Capabilities

Everything the platform does. One page.

From enterprise knowledge ingestion to review-ready pull requests — every capability, how the pipeline scales, and how it connects to your stack.

Pipeline tiers

The platform right-sizes to the work.

Spire classifies intent complexity and scales the pipeline automatically — from a single agent for simple tasks to the full 7-phase pipeline with parallel reviewers and verification gates for complex multi-repo features.

Four execution tiers, selected automatically
One-Shot
Simple change, codebase research, architecture review
1 agent
1 phase
Micro
Small targeted change, 1-2 files, single repo
4-6 agents
3 phases
Small
Moderate change, several files, single repo
8-12 agents
5 phases
Standard
Complex feature, multi-repo, breaking changes
15-25 agents
All 7

The Standard tier runs all 7 phases: Intake, Discovery, Review, Planning, Execution, Verification, Delivery. See how each phase works

Full capability list

What ships with Spire.

Enterprise knowledge

Upload runbooks, ADRs, design docs, policies, and wiki pages. Agents retrieve this context at execution time — reasoning over your documentation and your code. Built for legacy codebases where institutional knowledge matters.

Native Bedrock runtime

Runs natively on AWS Bedrock. No external dependencies. Model-agnostic — Claude, Llama, Mistral, or Nova at each pipeline stage. Swap models via Terraform.

Enterprise SSO

Admin-configurable SAML 2.0 and OIDC from the dashboard — Entra, Google Workspace, Okta, or any provider. Under 5 minutes. No Terraform required.

Multi-repo coordination

One feature request produces coordinated PRs across backend, frontend, mobile — however many repos are affected.

40+ specialized agents

Architecture reviewers, security auditors, implementation specialists, verification agents — purpose-built AI at each stage.

Codebase Copilot

Conversational AI integrated into every project. Searches your repos in real time, references enterprise docs via RAG, and maintains context across 25+ turns. When you're ready to build, describe what you need — the copilot proposes structured work with cost and timeline estimates, then escalates directly to the pipeline. Watch execution, approve gates, and review results in the same conversation.

Artifacts and mockups

Agents produce downloadable deliverables during execution — PDFs, documentation, spreadsheets, data exports, and interactive HTML mockups. Preview inline in the copilot or browse per-project in the Artifacts tab.

Human-in-the-loop

Approval gates at configurable checkpoints. Pause, redirect, or abort at any point. Multi-round feedback after delivery.

Budget controls

Real-time cost tracking in actual USD. Hierarchical budgets — organization, project, and per-execution caps.

Immutable audit trail

Every agent decision recorded. Independently verifiable. Compliance report generation for SOC 2, HIPAA, ISO 27001.

Slack and Jira

Type /spire in Slack. Import a Jira sprint and execute sequentially. Agents use your tools, in your environment.

Model-agnostic

Claude, Llama, Mistral, Nova — your team picks which AI models run at which stage. Controlled through Terraform.

Compounding knowledge

Every execution extracts conventions and decisions into a structured knowledge base. Combined with enterprise knowledge, agents operate with deep institutional context.

See if Spire fits your team.

30-minute conversation. We learn about your environment, you learn about the platform. No commitment.