The AI SRE landscape 2026: 66 tools evaluated

Last verified: July 2026.

In January 2026, Gartner published its first Market Guide for AI Site Reliability Engineering Tooling and predicted 85% enterprise adoption by 2029, from under 5% today. The market moved fast to validate that forecast: Resolve AI hit a $1.5B valuation in April, Dash0 raised $110M at a $1B valuation in March, Snowflake agreed to pay roughly $1B for Observe in January.

We put together this guide while mapping what teams are actually building and buying in this space — and specifically what each category of AI SRE tool needs from the observability data underneath it. The list covers 66 tools across 7 camps. It's not a ranking. The goal is a structured view of who is doing what and how to think about evaluating them.

Two community resources are worth knowing alongside this: awesome-ai-sre, also browsable as The AI SRE Watchlist, is a curated list that updates continuously; ai-sre.com is a community-driven comparison database. Both catch new entrants faster than a static guide will.

What "AI SRE" actually means

Ask ten vendors and you get ten different answers. They're genuinely solving different problems.

Autonomous incident response (Camp 1). An agent wakes on an alert, queries your observability stack, and produces a root-cause narrative. Most of the funding is going here. These tools sit on top of your existing observability infrastructure and investigate from the outside.

Observability with AI built in (Camp 2). A monitoring platform adds an investigation agent on top of its own data. The AI never crosses a network boundary to get telemetry, so it's faster and has richer context than an outside agent hitting the same APIs. More depth in exchange for deeper lock-in.

Incident management with AI on top (Camp 3). Purpose-built tools for the human side of incidents: who is doing what, what happened, and what do you tell the organization. The AI here is mostly summarization, postmortem drafting, and noise reduction. For diagnosis, you still need something from Camp 1 or 2.

AIOps and correlation (Camp 4). Event correlation at high volume — cutting tens of thousands of alerts to a manageable set of actionable incidents. Proven at enterprise scale. The pre-LLM correlation engine is often the mature core; the LLM assistant layer is newer and varies in depth.

Hyperscaler-native agents (Camp 5). AWS DevOps Agent and Azure SRE Agent, deeply integrated with their respective cloud primitives. Native integration depth that third-party agents cannot replicate through the same APIs.

Open-source Kubernetes copilots (Camp 6). Free, self-hostable diagnostic tools for K8s environments. Strong starting point for platform teams; diagnostic rather than autonomous.

Adjacent autonomous optimization (Camp 7). Tools that prevent a class of incidents before they happen: Kubernetes cost and resource optimization (CAST AI, Sedai), IaC drift remediation (Firefly), pipeline failure diagnosis (GitLab Duo).

This spectrum matters when evaluating vendors. A Camp 7 IaC drift-remediation agent and a Camp 1 on-call bot will both be pitched to you as "AI SRE." They're solving different problems from different data with different failure modes.

Build, extend, or buy

The choice isn't binary. Most teams land somewhere in the middle.

Full DIY. Maximum control; you pick a framework, write your own integrations, connect your own backend. The ops burden is real: agent reliability at 3 a.m., prompt versioning, LLM rate limits under incident load, context window management. It can be done — Vercel runs 100+ production agents on their own open-source TypeScript framework, eve. It takes serious engineering.

Build on open source. Proven foundations without starting from scratch: HolmesGPT or K8sGPT for Kubernetes investigation, Aurora for multi-cloud RCA, Keep for alert management. Smaller ops burden than DIY, with code you can inspect before trusting.

Buy, then extend. Commercial product for the core, custom integrations for the gaps. If the tool you buy doesn't connect natively to your observability backend, Bronto's MCP server gives any MCP-compatible agent direct access to your logs and traces without a custom connector.

Buy and run. Commercial product adopted as-is. Vendor's design decisions in exchange for skipping the build cost entirely.

Frameworks worth knowing for the build-heavy end: Claude Agent SDK (Python and TypeScript, runs against Anthropic's API, Bedrock, Vertex AI, and Microsoft Foundry); LangGraph (stateful agent orchestration, MIT-licensed, GA May 2025); MCP (open standard, Linux Foundation-governed since December 2025); Amazon Bedrock AgentCore (managed runtime for production agents, GA October 2025).

How to evaluate any tool

Ignore the word "autonomous." Ask four questions instead.

