In production at scale

AI engineering operations for Kubernetes platform teams.

Horizon unifies Kubernetes observability, Jira delivery intelligence, and AI-generated analysis of sprints, releases, and incidents. Built at D2 Solutions, running in production at scale across 7 Kubernetes clusters and 10+ tenants.

horizon · platform / d2-prod
Clusters Releases Sprints Incidents Live
Health
94
7 clusters · +3 last 24h
Deploy
88
v2.4.1 ready · 12m ago
Sprint
42SP
Dev 68% · QA 31%
Lead time
2.4d
down from 3.1d
Tenants
p95 latency
  • d2-prod
    94 42ms
  • d2-staging
    87 58ms
  • d2-qa
    71 124ms
  • d2-dev
    88 49ms

Live data from 7 production clusters, 10+ tenants, daily on-call use.

What Horizon does

One platform across the three layers most engineering organisations stitch together by hand.

01

Kubernetes Observability

Multi-cluster, multi-tenant runtime visibility that thinks in workloads, not just pods.

  • Per-namespace health score combining pod state, restarts, events, and deployments
  • Full resource browsing: ConfigMaps, Secrets, Ingresses, PVCs, RBAC, NetworkPolicies
  • Cluster, tenant, and environment comparison views
  • TV-mode dashboard for ops floors and standups
02

Jira Delivery Intelligence

Sprint, release, and DORA signal that takes carryover seriously.

  • Sprint dashboards with Story Points, Dev / QA pipeline, team breakdowns
  • Deploy confidence score (0-100) per release, weighted across services-ready, prerequisites, regression, and Jira completion
  • DORA metrics pulled from Helm history and incident records
  • Carryover detection through the Agile API, not the customfield that quietly returns NULL
03

AI Sprint, Release & Incident Analysis

Anthropic Claude and OpenAI wired in with grounded, board-aware context.

  • Per-sprint analysis: health, workload balance, dev/QA pipeline, risks, velocity comparison
  • Per-epic analysis with workload distribution, bottlenecks, stale tickets, release impact
  • Historical analysis across all 2026 sprints, velocity trends, recurring issues, recommendations
  • Incident timeline correlation across deploys, Kubernetes events, and MSSQL log streams

Why we built Horizon

Every engineering organisation we work with has the same two-layer problem. The delivery layer (Jira, sprints, releases) and the runtime layer (Kubernetes, observability, incidents) belong to the same team and describe the same product, but they almost never speak to each other in the same place.

The engineering intelligence vendors (LinearB, Faros, Swarmia) do Jira and git well but do not touch the cluster. The runtime tools (Komodor, Cortex, Port) see the cluster but do not understand delivery. Every team we know was running both, plus a custom dashboard, plus the bookmark to the Helm release history.

Horizon connects sprint reality to live cluster state, and lets an AI explain what you are looking at.

Built like a real SaaS

Horizon is not an internal tool that grew accidentally. The multi-tenant, billing, and auth surfaces are built in, not bolted on.

Multi-tenant from day one

Every model in Horizon carries an org_id and goes through scoped queries. Organizations, users, clusters, plugins, audit logs, AI cache, sprint boards, and releases are all org-isolated by design, not retrofitted.

Tier system already in place

Free, Starter, Pro, and Enterprise tiers with per-tier rate limits, resource quotas, audit retention, and plugin limits. The billing surface is ready to wire to Stripe; the policy is already enforced in code.

Enterprise auth out of the box

OIDC SSO, LDAP fallback with JIT user provisioning, TOTP MFA with backup codes, plus a super-admin impersonation flow via X-Org-Context header. Encrypted Kubernetes cluster tokens at rest.

Plugin architecture

Health score, deploy confidence, release tracker, cert calendar, MSSQL log analytics, sprint dashboard, regression management. Each is a plugin that can be enabled per organization. Building the next plugin is a contained change.

Stack

Boring on the serving layer, interesting on the analysis layer. Iteration speed where it matters, durability where it does not.

FastAPI Python HTMX Tailwind Kubernetes Anthropic Claude OpenAI Jira Agile API Prometheus MSSQL OIDC LDAP MFA JWT Helm Docker SQLite PostgreSQL
Integrations

Plugs into your stack.

Horizon reads from the systems your team already runs and pushes signal to the channels they already watch. No new dashboards to babysit.

Delivery sources

  • Jira (Agile API)
  • Bitbucket
  • GitHub
  • Jenkins

Runtime sources

  • Kubernetes API
  • Helm release history
  • Prometheus
  • MSSQL log streams
  • Harbor registry

AI providers

  • Anthropic Claude
  • OpenAI

Bring your own API key or use ours.

Identity & notify

  • OIDC (Keycloak, Auth0, Azure AD)
  • LDAP / Active Directory
  • TOTP MFA
  • Slack webhooks
  • Email (SMTP)

More on the roadmap: PagerDuty, Opsgenie, GitLab, ArgoCD, Datadog.

Early access open

Become a Horizon design partner.

We are opening early access to a small number of design partners this quarter. If you run a Kubernetes platform with more than a handful of tenants and your team lives in Jira, Horizon is built for you.

Get in touch