Back to Blog
AI & Search

Project Amelia's May Update: Amazon's AI Seller Assistant Can Now Forecast Inventory

KL
Kiet Lam
May 19, 20266 min read

Amazon launched Project Amelia in late 2024 as a generative AI assistant embedded directly in Seller Central. The first version was useful in a limited way: it could answer questions about your account metrics, summarize policy changes, and help draft responses to buyer messages. It felt like a capable prototype. The May 2026 update changes that assessment. A few of the new capabilities are genuinely meaningful for how brands manage their day-to-day operations.

What Is New in the May Update

The headline addition is inventory forecasting. Amelia can now analyze your historical sales velocity, seasonality patterns, and current FBA stock levels to generate a demand forecast and a recommended replenishment timeline. The model pulls from your own account data and from category-level trends that Amazon aggregates across the seller base.

I have been testing this feature with a few client accounts over the past few weeks, and the forecasts are reasonably accurate for established ASINs with at least 90 days of consistent sales history. For newer listings or products with high seasonality, the confidence intervals are wide enough that you should treat the output as a starting point rather than a plan. But as a quick sanity check before submitting an inbound shipment, it is better than most sellers' current process.

The second new capability is automated case drafting. If you have a stranded inventory issue, a listing suppression, or an account health flag, Amelia can now draft the Seller Support case on your behalf. You review it, edit as needed, and submit. For routine cases where the resolution path is well-established, this saves a meaningful amount of time. For complex situations involving account warnings or policy violations, the draft still needs significant human input before it is ready to send.

Third: Amelia now has access to your advertising performance data, which the initial version did not. You can ask questions like "which campaigns have the highest ACOS over the past 30 days" or "what is my total ad spend as a percentage of revenue in the tools category" and get a real answer in seconds. This is not a replacement for the advertising console, but for a quick cross-account overview, it is faster than pulling reports manually.

Where Amelia Still Falls Short

The forecasting feature has a notable limitation: it does not incorporate external signals. Amelia is working from your own data and Amazon's aggregate category data. It does not account for competitor launches, supplier lead time changes you have not encoded somewhere in your account, or macroeconomic shifts that would affect demand. A tool that does not know about a tariff change that affects your sourcing is going to generate forecasts that miss in ways that are hard to catch if you are relying on it passively.

The case drafting feature is similarly bounded. It is good at structural problems with known resolution paths. It struggles with cases that require narrative judgment, where you need to explain context, demonstrate good faith, or push back on an Amazon decision that seems incorrect. For those situations, the draft is more of a useful starting template than a finished document.

There is also a practical limitation worth noting: Amelia is only available in the Seller Central desktop interface as of May. The mobile app does not have it yet, which limits how you can use it for quick checks while away from a computer.

How to Get the Most Out of It Now

The most immediate use case is the inventory forecasting check before inbound shipments. Pull up Amelia, ask for a replenishment recommendation on your top 10 ASINs, and compare the output against what you were already planning to send. Discrepancies are worth investigating. If Amelia is recommending significantly higher or lower quantities than your current plan, find out why before you submit the shipment.

For case management, the workflow that works best is to start the case draft in Amelia, then move it into your case management process for review before submitting. Do not submit directly from Amelia without a human read-through. The drafts are a good starting point, but they are not tuned for the specific tone and structure that tends to get faster responses from Seller Support.

The advertising data access is genuinely useful as a quick-answer layer. Rather than training your whole team to navigate the advertising console, you can use Amelia to answer common questions faster, reserving the console for deeper analysis and campaign management.

How TKL Is Integrating Amelia

We are incorporating Amelia's forecasting output into our standard monthly inventory planning sessions for managed accounts. It does not replace our own modeling, but it adds a useful Amazon-sourced cross-check that can catch discrepancies our models might miss, particularly for categories where Amazon has richer demand-side data than we do.

If you want to understand how to get the most out of Amelia for your specific account type and catalog size, we are happy to walk through the setup and the use cases that make the most sense for your team. The tool is still maturing, but May's update moved it from "curious experiment" to "genuinely useful" for a meaningful subset of seller workflows.

Work with TKL

Ready to adapt your Amazon strategy?

Keeping pace with Amazon's changes is demanding. TKL helps ambitious brands stay ahead, protect margin, and grow. Tell us about your brand.

Get in touch