Where does it get its data? On top of your existing tools, or inside its own data lake? Portable versus deeper. The data layer matters more than the agent logic: sampled data, short retention, and slow search all produce confidently wrong answers.

Does it act, or only advise? Read-only is easy to adopt. Action-capable needs real guardrails — RBAC, rollback, blast-radius limits. Know which you're buying before granting access.

Can you see its reasoning? An explanation you can validate beats a confident black-box verdict every time. Tools that show their work are worth trusting when it actually matters.

What happens when it's wrong? Every tool on this list will be confidently wrong about something. Does it fail in a way an engineer will catch, or does it produce a wall of "analysis" that hides the uncertainty? Test failure modes before go-live, not just the happy path.

One more: will the company still exist in two years? In this list alone, four vendors were acquired within roughly a year (Squadcast, Zenduty, FireHydrant, Keep), one unicorn is being absorbed by Snowflake, and one vendor's website went dark while we were writing this. Wiring a startup into your 3 a.m. workflow is a business risk, not just a technical one.

The telemetry layer underneath

Every tool in Camps 1–7 reads from something we haven't put on this list: the telemetry layer. That's Bronto's job.

What AI agents need from the data layer is different from what humans need looking at dashboards. Complete data — no sampling, because the rare event they miss may be the root cause. Long retention — 12 months of history separates "this pattern happened before" from "no data found." Fast search — sub-second across petabytes, because a slow query during an incident costs time. Structured, discoverable data — an agent guessing your field names produces worse answers than one that can explore your schema directly.

We documented this concretely when we ran AWS DevOps Agent against different observability backends: Bronto scored 9.4/10 versus 5.3 and 4.6 for the alternatives, specifically because of data discovery, how each system handled silent query failures, and response completeness.

Camp 1: Pure-play AI SRE agents (Tools #1–25)

These tools sit on top of your existing observability stack and investigate from the outside. Teams choose this camp when they've invested in an observability stack and don't want to replace it. The core tradeoff: these tools are only as good as the signals your existing stack exposes. Most agents in this camp need broad read access across code repos, CI/CD systems, and cloud APIs, not just metrics and logs. Plan for that access surface.

1.Resolve AI

Autonomous incident responder founded by OpenTelemetry co-creators. Reads source control, deploy history, config, and telemetry to connect symptoms to code- and infra-level changes rather than stopping at surface signals. Published results: DoorDash (87% reduction in time to root cause, investigation time from roughly 40 minutes to under 1 minute), Coinbase (72–73% faster time to root cause), and Zscaler (roughly 30% fewer engineers per incident) — confirmed in TechCrunch, Forbes, and SiliconANGLE. Raised $125M at a $1.5B valuation. Enterprise-only with no self-serve path; all telemetry routes through Resolve's cloud infrastructure, which is a blocker for regulated industries with data-residency requirements. Setup typically involves several weeks and solutions-engineering engagement. resolve.ai

2.Cleric

An "AI SRE teammate" that takes the first pass on every alert, sweeps logs, metrics, traces, and changes, and drops a diagnosis into Slack. Gartner named it a Cool Vendor in AI for SRE and Observability 2025. Transparent reasoning: it ranks hypotheses, tests each against real data, and learns your failure patterns over time. Most specific public reference: BlaBlaCar, handling roughly 1,400 Kubernetes alerts per month. Read-only by design. $9.8M total raised; no self-serve evaluation path. cleric.ai

3.Traversal

Causal-ML root cause analysis built for petabyte scale. American Express made a $5M strategic investment and is a production customer (confirmed by a named Amex Ventures executive in SiliconANGLE). Other confirmed external customers: DigitalOcean, Eventbrite, Cloudways, and Kraken. Raised $48M from Sequoia and Kleiner Perkins. Causal rather than purely correlational reasoning; accurate output depends on an accurate service topology. No built-in incident lifecycle tooling. No public pricing. traversal.com

4.NeuBird (Hawkeye/Falcon)

Agentic AI SRE for enterprise IT. Gartner named NeuBird a 2025 Cool Vendor in IT Operations Leveraging Generative AI and listed it in the 2025 Hype Cycle for AI in IT Operations. Microsoft published a how-to for integrating Hawkeye via MCP with Azure SRE Agent; Datadog published official integration documentation. Triage Agent runs continuously at no charge; credits consumed on active investigations (roughly $25/investigation). $63.8M total raised. neubird.ai

