Statefulai/Use cases

The workflows persistent memory actually unlocks.

Memory is a foundation, not a feature. Below: the five scenarios where AI-first teams feel the difference inside their first week with Statefulai.

01OnboardingDay 1 productive 02Cross-agent handoffOne shared thread 03Long migrationsSix-week refactor 04Team knowledgeDecisions persist 05Enterprise contextAudit · RBAC · self-host
Who it's for

Three ICPs, one shared pain.

Different teams, same friction. Pick the row that sounds like yours and follow the recommended use case.

AI-first engineering teams

Daily Claude Code / Codex / Cursor users. Already comfortable with agents writing code. Wrestling with context limits, agent drift, and the cost of re-explaining the codebase.

› drift · token spend · onboarding loop

AI agencies & automation teams

Building automation on top of agents. Need durable memory across client workspaces and across bot runs. Strong opinions about cost-per-task.

› multi-tenant · cost · reliability

OSS & autonomous-agent builders

Building agents that run in CI, in nightly schedules, in production. Need memory that's portable, debuggable, and self-hostable.

› self-host · MCP · interop
01

Onboarding

An agent with full architectural memory is a better onboarding buddy than a 90-minute walkthrough. New hires ask their IDE; Statefulai answers with the team's actual conventions, decisions, and code patterns — not what your wiki said two years ago.

Day-1 productive−6 weeks ramp−12 hr/wk mentor
Read full case →
› where do we keep auth middleware? ⎯ src/server/auth/* uses requireSession() · refs 14 ⎯ ADR-007: cookies not JWT decided 03-21 · author @rena
02

Cross-agent handoff

Claude drafts the implementation. Codex reviews it against the architectural graph. Cursor finalizes the diff in the editor. Each picks up the thread where the previous one left it — no re-explaining the spec, the constraints, or yesterday's 4pm decision.

1 thread · 3 agents−92% tokens / handoffconsistent diffs
Read full case →
claude codex cursor › shared memory · one thread
03

Long-running migrations

Migrations span dozens of branches and sessions over six weeks. Statefulai keeps the migration plan, the constraints discovered along the way, the exceptions to the rule — all intact through every context window reset. Weeks-old gotchas reappear when they're needed.

6-week refactorpinned constraintsbranch-aware
Read full case →
migration · postgres-13 → 16 ⎯ pinned: jsonb path → seq scan ⎯ pinned: pgvector ≥ 0.6 required
04

Team knowledge

"Why did we do it this way?" is the most expensive question in software. Statefulai captures decisions from Slack and ADRs, links them to the code that resulted, and surfaces them automatically — six weeks later, six months later, for whoever asks.

ADR auto-capturedecision lineageno more stale wikis
Read full case →
@rena · #eng-billing · 04-12 › let's drop the queue here ⎯ ADR captured · billing.flush() rationale + alternatives stored @new-hire · 6 weeks later › agent: "no queue — see ADR-014"
05

Enterprise context

The same memory layer, hosted inside your VPC. Per-project isolation, RBAC, SSO, audit trails on every retrieval. Memory visibility controls let you mark sensitive code as "private to a sub-team." Self-hosted or dedicated cloud — your call.

VPC / self-hostRBAC + SSOaudit log
Read full case →
your VPC · region eu-west-1 STATEFULAI control plane data plane audit SSO · SAML RBAC · audit
Find your wedge

Start with the workflow that's bleeding today.

Pick a use case above. Each one is a 60-second install away.

Get early access Read the docs