Amazon Web Services
Deepest cloud coverage: dozens of service scanners for cost, security and reliability findings. NudgeBee runs the AWS CLI in a sandbox to investigate incidents and applies approved fixes.
Integrations
Plugs into the stack you already run
Read in place, in each backend's own query language. Nothing is shipped out to a third party.
What does NudgeBee integrate with?
NudgeBee connects to the tools you already run and queries them in place, in each backend's own query language. It integrates with 78 systems across cloud, Kubernetes, observability, ticketing, ChatOps, source control, databases and AI models, including Prometheus, Datadog, New Relic, Loki, Jaeger, PagerDuty, Jira, Slack, GitHub and OpenTelemetry, with nothing shipped to a third party.
Cloud & Kubernetes
One collector and one data model across AWS, Azure, GCP and every flavour of Kubernetes.
Microsoft Azure
Full inventory, cost and security scanning across 50+ Azure services, folding in Azure Advisor recommendations. Investigated with the az CLI and remediated behind approval.
Google Cloud
Inventory, cost and rightsizing across GCP, enriched with GCP Recommender. NudgeBee investigates with the gcloud CLI and opens a fix behind a human gate.
Kubernetes
A live topology of every workload, pod and node, kept current as resources change. NudgeBee runs kubectl and helm in a sandbox to diagnose and, with approval, remediate.
Amazon EKS
Managed Kubernetes on AWS, with node hardware, capacity and spot-versus-on-demand placement detected automatically for rightsizing.
Azure AKS
Managed Kubernetes on Azure, mapped into the same topology and cost model as every other cluster you run.
Google GKE
Managed Kubernetes on GCP, with fleet rightsizing and node placement handled alongside your EKS and AKS clusters.
Red Hat OpenShift
On-premises and hybrid OpenShift clusters run through the same in-cluster agent and topology model as any other Kubernetes distribution.
Rancher
Rancher-managed clusters, on-prem or in the cloud, are supported through the standard Helm-chart agent install.
AWS Fargate
Serverless containers on AWS, inventoried and cost-tracked in the same recommendation taxonomy as your clusters.
Amazon ECS
Container workloads on ECS, with rightsizing and configuration findings surfaced alongside the rest of your AWS estate.
AWS Lambda
Serverless functions are inventoried and analysed for cost and configuration alongside your containerised services.
Cloud Foundry
Cloud Foundry platforms are collected into the same multi-cloud inventory and cost model as AWS, Azure and GCP.
Metrics
Queried in place with PromQL and each backend's native dialect. Prometheus resolution is auto-tuned to the time range.
Prometheus
The primary metrics source. NudgeBee writes and runs PromQL against your Prometheus in place, tuning step resolution to the query window.
VictoriaMetrics
A drop-in PromQL-compatible backend, queried the same way as Prometheus for metrics investigation and rightsizing.
Grafana Mimir
Long-term, horizontally scalable Prometheus storage, queried in PromQL for historical baselines and anomaly detection.
Chronosphere
Cloud-native observability, queried in PromQL for metrics and used as a trace backend during investigations.
Last9
High-cardinality observability, read in place in its PromQL-compatible dialect for metric queries during an investigation.
Amazon CloudWatch
Native AWS metrics and alarms are ingested and correlated with the rest of your signals, and CloudWatch alarms can be created during remediation.
Azure Monitor
Azure-native metrics and logs, queried in KQL and folded into the same topology and correlation as your other backends.
Google Cloud Monitoring
GCP-native metrics are ingested for correlation and used as inbound alert sources during triage.
Logs
Read where they live, in LogQL, Elasticsearch DSL, ClickHouse SQL and more, and pulled just-in-time around an incident.
Grafana Loki
Log aggregation queried in LogQL. NudgeBee pulls the exact log lines before a crash or error to build the evidence chain.
Elasticsearch
Log and event search queried in Elasticsearch DSL, correlated against metrics and traces during an investigation.
SigNoz
Open observability for logs and traces, read in place via ClickHouse SQL as part of the investigation toolset.
