Weekly Amazon product opportunity brief in Notion

Build me an agent workflow that produces a weekly Amazon product opportunity brief in Notion. It should run on a cron trigger every Monday at 7am in my local time zone.

Inputs the workflow needs (make these configurable on the workflow, do not hardcode in the agent instructions): a list of target keywords I am considering launching products into (for example, silicone baby bibs, magnetic phone wallets, beard oil for sensitive skin), the Amazon marketplace country code (default US), the Notion parent page ID where the briefs should be filed as child pages, and the Slack channel that should receive the short headline.

For each target keyword, use Real-Time Amazon Data Search Products to find the top 8 to 10 best-selling ASINs for that keyword in the configured country. Then call Get Product Details for each of those ASINs to capture title, price, star rating, review count, and Best Sellers Rank (read it from product_information["Best Sellers Rank"], which is a human readable string like "#1 in Cell Phones (Unlocked)").

For each ASIN, also call Get Top Product Reviews to pull the 8 publicly visible recent reviews. Do NOT use Get Product Reviews or Get Product Review Details — those require a logged-in Amazon session cookie and are fragile. Get Top Product Reviews is the cookie-free endpoint and is what we want.

Now do the judgement work, which is why this is an agent and not a fixed pipeline. Across all ASINs for a given keyword, weight the review pool toward the 1 to 3 star reviews (those are where unmet needs hide). Cluster the recurring complaints and feature requests into 3 to 5 themed gaps per keyword. A gap should be a specific, repeatable customer pain ("strap slips off the shoulder", "battery dies after 6 months", "app does not work on Android") that shows up across multiple competitors, not a one-off review. For each gap, draft a one or two sentence differentiation angle: a concrete thing a new product could do to win on that axis.

Write one Notion page per keyword. Use Notion Create a Page as a child of the configured parent page, titled "Amazon Opportunity Brief — <keyword> — YYYY-MM-DD" (today's date in my local time zone). Then use Append Block Children to fill the page. The page should contain, in this order: a 2 to 3 sentence executive summary; a Competitors table with one row per ASIN (columns: ASIN, title, price, rating, review count, BSR); a "Top recurring gaps" section with one heading per gap and a short bullet list of the supporting complaints; and a "What to build" section that lists the differentiation angle for each gap as a punchy one-liner. Keep the whole brief to roughly one page per keyword. Do not pad.

After all per-keyword Notion pages are written, send ONE Slack message via the Slack Bot integration's Send a Message operation to the configured channel. The message should be a short summary with one bullet per keyword in the format: keyword name, a one-line punchy headline of the most promising opportunity, and a link to that keyword's Notion page. Use Slack mrkdwn (single asterisks for bold, <url|text> for links). Keep it scannable so the team sees the briefs exist without having to open Notion.

Tone and rules to bake into the agent: focus on real, repeating complaints, not single angry reviewers; quantify when you can ("6 of 10 top sellers have at least one 1-star review about the strap"); attribute observations to the ASIN they came from where useful; do not invent gaps if the reviews do not support them (if a keyword's review pool is too thin, say so in the brief and surface fewer gaps); avoid em dashes in user-facing copy; never include cookies, session tokens, or anything that would require a logged-in Amazon account.