TL;DR
Get privacy and security gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
AI-assisted site reliability engineering tools are gaining attention as companies report more deployments and on-call pages. Current systems mainly help investigate incidents and triage alerts; reports of agents independently fixing production problems remain conference demos rather than routine practice.
AI-assisted on-call tools are moving from conference demos into engineering teams’ incident workflows, but their current role is mainly to investigate alerts and help engineers triage them, according to a report by The Pragmatic Engineer. The report says some companies are building their own systems, while vendors are adding AI features; it found no evidence that agents routinely take production decisions and ship fixes without human oversight.
The report describes a change in attention since Microsoft demonstrated its Azure SRE Agent at its May 2025 BUILD conference. In the presentation, the agent investigated an alert and shipped a fix without waking an on-call engineer. The author said the demonstration seemed artificial and questioned whether teams would allow an AI to make production changes without supervision. Interest has since grown, with the report citing OpenAI’s internal Sevbot, an in-house effort at ING described by an engineer, and more vendor presence at technology conferences.
In practice, “AI SRE” usually refers to tools that gather incident context, investigate possible causes and work through chat platforms such as Slack or Teams. The report says systems may use multiple agents, engineers steering their work, and a context layer that stores service knowledge and prior incident information. Whether the tools should also modify code or production systems is contested. The author notes an investment in incident.io and says the companies discussed were not paid for inclusion.
The report cites Rootly data showing that pages are up compared with a year earlier and that outside-business-hours pages nearly doubled. It also says many companies have seen pull-request frequency double or more over the same period, linking higher change volume to more opportunities for failures. These are reported trends, not proof that AI-generated code alone caused the increase; the report does not provide a universal measure of incident rates across the industry.
More Changes Mean More On-Call Pressure
For engineering teams, the immediate promise is not to remove human responders but to reduce the time spent gathering information after an alert. If a tool can assemble logs, service ownership, recent changes and relevant past incidents, engineers may reach a useful hypothesis sooner. Alert triage could also reduce pages that do not require a person’s attention, easing disruption outside working hours.
The trade-off is that incidents reaching a human may be more difficult, while engineers could lose practice if they rely too heavily on automated investigation. The report flags this deskilling risk and points to incident-response simulations, including those offered by Uptime Labs, as one way to maintain response skills. The net effect on reliability, workload and recovery time is not yet established by the material cited.
Products Worth Considering
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
DJI Mic 3 Wireless Microphone System (2 Transmitters, 1 Receiver, Charging Case)
As an affiliate, we earn on qualifying purchases.
From Demo to Incident Workflow
“AI SRE” is an imprecise label: site reliability engineers work on system reliability well beyond responding to alerts. The report says vendors themselves have not settled on a better term, despite alternatives such as automated incident investigation or remediation agents. For now, the category covers a range of tools, from alert triage to guided investigation, rather than one standard capability.
The report frames rising change frequency as one pressure on reliability. Its simplified relationship is that incident frequency depends on both how often systems change and the chance that any individual change causes a failure. More deployments can raise incident risk even if the failure probability per change stays the same. The report says AI SRE tools aim to help teams learn from incidents and improve resilience, while pre-deployment testing addresses risk before changes reach production.
“The fully autonomous agents demoed on conference stages are still not a reality.”
— The Pragmatic Engineer report
Accuracy and Autonomy Remain Open
It remains unclear how much AI SRE tools improve incident outcomes across different companies. The report says accuracy is difficult to define and measure because each company and engineering team has different systems, incident histories and expectations. Suggested methods include testing hypotheses with adversarial agents, using multiple models, running evaluations and backtesting against past incidents, but no common industry benchmark is described.
The source material also does not establish that rising pages are caused specifically by AI-written code, or provide enough detail about Rootly’s data to assess its representativeness. It says fully autonomous response is not yet routine, but does not give a timeline for when it might become safe or reliable. The scope of production access granted to the cited in-house systems is also not specified.
Teams Will Test Triage Before Remediation
The near-term development to watch is whether teams use AI to sort alerts and assemble incident evidence, and whether that reduces unnecessary pages or speeds human diagnosis. The report says it is becoming more common for agents to triage pager calls and silence alerts judged unimportant, while the extent and effectiveness of that practice remain uncertain.
For organizations adopting these tools, the next steps are likely to include testing them against past incidents, checking their hypotheses and limiting how much authority they have over production systems. The report identifies model choice, prompt design and cache performance as ways teams balance speed, cost and accuracy. It does not identify a shared industry standard or a scheduled milestone for autonomous remediation.
Key Questions
What does “AI SRE” mean?
It generally refers to AI tools that support on-call incident response, especially investigation and alert triage. The term is broader and less precise than the work of a site reliability engineer.
Are AI agents fixing production incidents on their own?
The report says fully autonomous agents shown at conferences are not yet routine in practice. Current use is more focused on investigating incidents and helping prioritize alerts; the level of human oversight varies and is not fully detailed.
Why are companies paying more attention to these tools?
The report points to more frequent software changes and rising on-call pages. It cites Rootly data showing pages are up year over year and outside-business-hours pages have nearly doubled, though the supplied material does not describe the dataset in detail.
Do AI SRE tools measurably improve reliability?
The report does not establish a shared measure or industry-wide result. Accuracy and usefulness depend on each company’s systems and incident patterns, making cross-company comparisons difficult.
Source: newsletter.pragmaticengineer.com