5.Ciroos

Multi-agent AI SRE built on open interop standards (MCP and agent-to-agent protocols). Founded February 2025 by former Cisco/AppDynamics executives; $21M seed led by Energy Impact Partners. Constellation Research Shortlist 2025; SiliconANGLE 2026 CUBEd Award for Most Innovative Observability/Monitoring Platform. The open-standards approach makes it extensible across existing tools rather than a walled garden. Very early in market penetration. ciroos.ai

6.Deductive AI

AI SRE agents using a code-aware knowledge graph linking codebase structure with telemetry. Named customers confirmed in TechCrunch: DoorDash, Foursquare, Kumo AI, and Apoha. Elastic agreed to acquire Deductive for up to $85M in June 2026; product direction and availability will change during integration. Code-awareness requires codebase access; evaluate against your security posture. deductive.ai

7.Causely

Causal-AI engine maintaining a live causality graph for root cause identification in Kubernetes environments. Gartner Cool Vendor for AI in IT Operations (December 2025). Public AWS Marketplace pricing: $24,000/year for up to 500 services. Diagnosis-first, no-side-effects architecture — identifies root causes but does not execute remediation. Accurate output requires an accurate service dependency map. causely.ai

8.Wild Moose

AI first responder from alert to root-cause candidate. Public customers: Wix (30,000+ alerts/month enriched), Redis (investigation time from 20 minutes to under 1 minute), GoFundMe, and Lemonade. Processes telemetry in-memory without persisting customer logs (SOC 2 compliant). $7M seed; YC W23. Team is roughly 10 people; factor into enterprise support expectations. wildmoose.ai

9.TierZero AI

Autonomous incident investigation and production remediation. Confirmed customers: Brex, Discord, and Drata (Drata case study: 42% MTTR reduction, 7,000+ engineering hours saved per year). Acquired by Cognition (maker of Devin) on July 20, 2026. Evaluate as an acquisition integration, not a standalone product. tierzero.ai

10.DrDroid (Doctor Droid)

AI SRE agent with investigation knowledge graph, PlayBooks automation, and a Slack AlertOps bot. One of the few tools with transparent public pricing: Teams plan at $99/month for 99 investigation credits, unlimited users and integrations. Bring-your-own-LLM option (OpenAI, Anthropic, Bedrock, or local) is rare in this category and valuable for regulated industries. Open-source PlayBooks on GitHub. YC W23. drdroid.io

11.Bacca.AI

A "virtual SRE" that forms hypotheses from architecture, incident history, and institutional memory, then validates them with read-only queries. Available on AWS and Google Cloud marketplaces. Base pricing $60,000/year. Hypothesis-first investigation produces targeted analysis rather than raw telemetry summaries. Very early stage; verify current status directly before committing. bacca.ai

12.Sherlocks.ai

AI-native SRE assistant built around an "institutional memory" awareness graph. G2 4.9/5 across 27–28 reviews — one of the few Camp 1 tools with any G2 presence. Free tier: 30 investigations per month, no credit card required. Publishes their own agent success rate over time in their changelog (from 35% to roughly 75% by early 2026) — an unusual level of transparency. India-based team; customer base concentrated in Indian startups. sherlocks.ai

13.Hyground

Hamburg-based AI SRE built for data sovereignty: fully self-hosted, air-gap capable, bring-your-own-model, zero external data transmission. The only pure-play on this list built for GDPR and regulated-industry constraints from day one. Confirmed customers: Deutsche Bahn (Rail Information Systems division) and ifm (German industrial automation). EUR 3M pre-seed led by Partech. Self-hosting means you carry the ops burden and model inference cost. hyground.ai

14. IncidentFox

Open-core, self-hostable AI SRE agent. YC W26. GitHub repository archived May 31, 2026; the product has likely shut down or pivoted entirely. Do not evaluate as a current product.

