights and tokens.
Takeaway: use the readily available starting point, then spend judgment on a specific problem, distribution, and a defensible difference.
More from coding & agent
- Zenzap raises $44M and launches AI agents in its workplace chat app — testingcatalog · 2026-10-15
- LangChain Deep Agents demo: drop a CSV in Slack, get a slide deck back — Hacubu · 2026-10-15
- 'Slop' perception about to reverse: soon not using AI may signal sloppy coding — mark_k · 2026-10-15
- Workday and Anthropic connect HR and finance agents via MCP — emmanuelvivier · 2026-10-15
- Shopify opens its billion-listing catalog to any AI agent with one-click Shop Pay checkout — emmanuelvivier · 2026-10-15
- DuoDiscover turns evaluation into agent search, 2x experts on SWE-fficiency — OfirPress · 2026-10-15