🔍 Read the full analysis: Amazon’s Push Against Tokenmaxxing Puts Outcomes In Focus on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Amazon published engineering guidance on October 27 rejecting AI token use and tool adoption as performance targets, while saying those measures can still help identify trends. The memo says teams should judge AI by its effect on customer, business and team outcomes. Amazon’s third-quarter results were due October 29, but the source provides no results or new financial figures.
Amazon published new guidance for its software engineers on October 27, warning that targets based on AI token use and tool adoption can reward activity rather than results. The memo says engineers should be evaluated on their impact on customers, the business and their teams, placing renewed focus on whether AI tools improve work rather than how often employees use them.
The memo was written by Terk Terkowitz, a vice president and distinguished engineer in Amazon’s stores business, and presents a revised role guideline for software engineers. It says long-term outcomes are the most relevant measures but can take time to assess. Some teams have used measures such as token consumption, AI-tool adoption, code changes and code reviews as quicker proxies.
Amazon calls those measures “deeply flawed” when used as targets, arguing that they are not directly tied to the results that matter and can increase costs or system complexity. It gives tokenmaxxing—using AI tools to consume more tokens for the sake of a score—as an example that conflicts with Amazon’s principle of frugality. It also warns that setting lines of code as a goal conflicts with simplifying systems.
The guidance does not say that usage data has no value. It says such inputs currently “correlate well” with important outputs, and notes that teams adopting AI more heavily are delivering more while teams that do not adopt it are falling behind. Amazon’s stated distinction is between using metrics to spot patterns and using them to score or manage people.
The memo asks leaders to examine both low and high adoption. Looking at teams around the 10th percentile can help identify why tools are not useful or being adopted; examining those around the 90th percentile can test whether heavy use is producing better results. It also tells leaders to try the tools on their own production work and explain where they helped and where they did not.
Six months at Amazon
From token leaderboards to an outcomes memo
Amazon spent 2026 learning the same lesson twice: once with how its engineers use AI, once with its own AI models.
- May 2026Token leaderboards backfireFinancial Times (via The Decoder): staff give an internal agent tool pointless tasks to inflate token counts. Reported, anonymous sources.
- Late May 2026KiroRank “deprecated”Amazon spokesperson to Business Insider. SVP Dave Treadwell: “Please don’t use AI just for the sake of using AI.”
- 24 Jul 2026San Francisco AGI Lab site closesConfirmed by Amazon to GeekWire. Frontier model research continues under Pieter Abbeel.
- 28 Jul 2026Most flagship Nova models wound downBusiness Insider, people familiar: Premier, Omni, Reel, Canvas to “keep the lights on”. Amazon: models “remain one of the most important things we’re working on”.
- 14 / 30 Sep 2026Nova Sonic v1, Nova Reel, Nova Canvas reach end of lifeAWS Bedrock model lifecycle page. Requests to them now fail.
- 26 Oct 2026Anthropic files its S-1Amazon notes: $6.2 billion principal, 12% ownership cap that Amazon can waive. Anthropic commits at least $110 billion to Amazon through April 2036.
- 27 Oct 2026The memo: outcomes, not tokenmaxxingUsage and throughput metrics are “deeply flawed” as targets. “Tokenmaxxing is at odds with frugality.”
- 29 Oct 2026Amazon’s third-quarter resultsAnnounced by Amazon. Not yet published when this article went live.
Sources: Amazon engineering memo (27 Oct 2026); Business Insider (29 May and 28 Jul 2026); GeekWire (24 Jul 2026); The Decoder (12 May 2026, summarising the Financial Times); AWS Bedrock model lifecycle (read 28 Oct 2026); Anthropic S-1 (26 Oct 2026).
Powered by Thorsten Meyer AI
Why AI Usage Targets Can Backfire
The guidance addresses a measurement problem facing employers adopting generative AI: usage is easier to count than value. Token totals and adoption rates can show whether tools are being tried, but on their own they do not establish that work improved. Turning those figures into performance targets risks encouraging employees to generate activity that raises costs without improving customer outcomes.