15.Kestrel AI

AI agents for platform engineering: incident response, cloud provisioning, CI/CD, and developer requests as deterministic workflows. Public pricing: Workflows ~$0.50/execution, Incident Response $200/seat/month, On-call $50/user/month. 140+ pre-built actions; one-click YAML fixes for Kubernetes incidents. YC Fall 2025. usekestrel.ai

16.Parity

Read-only Kubernetes incident response agent. Self-published SREBench benchmark: 70% incident success rate at a 2-minute MTTR target. RAG over your runbooks is useful for teams with strong runbook discipline. Kubernetes-only. YC S24. Very limited public activity through most of 2026; verify current roadmap status before committing. tryparity.com

17.Komodor (Klaudia)

Kubernetes operations platform with Klaudia, a multi-agent AI SRE for K8s troubleshooting. G2 4.4/5 from roughly 40 verified reviews. Representative Vendor in Gartner's inaugural AI SRE Market Guide (January 2026). Confirmed enterprise customers: Nebius (NASDAQ: NBIS) and Qualcomm. Klaudia Memory (July 2026) adds cross-investigation learning and runbook/postmortem Knowledge Base. Kubernetes-only; no native log management or distributed tracing. komodor.com

18.Kura

AI DevOps copilot for AWS infrastructure management: runbooks, provisioning, log monitoring, cost monitoring, and security checks via chat. AWS-only at launch; Azure and GCP listed as coming soon. YC S24. In pilot/waitlist mode with no general availability yet. usekura.com

19.Beeps

On-call platform routing incidents to both developers and AI coding agents (Devin, Cursor Cloud Agent, AWS DevOps Agent, Claude Code, Codex). Treats AI agents as first-class on-call participants on the same footing as humans. Pricing: $70 per active relay per month. Genuinely ahead-of-the-curve design; very early product. beeps.dev

20.NudgeBee

Agentic platform for SRE, FinOps, and Kubernetes operations with pre-built assistants and customizable workflows. Rackspace is a confirmed named customer (CTO quoted describing 1,200+ automated remediation workflows at scale). Open-source self-hosted tier; cloud plans from $150/month. $3M seed from Kalaari Capital. nudgebee.com

21.Kubiya

Agentic engineering platform driven by natural-language Slack and Teams commands, executing Terraform, CI/CD, Kubernetes management, cloud cost reporting, and access grants. Gartner Cool Vendor in Platform Engineering (April 2024). Confirmed enterprise customers: Aston Martin and A+E Networks. Action-capable — executes tasks against real infrastructure — which makes RBAC evaluation load-bearing. Website was offline at the time of research; verify vendor status before committing. docs.kubiya.ai

22.SRE.ai (rebranding to Laminar)

AI DevOps agents for Salesforce ecosystem workflows: deployments, CI/CD, and testing. $7.2M seed co-led by Salesforce Ventures. A real niche for a genuinely painful workflow most of this list ignores. Not a general production-incident SRE agent. sre.ai

23.RunWhen

SRE automation platform discovering Kubernetes and cloud infrastructure, generating pre-built "CodeBundle" skills exposed via MCP server. MCP-first: AI agents can invoke RunWhen's troubleshooting skills without custom integration. CodeBundles are inspectable and owned by your team. Public pricing: 5,000 sessions/year at $25K. Last blog post May 2025; confirm the product is active before committing to annual pricing. runwhen.com

24.AlertD

Agentic AI SRE for AWS. A background daemon continuously indexes your cloud environment, surfacing AI agents for cost, security, compliance, performance, and troubleshooting. VPC-resident; your data doesn't leave your account. SOC 2 Type II certified; AWS Marketplace listed. Founder is former Principal Architect at Splunk and Chief Architect at AppDynamics. $3M seed (True Ventures). AWS-only. alertd.ai

25.NOFire AI

AI SRE platform built around a live "Production Context Graph" connecting services, dependencies, code changes, deployments, and past incidents. Representative Vendor in Gartner's 2026 AI SRE Market Guide. SOC 2 Type II and GDPR compliant. Deployment-risk prediction (catching bad deploys before production) is the main differentiator to evaluate. $2.5M seed from Marathon Venture Capital. nofire.ai

Camp 2: Observability platforms with AI built in (Tools #26–42)

These platforms have built the AI directly into their own data store, so the agent never crosses a network boundary to get telemetry. The fundamental advantage is completeness of context. The tradeoff is lock-in, and it compounds over time. Before committing, model what an active incident month looks like on the platform's pricing model, not just the baseline subscription.

