<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel>
<title>Akshay Sharma</title>
<link>https://akshaysharma.io/writing/</link>
<description>Senior Product Manager at Avalara. 10+ years building enterprise AI products.</description>
<item><title>Your AI roadmap should be a list of questions</title><link>https://akshaysharma.io/writing/ai-products-need-a-different-roadmap/</link><guid>https://akshaysharma.io/writing/ai-products-need-a-different-roadmap/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>Feature roadmaps assume you know what the software will do once it is built. With AI, you find that out after you ship.</description></item>
<item><title>Evals are a product decision</title><link>https://akshaysharma.io/writing/evals-are-a-product-decision/</link><guid>https://akshaysharma.io/writing/evals-are-a-product-decision/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>If the product manager does not define &quot;good&quot;, the model team will, and they will optimise for whatever is easiest to measure.</description></item>
<item><title>Agents in regulated work need a place to stop</title><link>https://akshaysharma.io/writing/agentic-ai-for-regulated-work/</link><guid>https://akshaysharma.io/writing/agentic-ai-for-regulated-work/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>In tax, finance or healthcare, an agent that is right most of the time is not enough. Someone is accountable for every output.</description></item>
<item><title>Six questions before an AI idea reaches the roadmap</title><link>https://akshaysharma.io/writing/choosing-an-ai-use-case/</link><guid>https://akshaysharma.io/writing/choosing-an-ai-use-case/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>Most AI ideas fail before a model is involved, because the use case was never worth solving.</description></item>
<item><title>The demo worked. Now ship it.</title><link>https://akshaysharma.io/writing/from-demo-to-production/</link><guid>https://akshaysharma.io/writing/from-demo-to-production/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>Five things that break between a good demo and a product people use every day, and what to plan for each.</description></item>
<item><title>A 90-day plan for moving into AI product management</title><link>https://akshaysharma.io/writing/moving-into-ai-product-management/</link><guid>https://akshaysharma.io/writing/moving-into-ai-product-management/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>You do not need a machine learning degree. You need working knowledge, one thing you built, and a way to talk about both.</description></item>
<item><title>The cost of a right answer</title><link>https://akshaysharma.io/writing/cost-of-a-right-answer/</link><guid>https://akshaysharma.io/writing/cost-of-a-right-answer/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>What cloud cost governance taught me about pricing an AI feature, and why cost per request is the wrong number.</description></item>
<item><title>Build, buy or wrap?</title><link>https://akshaysharma.io/writing/build-buy-or-wrap/</link><guid>https://akshaysharma.io/writing/build-buy-or-wrap/</guid><pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate><description>Three ways to bring generative AI into an enterprise product, and how I choose between them.</description></item>
</channel></rss>
