Aira
Login
Back to home

Nine AI agents on a verifiable Project Ledger.

Five user-facing specialists — planning, assignment, coding, analysis, reporting — plus four background agents that route requests, fire heartbeats, catch drift, and grade quality. They write into a shared, evidence-backed ledger that your other AI tools (Claude Code, Cursor, custom CLIs) can read too.

Powered by Claude · Orchestrated by LangGraph · Grounded in your project state

Planner Agent

Sprint planning in minutes, not all-morning grooming

90-minute grooming ceremonies? There's a better way.

Describe what you want to build. The Planner agent breaks it into tasks with acceptance criteria and effort estimates, then designs a sprint that respects your team's real capacity. Handoff to the Assigner happens automatically.

  • Creates tasks with acceptance criteria and effort estimates
  • Designs balanced sprints respecting team capacity (members × hours/week × buffer)
  • Hands the assignment decision off to the Assigner agent for skill-and-workload matching
  • Cycle detection on the dependency graph — no impossible-to-ship sprints
Aira Planner Agent
>
Analyzing codebase and planning sprint...
Set up OAuth provider config
3h
AS
Design login & register UI
5h
MK
Implement auth middleware
2h
JR
Add session management
4h
LP
Build RBAC permission layer
3h
AS
Write integration tests
2h
JR
Assigner Agent

No more "who has bandwidth?" in Slack

Gut-feeling task allocation leads to burnout and bottlenecks.

Aira scores every team member across six dimensions — skill match, workload, performance history, interest, growth opportunity, and context fit — then assigns automatically with explainable rationale.

  • 6-factor scoring (skill 30%, workload 20%, performance 15%, interest 15%, growth 10%, context 10%)
  • Workload rebalancing detects and fixes overload automatically
  • Pair programming suggestions for complex tasks
  • Growth-aware: factors in learning goals, not just capacity
Aira Assigner Agent
Refactor payment gateway integrationHIGH
Scanning team capabilities...
SK
Sarah K.
Senior Backend
Skill match
Capacity
Timezone fit
Growth opp.
AM
Alex M.
Full-Stack
Skill match
Capacity
Timezone fit
Growth opp.
JR
Jordan R.
Backend
Skill match
Capacity
Timezone fit
Growth opp.
AI Team Members

AI agents that ship code alongside your team

"We need more engineers." What if you could just add one — instantly?

Add AI agents as first-class team members. They get assigned tasks through the same Assigner scoring algorithm as humans, execute autonomously in isolated containers, create PRs, and report back — with per-action cost transparency.

  • First-class team members: same assignment scoring as humans
  • Autonomous execution in isolated containers — analyze, edit, push PRs
  • Clarification flow: agents ask before guessing on ambiguous tasks
  • Per-action cost tracking with LangChain wrapper — know what each agent costs
AI Team Members
AA
Agent Alpha
Software Engineer
Idle
codingtesting
AB
Agent Beta
Software Engineer
Idle
codingdevops
Execution log
Analyst Agent

Your best insights are buried in 200 Slack messages

Customer feedback scattered across interviews, tickets, and threads.

Feed Aira interviews, support tickets, survey results, or GitHub issues. It writes evidence-backed atoms straight into the Project Ledger, flags contradictions with full critic and reviewer rationale, and resolves conflicts — every claim traceable to its source.

  • Ingests 11 source kinds: interviews, feedback, support tickets, analytics, documents, text, markdown, CSV, JSON, PDF, and repositories
  • Writes evidence-backed atoms to the Project Ledger — every claim traceable to its source
  • Contradiction engine flags conflicting atoms with critic + reviewer rationale
  • Human-in-the-loop verification for high-stakes decisions
Aira Analyst Agent
Sources ingested
🎙️
Interview
"Users hate the onboarding flow"
🎫
Support Ticket
Ticket #4821: Export fails for large datasets
📊
Survey
62% want better reporting dashboards
🐙
GitHub Issue
#142: API rate limits too aggressive
Extracting patterns across 4 sources...
Insights extracted
Pain PointHIGH
Onboarding drop-off at step 3
Feature RequestMEDIUM
Scalable data export pipeline
OpportunityCRITICAL
Self-serve analytics dashboard
Reporter Agent

Executive-ready reports in seconds

"Stop manually assembling reports nobody reads."

Ask a question in natural language. Get a structured report with sprint progress, team health, and a delivery forecast with confidence intervals. Export to PDF in one click.

  • Natural language questions: "Will we hit the deadline?" "Who's overloaded?"
  • Delivery forecast with confidence intervals + per-status sprint progress
  • Team-health assessment surfacing blockers and overloaded members
  • PDF export for stakeholder distribution
Aira Reporter Agent
>
Behind the Scenes

Four agents you never talk to directly.