26.Datadog Bits AI SRE

Coordinated agents that fire on alerts and pull telemetry across every connected Datadog product. Gartner MQ Leader for the sixth consecutive year, highest by Ability to Execute in 2026. G2 4.4/5 from 691+ reviews. Confirmed enterprise customers: Uber Freight and iFood (70% MTTR reduction). Pricing: $500/20 investigations/month on annual commitment. Structural constraint: Bits AI can only investigate signals within Datadog — Grafana, Sentry, Splunk, and Prometheus that don't flow in are hard blind spots. datadoghq.com

27.Dynatrace Davis AI

Full-stack observability with deterministic causal-AI and Davis CoPilot for natural-language remediation. Gartner MQ Leader for 16 consecutive years. G2 4.4/5 across roughly 9,200 aggregated reviews. Confirmed press-covered customers: BMO (80% MTTI and RCA reduction), TD Bank (45% cost reduction). Stale Smartscape topology causes confident-but-wrong root cause attributions — calibration is required before the AI is reliable. dynatrace.com

28.New Relic AI (NRAI / SRE Agent)

Full-stack observability with NRAI and an SRE Agent launched February 2026. Gartner MQ Leader for 13 consecutive years. G2 4.4/5 from 598 reviews. Well-known confirmed customers: Starbucks, Home Depot, Lowe's. Free tier (100GB/month, no credit card) is a genuine evaluation path. Billing opacity and forced sampling to manage costs are the most documented practitioner complaints. newrelic.com

29.Grafana Assistant

Open-source-rooted observability with Assistant, Sift, and Asserts/RCA Workbench. Gartner MQ Leader 2024–2026, furthest in Completeness of Vision. G2 4.5/5. 25M+ open-source users. Confirmed enterprise customers: Dell, Microsoft, Salesforce, Citigroup, Anthropic, NVIDIA. Free tier available. AI features are augmentive rather than autonomous. grafana.com

30.Splunk AIOps (Cisco)

Enterprise observability plus ITSI with ML-based anomaly detection. Part of Cisco since March 2024 ($28B acquisition). Gartner MQ Leader for Observability Platforms and SIEM. AI capability is distributed across products from growth-by-acquisition history; pin down which specific product carries what you're evaluating. splunk.com

31.Elastic AI Assistant

Natural-language querying of logs, metrics, and traces in Kibana with RAG-backed knowledge base. GA September 2023. Also acquiring Keep (Camp 4) and Deductive AI (Camp 1). RAG over your own runbooks is a meaningful grounding feature. Enterprise subscription required. Elastic's own documentation states that data sent to the assistant is stored by the third-party AI provider — evaluate against your compliance requirements. elastic.co

32.Honeycomb

Event-based observability with Query Assistant, Honeycomb Intelligence, and AI-guided Canvas. G2 4.5–4.7/5. Gartner MQ Leader 2022–2024. Confirmed press customers: Stripe, Slack, Vanguard, Vercel. Hosted MCP server. AI augments Honeycomb's strong high-cardinality query model; features are query assistance and anomaly surfacing, not autonomous incident investigation. honeycomb.io

33.Dash0 Agent0

OpenTelemetry-native observability with specialized agents for trace explanation, PromQL drafting, dashboard building, and instrumentation-gap analysis. G2 4.8/5 from 42 reviews; 93% five-star on AWS Marketplace — highest-rated observability platform on this list by review score. Confirmed customer: Telegraph Media Group. Consumption pricing: $0.20/M metric data points, $0.60/M spans and log records; no base subscription. Founded by the creator of Instana. Agent0 is augmentation, not autonomous incident response. dash0.com

34.Observe (by Snowflake)

Unifies logs, metrics, traces, and business context into an open data lake (Apache Iceberg + OTel) with an AI SRE agent for automated investigation. Snowflake acquisition for roughly $1B (January 2026). Gartner Peer Insights 4.5/5. Confirmed customers: Capital One, Tekion, Topgolf/Toptracer. AWS Marketplace entry: $20,000/year. Proprietary OPAL query language. observeinc.com

35.Metoro

eBPF-based Kubernetes observability with an AI SRE layer ("Guardian"). Single Helm install, no code changes. Complete telemetry from one install without instrumentation. Very early stage: approximately 3 employees, seed-stage. Kubernetes-only. metoro.io