For engineers, the memo makes room for a reasoned decision not to use AI on a particular task. It says that if current tools do not help, “that is a finding, not a failing”, and asks employees to document and share the reason. For managers, the practical test is whether the work became better, faster or more useful—not simply whether an AI tool was involved.
The guidance is also a signal about how Amazon wants its internal AI rollout assessed. It supports adoption where it improves work, while cautioning against treating adoption itself as proof of success. The memo does not give a new companywide measure of AI productivity or quantify how much value the tools have delivered.
Products Worth Considering
Apple iPad Air (2025) 13‑inch Wi‑Fi 256 GB – Space Gray (M3 chip)
Logitech MX Brio Ultra HD 4K Webcam, Graphite
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Employee Leaderboards to Guidance
The memo follows earlier reporting about informal AI-use tracking at Amazon. The Financial Times reported in May that some employees used an internal agent tool called MeshClaw for unnecessary tasks to increase their token counts. Those accounts were attributed to employees and are not independently established by the memo. The report also described a weekly AI-use target and token leaderboards.
At the end of May, Amazon confirmed to Business Insider that a leaderboard called KiroRank had been deprecated. The company said it was an informal tracker created by a group of employees and was not intended to promote AI use for its own sake. Senior vice president Dave Treadwell also told staff, “Please don’t use AI just for the sake of using AI.” The October guidance takes that message into an official engineering role guideline.
The source article also compares Amazon’s internal measurement debate with its broader AI strategy, including its Nova models and its investment in Anthropic. It cites Business Insider reporting that most flagship Nova models were moved to maintenance and Amazon lifecycle listings showing some Nova products retired. Those developments provide a wider business backdrop, but the October memo itself does not announce a change to Amazon’s model strategy or provide a direct financial comparison between the two efforts.
““Tokenmaxxing is at odds with frugality.””
— Amazon engineering memo
What the Memo Does Not Measure
The guidance does not specify a replacement score, a standard method for measuring AI’s effect on productivity, or how managers should compare work across different teams. It also does not state whether Amazon will formally change performance reviews or remove all usage dashboards. The memo instead calls for judgment and for examining adoption patterns at both ends of the range.
The source material reports that Amazon’s third-quarter results were scheduled for October 29, but it does not include those results. Claims about the financial value of Amazon’s Anthropic investment, or the performance of its Nova models, should not be read as findings of the engineering memo. The effect of the new guidance on costs, product quality and employee evaluations remains unknown.
Results and Rollout to Watch
The next dated corporate milestone in the source material is Amazon’s third-quarter earnings report, scheduled for October 29. That report may give investors information about the company’s broader financial performance, but the source does not establish that Amazon will disclose AI-specific productivity measures or discuss this memo.
Within engineering, the guidance points to continued review of how different teams use AI and whether that use improves outcomes. Leaders are asked to try tools on production tasks, compare where they help and where they do not, and share those findings. Further details on how Amazon will apply the guideline in performance reviews, or whether it will publish results of that process, have not been provided.
Key Questions
What did Amazon tell its engineers about tokenmaxxing?
Amazon’s October 27 memo says token use should not be a performance target. It warns that consuming tokens for a score can increase costs without showing that work has improved.
Is Amazon telling employees to stop using AI?
No. The memo says AI adoption can correlate with better outputs and notes that teams that do not adopt AI may be falling behind. Its guidance is to judge tools by their effect on work, not usage alone.
What does Amazon say engineers should be measured on?
The role guideline centers on impact on customers, the business and the team. It does not set out a single replacement metric, and it recognizes that longer-term outcomes can be slow to measure.
What happened to Amazon’s KiroRank leaderboard?
Amazon told Business Insider at the end of May that KiroRank had been deprecated. The company described it as an informal tracker made by employees and said it was not intended to encourage AI use for its own sake.
Does the memo say how much AI improves Amazon’s productivity?
No. The memo says some AI-use measures currently correlate with important outputs, but it does not provide a quantified productivity gain or a companywide measure of AI’s impact.
Source: ThorstenMeyerAI.com