Logstash
Log pipelines feeding Elasticsearch are part of the same log-analysis flow used to reconstruct what happened before an incident.
Loggly
Cloud log management, read in place and correlated with the rest of your signals during triage.
Google Cloud Logging
GCP-native logs pulled into an investigation without moving them out of your project.
Azure Application Insights
Application telemetry queried in KQL, used for logs and traces during an Azure-side investigation.
ClickHouse
High-volume logs and traces queried in ClickHouse SQL, including as the store behind SigNoz and OpenTelemetry.
Traces & APM
Distributed traces and APM suites feed the topology graph as behavioural edges, so correlation follows real call paths.
Jaeger
Distributed tracing queried during an investigation, with trace co-participation feeding the causal correlation model.
Grafana Tempo
Scalable trace storage in the Grafana stack, read in place to reconstruct request paths across services.
OpenTelemetry
Vendor-neutral traces and metrics, collected by the in-cluster agent and used as a behavioural source for the topology graph.
eBPF
Kernel-level flow data captured by the node agent, giving the knowledge graph real service-to-service edges with no code changes.
Datadog
The full Datadog API: metrics, logs, traces, events, hosts, services and APM, queried in place and its APM used as a topology source.
New Relic
APM and infrastructure telemetry queried in NRQL, with New Relic APM feeding the dependency graph.
Dynatrace
Full-stack observability queried in Grail DQL, with Davis findings folded into correlation.
Splunk Observability
Metrics and traces queried in SignalFlow, read in place and correlated with the rest of your stack.
Observe
Data-lake observability queried in OPAL as part of the investigation toolset.
SolarWinds
Infrastructure monitoring ingested for correlation and used as an inbound alert source.
ScienceLogic
AIOps monitoring data folded into NudgeBee's own correlation and ranking.
Zabbix
Open-source infrastructure monitoring, ingested and correlated with cloud and Kubernetes signals.
Nagios
Classic host and service monitoring, brought into the same ranked incident view as everything else.
Google Cloud Trace
GCP-native distributed tracing read in place during investigations of Google Cloud workloads.
Incident & Alerting
Alerts flow in, get de-duplicated and ranked, and the root-cause analysis is written back onto the originating incident.
Alertmanager
Prometheus alerts are received, matched to the owning workload, de-duplicated and scored before an investigation starts.
Grafana
Grafana alerts are ingested as incident sources, and its data sources are read in place during triage.
PagerDuty
Two-way incident management: NudgeBee creates, acknowledges and resolves incidents, and writes the cited root-cause analysis back onto the PagerDuty incident automatically.
Zenduty
Incident response with the root-cause analysis written back onto the originating Zenduty incident, so responders see the why, not just the alert.
Ticketing & ITSM
Create, comment, transition, assign and resolve, all behind one normalized API with per-provider field mapping.
Jira
Create and update issues with live field discovery, and keep status in sync so findings track through your existing workflow.
ServiceNow
Incidents and change records created and transitioned through ServiceNow's state model, with severity and urgency mapped in code.
GitHub Issues
Findings and follow-ups filed as GitHub Issues, kept in the same normalized ticket model as your ITSM tools.
GitLab Issues
Issues created and updated in GitLab, so remediation work lands where your engineers already track it.
ChatOps
Conversational, not slash-commands. Mention Nubi in-channel to investigate, and approve every change from the same thread.
Slack
Mention Nubi to kick off an investigation in-thread, receive ranked findings, and approve or reject remediation with an interactive action.
Microsoft Teams
The same conversational investigation and approval flow, delivered natively inside Microsoft Teams.
Google Chat
Findings, investigations and approvals delivered where Google Workspace teams already work.
Discord
Alert delivery and notifications for teams that run their operations in Discord.
Email / SMTP
Notifications and approval links delivered over your own SMTP, with white-labelled templates.
Source Control & GitOps
From a log line to the exact commit and author, and back out as a ready-to-merge pull request.
GitHub
Traces a stack trace to the exact line, commit and author, and in fix mode opens a build-verified, ready-to-merge pull request.