The five user-facing agents above wouldn't work without these four. They route requests, fire heartbeats, catch drift, and grade quality — quietly, in the background, so you don't have to think about them.

Supervisor
Routes your request to the right specialist

You don't have to know whether your question goes to the Planner, the Analyst, or the Reporter. The Supervisor parses intent, picks the specialist, coordinates handoffs, and stitches the final response together.

Scheduler
Wakes Aira up when it should be working

Manages heartbeat intervals for autonomous work, respects your team's timezone and quiet hours, and processes triggers from webhooks. Aira doesn't need a human to "run" — it picks up at the right moments.

Guardian
Catches drift between Aira and reality

Watches the ledger event bus and asks pluggable validators (GitHub, Jira, Slack, …) — is Aira's recorded view still matching the source of truth? Each divergence becomes a drift report you can act on. Powers the Project Ledger's accuracy guarantee.

Quality
Reviews AI-generated PRDs + features against a rubric

Procedural agent (not a LangGraph node) called by route handlers to score every AI-generated PRD or feature — incomplete / basic / intermediate / advanced — and surface iteration feedback before sloppy output ships.

Pulse Monitor

Risks flagged weeks early, not days late

"Radio silence from the team — until deadlines blow up."

Aira continuously monitors sprint health, flags blocked dependencies, detects overloaded team members, and surfaces risks weeks before they become fires. Static rules plus AI review — nothing slips through.

  • Detects high-overdue tasks, idle sprints, stale analysis automatically
  • AI-generated action suggestions with priority scoring
  • Role-based alerts: different actions for PMs vs engineers
  • Delivery forecasting with confidence intervals
Sprint Risk MonitorDay 0/10
Without Aira
1
2
3
4
5
6
7
8
9
10
With Aira
1
2
3
4
5
6
7
8
9
10
Project Ledger

The verifiable substrate your AI tools can plug into

Every AI tool maintains its own out-of-sync project context. Your work splinters across them.

Every event in your project — Slack message, GitHub PR, task update, decision — flows through one ordered log into a per-project context graph. A verifier called Guardian continuously checks Aira's view matches reality. Your AI tools read the same source of truth Aira does, via plain-text files on disk or over MCP.

  • Guardian verifier — continuously checks Aira's view matches reality; flags drift before it bites
  • Live on-disk context graph at ~/.aira/context/ (local) or GCS (cloud) — every AI tool reads the same source of truth
  • Read via MCP — Claude Code, Cursor, custom CLIs ground their answers in your project's actual state
  • One ordered log of every event — trace any claim back to its source in seconds
~/.aira/context/<your-project>/
~/.aira/context/<project>/
├── summary.md            # always up to date
├── entities/             # people, services, decisions
│   └── <name>.md
├── contradictions/
│   ├── open/             # things that don't add up yet
│   └── resolved/         # how we sorted them
├── evidence/
│   └── <YYYY-MM-DD>/     # proof behind every claim
└── timeline.md           # what happened, in order
Guardian verified
Read by:Claude Codelocal + MCP
Integration Hub

One source of truth across your entire stack

"Jira + Confluence + Slack + email + Excel... and none of them talk to each other."

Aira plugs into GitHub, Jira, Bitbucket, and Slack via OAuth. It synthesizes scattered data into a single view — webhook-driven, always in sync, zero manual updates required.

  • GitHub: PR automation, commit linking, webhook-driven sync
  • Jira: bidirectional issue sync with status mapping
  • Slack: two-way Q&A — ask Aira questions right from your channel
  • Webhook-driven with per-project DB-backed idempotency fence — safe replay on retries
Integration HubLive sync
GI
GitHub
JI
Jira
SL
Slack
BI
Bitbucket
EM
Email
Aira synthesizing
Awaiting events...
Coming Soon

Your entire Apple ecosystem. One AI PM.

Aira goes beyond the browser. Native apps for every Apple platform — designed for how you actually work, not just where you sit.

iPhone
iOS

Dynamic Island pulse. Voice commands. 30-second decisions between meetings.

iPad
iPadOS

Sketch with Pencil → structured tasks. Your meeting command center.

Mac
macOS

Menu bar presence. Global ⌘⇧A command palette. Always-on agent monitor.

Vision Pro
visionOS

3D project galaxy. Walk through your sprint. Spatial team presence.

Ready to stop updating tools and start shipping?

Join the pilot. 90-day free trial · 60% off first year · Private onboarding · Direct founder access.

Join the Pilot — Free for 90 Days