- AWS is rolling out agentic AI products at its AWS re:Invent conference today
- At the same time, more than 1,000 Amazon employees (and counting) have signed an open letter saying the company needs to slow down with AI
- The letter says AI has caused Amazon to abandon its climate goals, among other charges
AWS RE:INVENT, LAS VEGAS – AWS is pushing the boundaries of what AI can do, introducing new autonomous Frontier Agents here at re:Invent. But a large group of Amazon employees seem to think the "move fast and break things" ethos that has dominated the tech industry over the past decade shouldn’t apply to AI.
“We believe that the all-costs-justified, warp-speed approach to AI development will do staggering damage to democracy, to our jobs, and to the earth,” reads an open letter signed by more than 1,000 Amazon staffers.
The letter purports to be penned by “the workers who develop, train and use AI.” Though it is undated, reports about the letter surfaced shortly before AWS CEO Matt Garman noted onstage at re:Invent that Amazon last week standardized internal development on Kiro, an AI-powered environment that features a new Kiro Autonomous Agent capable of reducing the time and manpower required for development tasks.
The missive accuses Amazon of forcing workers to use and train AI that could be used to displace them, abandoning climate goals in the frantic push to deploy AI at scale, and helping build a surveillance state that could be misused by governments and other entities.
It calls on Amazon to correct course. Amazon signatories were purportedly joined by more than 3,600 workers from other companies, including Microsoft, Google, Meta and Apple.
Frontier Agents
Concerns about AI stealing jobs aren’t new. But they – along with climate concerns – are resurfacing as the technology advances and cloud companies scramble to deploy the compute power required to make them work.
At re:Invent, Garman introduced AWS' Frontier Agents, “a new class of agents that are a step function change more capable than what we have today.” Unlike other agents, Garman said Frontier Agents are autonomous, massively scalable and long-running – that is, they can run without intervention for hours or days.
The aforementioned Kiro Autononous Agent is one of a trio of Frontier Agents that AWS debuted at the show. Garman recounted how Kiro was used for an internal rearchitecture project that was expected to take 30 people 18 months to deliver. Using the tool, a team of six people was able to finish the project in 76 days.
“This is not just the 10-20% efficiency gains that people were seeing with the first generation of AI coding tools,” Garman said.
The two other initial Frontier Agents include an AWS Security Agent, to consult on app design, code reviews and penetration testing and the AWS DevOps Agent, which can help prevent and resolve incidents.
“I believe that the advance of the AI agents has brought us to an inflection point in AI’s trajectory,” Garman said. “It’s turning from a technical wonder into something that delivers us real value.”
Ryan Shrout, president at Signal65, told Fierce that the industry has hit a point where AWS has the potential to turn agentic AI “into a real system rather than a lab experiment.” He added that if AWS can ship a truly coherent agentic stack – something that doesn’t break when developers push it – then “it puts them in a different position: Not just hosting the AI boom – directing it.”
It seems its workers think it’s high time for the company to start thinking more deeply about what Frontier Agents and the like mean not just for the bottom line, but society at large.