36.Middleware OpsAI

Full-stack observability with OpsAI that detects, diagnoses, and ships fixes as a pull request. G2 4.7/5 from roughly 55 reviews. OpsAI went GA May 2026. Can read from Datadog and Grafana without requiring a full migration. YC W23; $6.5M seed (8VC). middleware.io

37.Mezmo AURA

Telemetry pipeline company (formerly LogDNA) with AI SRE agent AURA on an open-source agentic harness. G2 4.6/5 from 224 reviews for the Mezmo platform. IBM Cloud is a confirmed long-standing customer. A mature telemetry pipeline underneath the agent is a genuine advantage. AURA was still in beta with design partners at research time. mezmo.com

38.Coroot

Open-source observability with eBPF auto-instrumentation and AI-powered RCA. Free Community edition; Pro at $100/month for 5 nodes. Actively developed; low adoption risk for self-hosting teams. eBPF depth varies by kernel version and cloud environment. coroot.com

39.Last9

Unified observability with an Agentic SRE SDK and MCP integration with AWS DevOps Agent (June 2026). G2 4.7/5 from 51 reviews. Gartner Cool Vendor in AI for SRE and Observability 2025. SDK-first approach lets you wire agentic SRE into your own workflows. Pairs with PagerDuty or similar for on-call. last9.io

40.SigNoz

Open-source, OTel-native platform for logs, metrics, and traces. Free Community edition (self-hosted); Teams from roughly $199/month cloud. 20,000+ GitHub stars. No vendor lock-in. AI and agent features are earlier-stage than the incumbents. signoz.io

41.Lightrun

AI SRE platform for autonomous code remediation. Adds logs, snapshots, and metrics to running production code without redeploys. G2 4.8/5 from 67 reviews. Gartner Cool Vendor. $70M Series B (Accel, Insight, April 2025). Owns a runtime-debugging capability unique on this list. Currently limited to Java, JavaScript, Python, and .NET. lightrun.com

42.Sentry Seer

AI debugging agent built on production error telemetry. Identifies actionable issues, does RCA, and generates code fixes. Sentry platform G2 4.5/5 from 1,280+ reviews. Strongest for app-level bugs; weaker on infra-level and systemic incidents. Requires Sentry for error monitoring. sentry.io

Camp 3: Incident management with AI on top (Tools #43–49)

These tools handle the human side of incidents: coordination, communication, and postmortems. The AI is mostly summarization, timeline reconstruction, noise reduction, and postmortem drafting. For root cause analysis, you still need something from Camp 1 or 2. The best setups pair a Camp 1 agent for diagnosis with a Camp 3 tool for the incident workflow. Three of the seven tools below were acquired within roughly a year of each other.

43.PagerDuty AIOps + SRE Agent

Enterprise incident management with ML noise reduction and an AI Agent Suite. G2 4.5/5 from 900+ reviews — the largest review corpus in this camp. The incumbent on-call and routing backbone. The SRE Agent is newer; confirm diagnostic depth beyond summarization in your evaluation. Pricing for AIOps add-on requires sales engagement. pagerduty.com

44.incident.io

Slack-native incident management with AI SRE, AI alert triage, AI postmortems, and Claude and Cursor integration. G2 4.8/5 from roughly 180 reviews. Confirmed named customers: Netflix, Etsy, Airbnb, Canva. Best-loved Slack-native workflow in this category. Platform lock-in to Slack or MS Teams. incident.io

45.Rootly

AI-native incident management analyzing code changes, telemetry, and past incidents to surface probable root causes with confidence scores. G2 4.8/5 from roughly 77 reviews. On-call available standalone or bundled. Visible reasoning chain with ranked hypotheses is the standout feature. On-call is a paid add-on; factor that into total cost for full incident lifecycle coverage. rootly.com

46.FireHydrant (Freshworks)

Incident response with AI summaries and AI-drafted retrospectives. Acquired by Freshworks December 2025. G2 4.5/5 from roughly 143 reviews. Confirmed enterprise customers: Palo Alto Networks, BP, Qlik. Roadmap will increasingly serve Freshservice integration needs. firehydrant.com

47.SolarWinds Incident Response (formerly Squadcast)

Incident management with AI-driven alert clustering. Acquired by SolarWinds March 2025; rebranded March 2026. G2 roughly 4.5/5. Confirmed customer: Redis. Evaluate as part of the SolarWinds ecosystem, not as an independent product. solarwinds.com

