| Issue 19 | September 22, 2026 |
The RatioA weekly newsletter on reliability economics | | |
|
The Number
6 of 9
Six of the nine organizations classified as near-optimal reliability investment are in Financial Services — a sector that represents just 32% of the benchmark.
Six of the nine near-optimal organizations in this benchmark are in Financial Services. That sector is 32% of respondents.
Neutral
Regulation forced Financial Services firms to price failure to the dollar, while everyone else budgets reliability by habit and guesses wrong in both directions, overspending or underspending with no logic behind either. Same as insurance underwriting: actuaries who price from loss history beat the ones pricing from gut, every time.
| Financial Services is 32% of this benchmark but holds 67% of the organizations getting reliability investment right. |
|
|
This Week in Reliability
Autonomous Operations Land in Production
Multiple vendors shipped self-healing systems that act without human approval in February-March 2026, moving AI SRE from proof-of-concept to closed-loop automation. The question is no longer whether agents can diagnose—it's whether organizations will let them fix.
|
Cast AI · Primary evidence—convergence thesis Komodor combined autonomous self-healing with cost optimization in a single February 2026 product release. Cast AI argues reliability decisions and cost decisions share the same prerequisites—infrastructure visibility, policy guardrails, and real-time actuation—so the market is converging them into unified platforms. Neutral When cost automation and incident remediation merge into one product category, we're watching the reactive/preventive boundary collapse. Organizations that treat FinOps and SRE as separate budget lines will find vendors forcing the conversation—because the platform already made the choice. | Cost and reliability now ship in the same release. |
|
Qyrus · Thematic anchor—autonomous healing Qyrus frames 2026 as the year closed-loop autonomous systems move from experiment to requirement. The article argues infrastructural resilience must be engineered into systems by design rather than bolted on as reactive backup, using autonomous healing to close the alert-fatigue loop. Prevention The rhetoric is finally catching up to the product reality: nobody is pitching 'AI-assisted' anymore. If your reliability strategy still assumes a human will always be in the loop, this is the year that assumption becomes the bottleneck—and vendors will sell around you to your CFO. | Alert fatigue doesn't get fixed; it gets automated away. |
|
Cast AI · Implementation detail—guardrails Cast AI publishes the decision logic behind its automation: actions happen within user-defined limits, decisions derive from observed usage over workload variation windows, and every action is recorded for audit. Prevention Transparency in autonomous decision-making is the new trust tax—vendors that can't explain how the agent chose option B over option A won't survive procurement. | Autonomous systems earn trust through audit logs, not promises. |
|
Causely · Adjacent—RCA automation Causely's causal reasoning engine automatically identifies cause-and-effect chains between problems and symptoms in real time when performance degrades, using instrumentation to map the detailed path from root cause to observable failure. Firefighting Causal reasoning closes the gap between 'we detected it' and 'we know why'—the missing middle that keeps MTTR high even when observability spend is off the charts. | Root cause analysis that doesn't wait for the war room. |
|
Densify (Kubex) · Cultural signal—sentiment shift Viktor Farcic, a former AI skeptic, now claims there's no professional life without AI, discussing agents, the Model Context Protocol (MCP), and the rapidly changing automation landscape in a Kubex Talks episode. Neutral When vocal skeptics flip to 'no life without AI,' that's a lagging indicator the market already moved—useful for calibrating how far behind your org might be. | The skeptics have left the building. |
|
Grafana Labs · Infrastructure requirement—alert scale Grafana Alerting introduces multiple notification policies to prevent alert routing configurations from collapsing under organizational growth, addressing the problem where simple early-stage policies become unmanageable trees as teams and services multiply. Firefighting Alert routing is the plumbing nobody budgets for until it breaks—this is Grafana acknowledging that autonomous systems produce more alerts, not fewer, and you need routing architecture that scales with agent activity. | More automation means more alerts, not fewer. |
|
ilert · Ecosystem indicator—integration activity Bleemeo monitoring now connects natively to ilert, routing threshold breaches through escalation policies to on-call responders and closing incidents automatically once issues clear. Firefighting The integration economy around autonomous healing is heating up—every monitoring tool needs a native path to incident management, because manual ticket creation is now the bottleneck agents expose. | Integration velocity signals where automation is landing. |
|
|
The Challenger — Vendor Landscape
On-call intelligence platforms split cleanly on the U-Curve.
Prevention side: incident.io surfaces recurring failure patterns from past incidents to prevent repetition. Rootly automates retrospective workflows so findings reach the team that can actually fix things.
Reaction side: PagerDuty dominates alert routing and escalation. Fastest path from signal to engineer. Not the path that reduces signal count.
The gap: None of these platforms reports your prevention-to-firefighting ratio. Measuring response time is solved. Measuring whether you're preventing incidents is still manual.
Neutral
| Reaction tooling solved. Prevention measurement still manual. |
| Should this replace the favorite? | |
|
|
The Ratio is a weekly newsletter by Florian Hoeppner. Early edition, you are one of the first.
Forwarded this? Get The Ratio in your inbox every week.
|
Take the assessment → reliabilityeconomics.com/benchmark Reply to this email with your take.
How this newsletter gets better: Every link is tracked and every vote is counted. Sections that get a down vote get replaced. Topics that get an up vote stay. No personal data is stored — just click and vote counts per section. Your feedback shapes the next edition.
|
|
|