This week's release is all about accuracy. We shipped three bug fixes across our Amazon and Google API suites, all aimed at the same underlying goal: making sure the data you get back is exactly the data you asked for no wrong regions, no duplicate clutter, no garbage payloads. Here's what changed.
Amazon Product & Search API: Regional Data Now Matches Your Postal Code
If you've ever passed a postal code into the Amazon Product API or Amazon Search API and gotten results that clearly belonged to a different region, this fix is for you. We identified an issue where both APIs would occasionally ignore the specified postal code and return data from the wrong region pricing, availability, and listings that didn't reflect where you actually wanted to scrape from.
This mattered most for teams doing localized price monitoring, competitor tracking, or availability checks across different markets. A silent regional mismatch is worse than an obvious error, because it can quietly skew an entire dataset before anyone notices.
The issue has now been resolved. Both the Amazon Product API and Amazon Search API correctly honor the postal code you pass in, returning region-specific data every time. If you run location-sensitive Amazon scrapes, this is a meaningful reliability improvement and something you should notice immediately in your next pull.
Google AI Mode & AI Overview: No More Duplicate References
Our second fix targets the Google AI Mode API and AI Overview API. Some queries were returning duplicate reference entries in the response, the same source cited more than once for a single answer. Beyond being untidy, duplicate references made it harder to build clean citation lists, reference counts, or source-attribution features on top of our data without adding your own deduplication logic first.
We've resolved this at the source. Reference data returned by both APIs is now deduplicated automatically, so what you get back is a clean, accurate list of the sources actually backing each AI-generated answer. If your workflow parses references for citation display, fact-checking, or content sourcing, you should see immediately cleaner output with no extra post-processing required on your end.
Google AI Overview & AI Mode: Clean JSON and Markdown, Every Time
The third fix addresses a more disruptive issue: garbage data occasionally showing up in the JSON and Markdown responses from the Google AI Overview API and AI Mode API. This is the kind of bug that's especially painful because it can break downstream parsing, malformed JSON fails silently or throws errors, and garbled Markdown renders incorrectly or pollutes text pipelines.
We've traced and fixed the root cause, and both APIs now consistently return clean, well-formed data in the format you request. Whether you're consuming JSON for structured processing or Markdown for direct display, you can expect reliable, parseable output going forward.
Here is a video tutorial about our weekly updates.
Why This Batch of Fixes Matters
Individually, each of these might look like a small fix. Together, they point to a theme: precision and trustworthiness in the data layer. Whether it's making sure a postal code actually determines your region, making sure a reference list doesn't double-count sources, or making sure a JSON payload is actually valid JSON, these are the kinds of issues that erode trust in an API silently, over time, until someone downstream finds a broken pipeline or a skewed report.
Fixing them proactively means:
- Amazon scraping workflows that depend on regional accuracy (pricing comparisons, localized availability, market-specific listings) can trust the postal code parameter again.
- Teams building citation or source-tracking features on top of Google AI Mode and AI Overview data get clean reference lists without extra deduplication work.
- Anyone parsing structured output (JSON) or rendering formatted output (Markdown) from Google AI Overview and AI Mode APIs gets consistently valid, usable data.
What You Should Do Next
If you're actively using any of these four APIs, Amazon Product API, Amazon Search API, Google AI Mode API, or Google AI Overview API and no changes are required on your end. The fixes are live now and apply automatically to your existing integrations. That said, it's worth doing a quick spot-check on your next few API calls, especially if you had built workarounds for any of these issues (like manual deduplication of references, or extra validation steps for malformed JSON). Those workarounds are no longer necessary and can be safely removed.
As always, if you notice any unexpected behavior in your responses, reach out to our support team with a sample request and response, the more detail you provide, the faster we can investigate.
That's it for this week's update. We'll be back next week with more improvements as we keep tightening up data accuracy across the board.