48.Xurrent IMR (formerly Zenduty)

On-call and incident management with AI summarization and postmortems. G2 4.6/5 from roughly 141 reviews. Starter at $6/user/month — one of the lowest entry points in this camp. Acquired by Xurrent February 2025. zenduty.com

49.BetterStack

Developer-friendly uptime monitoring, on-call, log management, and incident management in one product. G2 4.8/5 from 303 reviews. Free tier available; on-call from $20/responder/month. Easy to set up; genuinely useful for developer-first teams. AI features are modest compared to dedicated AI SRE tools. betterstack.com

Camp 4: AIOps and correlation platforms (Tools #50–55)

These serve large organizations where high alert volume is the primary problem. Event correlation that cuts tens of thousands of alerts to a manageable set of actionable incidents is transformative on its own. The pre-LLM correlation engine is often the mature core; ask specifically what the LLM layer adds beyond what the pre-LLM engine already did.

50.BigPanda

AIOps for high-alert-volume environments: event correlation cutting alert volume by 95%+, plus an AI Incident Assistant. Acquired AI SRE startup Velocity (November 2025). G2 4.5/5 from 134 reviews. Proven at enterprise scale in large financial services and telecom. No native telemetry collection. bigpanda.io

51.Dell APEX AIOps (formerly Moogsoft)

Event dedup, correlation, and automated RCA. Confirmed enterprise customers: American Airlines, Fannie Mae, Fiserv. One of the original AIOps names with a strong correlation engine. AWS Marketplace: $39,286/year for up to 50,000 events/month. dell.com

52.LogicMonitor Edwin AI

Cloud infra monitoring with Edwin AI for plain-language summaries, predictive analytics, and capacity forecasting. G2 4.5/5 from roughly 829 reviews. Gartner Visionary in Observability MQ. Infra-monitoring roots mean stronger on hosts and network than on application-layer distributed tracing. logicmonitor.com

53.Selector AI

AI network observability with a "Network LLM" and roughly 90% alert-noise reduction. $32M at $375M valuation (February 2026). Gartner Cool Vendor 2023. Real network specialization in a domain most AI SRE agents ignore. Right for NetOps teams; limited value for application-layer incidents. selector.ai

54.Keep (Elastic)

Open-source AIOps and alert management with 50+ integrations in a "GitHub for alerts" workflow-as-code style. Acquired by Elastic May 2025. Three pricing tiers: Starter (free, self-hosted), Team ($100/month), Enterprise (custom). The Elastic acquisition gives it a stable corporate home. keephq.dev

55.OtterMon AI

Cross-vendor DevOps intelligence platform training on each customer's environment and incident history. Founded by a former New Relic Director of Product Management. Vendor-agnostic posture is the differentiator. Very early stage: $250K angel round, roughly 6 employees, private beta as of mid-2026. A company to watch. ottermon.ai

Camp 5: Hyperscaler agents (Tools #56–57)

For teams deeply committed to one cloud provider. The native integration advantage is real; a third party can integrate with the same APIs but not with the same depth or latency. The structural constraint is also real: these agents are single-cloud by design.

56.Azure SRE Agent

AI agent for monitoring, diagnosing, and resolving issues in Azure-hosted apps with a no-code sub-agent builder. GA March 2026. Microsoft uses it internally: 1,300+ agents running, 35,000+ incidents mitigated on Microsoft's own estate. Native Azure Monitor, Resource Graph, and Service Health integration. Azure-only. azure.microsoft.com

57.AWS DevOps Agent

AWS's production operations and release management agent. Announced re:Invent 2025; production-operations GA April 2026. Deepest native integration with CloudWatch, Systems Manager, CodeDeploy, and every AWS API. Agent credits bundled with AWS Support plans. We published a benchmark of what AI agents need from a telemetry layer when running on AWS. AWS-only. aws.amazon.com/devops-agent

Camp 6: Kubernetes copilots and open-source building blocks (Tools #58–62)

For K8s-native teams and engineers who want to read the code before trusting the agent. Cost floor is low, often free; transparency is high. Primarily diagnostic: they investigate the K8s layer and surface a finding. For autonomous remediation you need something alongside them.