GitLab
The same code-level root cause and merge-request generation flow, against GitLab repositories.
Bitbucket
Repository access for source-level investigation across Bitbucket-hosted code.
Argo CD
GitOps deploys are read as part of the incident window, so a recent rollout is correlated with what broke.
Confluence
Runbooks and internal docs are indexed so investigations can ground answers in your own knowledge base.
Databases & Queues
Queried through the in-cluster agent as a proxy, so credentials and data never leave your environment.
PostgreSQL
Read-only queries during an investigation, proxied through the in-cluster agent so the connection stays inside your network.
MySQL
Query MySQL as an evidence source during an incident, proxied and credential-injected in your cluster.
Microsoft SQL Server
SQL Server reads for investigation, run through the same proxy so nothing is exposed externally.
Oracle
Oracle Database queried in place as part of an investigation, proxied through the agent.
Redis
Inspect Redis state during an incident, run through the proxy with commands classified for approval.
MongoDB
Document-store queries proxied through the in-cluster agent for read-only investigation.
Apache Kafka
Topic and consumer-group state inspected during an investigation, proxied through the agent.
RabbitMQ
Queue and exchange state read during triage of messaging-related incidents.
SSH
Run allowlisted diagnostic commands on hosts over SSH, proxied through the agent and gated at the tool layer.
AI Models
Bring your own model. Nine provider routes, no lock-in, and embeddings that can run fully on-device for air-gapped deployments.
AWS Bedrock
The default managed route, giving the agents access to Bedrock's model catalog inside your AWS account.
OpenAI
Connect your own OpenAI account. Model choice is resolved per account, agent and conversation.
Azure OpenAI
OpenAI models served through your Azure tenancy for teams standardised on Azure.
Google AI (Gemini)
Gemini models via Google AI, selectable as the route for any agent.
Anthropic Claude
Claude models via the Anthropic API for teams that prefer them.
Vertex AI
Google Cloud's managed model platform, including custom Vertex endpoints.
Amazon SageMaker
Self-hosted models served from SageMaker endpoints in your own AWS account.
Hugging Face
Open-weight models served from Hugging Face, for teams running their own inference.
Ollama / self-hosted
Run open-weight models on your own GPUs, with embeddings that can run fully on-device for air-gapped setups.
Queried in native dialects including PromQL, LogQL, KQL, NRQL, Grail DQL, SignalFlow, OPAL and Elasticsearch DSL. New integrations are added regularly, and NudgeBee is extensible to any MCP server.
Integrations FAQ
Questions about connecting your stack
Everything is queried in place, in each backend's native dialect.
Request an integrationYes. NudgeBee queries the full Datadog API in place, including metrics, logs, traces, events and APM, and uses Datadog APM as a source for its topology graph. Your Datadog data is read where it lives and is not shipped to a third party.
Yes. NudgeBee writes and runs PromQL against your Prometheus or any PromQL-compatible backend such as VictoriaMetrics or Grafana Mimir, and reads Grafana data sources and alerts in place. Prometheus resolution is auto-tuned to the query window.
Ticketing and ITSM across Jira, ServiceNow, PagerDuty, Zenduty, and GitHub and GitLab issues. ChatOps in Slack, Microsoft Teams, Google Chat, Discord and email, where you mention Nubi in-channel to investigate and approve changes from the same thread.
No. NudgeBee is self-hosted and queries your observability, cloud and database backends in place, in their native query languages. Databases are reached through an in-cluster proxy, so credentials and data stay inside your own network and nothing is shipped to a third party.
NudgeBee connects to 78 systems across cloud, Kubernetes, observability, ticketing, ChatOps, source control, databases and AI models, and is extensible to any MCP server. New integrations are added regularly, and you can request one from the contact page.
No. NudgeBee reads your existing metrics, logs and traces in place across 19 or more observability platforms, in PromQL, LogQL, KQL, NRQL and other native dialects. It layers on top of what you run rather than replacing it.
Don't see yours?
Bring the tools you already run.
NudgeBee reads them in place, in their own query languages, and nothing leaves your environment. If something is missing, it is usually quick to add.