58.K8sGPT

CNCF project for AI-powered Kubernetes diagnostics. Multiple LLM backends including local models via Ollama — valuable for air-gapped or data-sovereignty-constrained environments. 6,000+ GitHub stars. Free, CNCF-backed, bring-your-own-model. Kubernetes-scoped; surfaces problems but does not remediate. github.com/k8sgpt-ai/k8sgpt

59.HolmesGPT

Open-source AI agent (Apache 2.0, CNCF Sandbox since October 2025) running an agentic investigation loop over Prometheus, Grafana, Datadog, PagerDuty, and Kubernetes. 2,000+ GitHub stars. Can be triggered by PagerDuty alerts to auto-investigate in your existing workflow. Production issues get GitHub issues, not an SLA. github.com/HolmesGPT/holmesgpt

60.Aurora (Arvo AI)

Apache-2.0 open-source agentic incident management and RCA. LangGraph agents investigate across AWS, Azure, GCP, and Kubernetes with PagerDuty, Datadog, Grafana, and Slack integrations. Broader multi-cloud scope than the K8s-only tools. Started January 2026; actively developed. github.com/Arvo-AI/aurora

61.Lens Prism

AI-powered Kubernetes copilot built into Lens IDE. Answers natural-language questions from live kubectl output, Prometheus metrics, and logs; generates copy-pasteable commands. MCP server since March 2026. Included in Lens paid plans at no extra cost. A productivity copilot rather than an autonomous incident responder. lenshq.io

62.Kube-Copilot

Open-source Kubernetes copilot CLI written by a Kubernetes maintainer. Bring-your-own-model; kubectl and trivy under the hood. Simple, free, good starting point. No commits since November 2025; fine as a personal tool, not a foundation for a production workflow. github.com/feiskyer/kube-copilot

Camp 7: Adjacent tools — autonomous optimization to lifecycle AI (Tools #63–66)

These tools prevent a class of incidents rather than investigating them. The right mental model: Camp 7 handles conditions that would have caused incidents; Camp 1 handles the incidents that happen anyway. They complement each other rather than compete.

63.CAST AI

Kubernetes cost and performance optimization with "agentic runbooks" for autonomous operational decisions with human approval. Unicorn as of January 2026. Representative Vendor in Gartner's 2026 AI SRE Market Guide. Confirmed customers: BMW, Cisco, FICO. Prevents a class of performance-related incidents through continuous rightsizing. Center of gravity is optimization, not root-cause investigation. cast.ai

64.Sedai

"Self-driving cloud": autonomous platform using reinforcement learning to optimize cloud resources for cost, performance, and availability. $20M Series B (June 2025). 25M+ production actions; Palo Alto Networks case study: $3.5M saved, 2M+ autonomous remediations per year. Adjacent by Sedai's own positioning. sedai.io

65.Firefly Thinkerbell AI

Cloud asset management and IaC platform with AI agents that remediate drift and policy violations and can rebuild environments autonomously. $23M Series A. Named in Gartner's 2026 AI SRE Market Guide. IaC-native: catches drift before it becomes an incident. Pair with a Camp 1 tool for full incident lifecycle coverage. Review approval workflows carefully before enabling Autopilot mode. firefly.ai

66.GitLab Duo Agent Platform

GitLab's AI-native agent layer across the SDLC. GA since GitLab 18.8 (January 2026). SRE-relevant features: pipeline failure diagnosis and auto-fix, Root Cause Analysis for CI/CD job failures. Included in GitLab Premium and Ultimate subscriptions. Scope is the development lifecycle — it fixes your pipeline, not your outage. Value is proportional to how much of your incident root cause originates in deployments and pipeline failures. docs.gitlab.com

Keep watching

A list like this starts aging the day it's published.

• ai-sre.com: community-driven, continuously updated.
• awesome-ai-sre, also browsable as The AI SRE Watchlist.

For a deeper framework on the full spectrum of what "AI SRE" can mean, the CNCF whitepaper on levels of service reliability automation is in progress and open for collaborators.

A more detailed version of this guide is available as a six-part series on the Bronto blog: part 1 (framework), parts 2–3 (pure-play agents), part 4 (observability with AI built in), part 5 (incident management and AIOps), and part 6 (hyperscalers, K8s copilots, and adjacent tools).

If you're building an AI SRE tool and want the best possible observability backend for it, that's what Bronto is for. Reach out.