Questions worth answering
before you write to us.
Answers grouped by audience. Skim what matters. Email the rest.
How is this different from LaunchDarkly, Statsig, or Flagsmith?
Those tools target users, contexts, and segments well, with strong experimentation and approvals primitives. Argus stores tenant decisions as tenant-owned override state with a native per-tenant audit stream — not as a parking lot of segment clauses. Most teams keep their existing tool for experimentation and personalization, and use Argus for tenant-scoped operations.
Is this an entitlement system?
Adjacent. Entitlement systems answer "what is this customer contractually allowed to use." Argus answers "is this feature on for this tenant right now, who approved it, and how do I roll it back without touching the others." A clean entitlement system is a great upstream input.
Does this replace our app release tooling (Runway, Codemagic, Bitrise)?
No. Release tooling orchestrates app store submissions, certificates, and binary variants. Argus owns the persistent cross-brand truth that outlives any one release window. They compose.
Can Argus work alongside our existing feature flag system?
Yes — by design. Keep your existing flags for experiments, broad runtime targeting, and personalization. Use Argus for tenant-scoped decisions that need approvals and audit.
What's the SDK story?
Web SDK ships today — @argus/client for the browser and @argus/node for server-side pull / SSR. iOS, Android, and Web all deliver flag changes in real time via Firestore listeners (~1s, no polling). Edge resolution is on the v1.5 roadmap. The evaluation API requires a tenant — this is enforced server-side, not optional.
Can I manage flags from my AI coding agent / Claude Code?
Yes. Argus ships a local MCP server (argus-mcp) that lets an AI coding agent — Claude Code, Cursor, and the like — list, create, and publish flags, set per-environment values, and toggle flags from inside a coding session. It's governed by default: new flags are created as DRAFT and stay invisible to SDKs until published, changing a prod value requires explicit confirmation, and every mutation is written to the audit log, tagged as the agent. Today it's founder / local — it authenticates with a local service-account key and runs as a local stdio server. A productized, multi-tenant MCP with per-agent revocable tokens is on the roadmap.
What does the architecture look like today?
React frontend, Firestore as data plane and change log via onSnapshot, Firebase Auth, and Cloud Functions enforcing approvals and audit writes. We treat custom Postgres or streaming infrastructure as a post-PMF decision, not day one.
What's the latency story?
v0 latency is bounded by Firestore. We publish a real p99 number per region as we measure it across the iOS, Android, and Web SDKs. We won't quote a number we can't reproduce in front of you.
Can we self-host?
Self-host parity (control + data plane) is a stated v1 target. Today's MVP is hosted. We're explicit with design partners about this gap.
What's the security posture?
Tenant-scoped RBAC, two-engineer production approval, immutable audit rows with actor / timestamp / env / tenant / before / after. SOC 2 Type 1 evidence collection is in flight. Full security page is at /security as it matures.
What's the data model for audit?
Append-only. Tenant-scoped. Each row carries source metadata that travels in the resolution response. Exportable to your SIEM.
What does pricing look like?
Per tenant, not per MAU. Free is $0 with no card — 1 tenant, 10 flags, 100k resolves a month. Growth is $99/month for one tenant with real-time streaming and unlimited flags. Add tenants and each one is discounted further than the last, down to a $40 floor (~40% off at ten). Past ten tenants it's custom — see the full curve or contact us.
Why no calendar booking?
Async-first is a commitment, not a quirk. We've found that the best written reply within two business days beats a calendar tag and a half-prepared discovery call. If you want a call, we'll set one up after a written exchange.
Are you taking customers right now?
We're taking three design partners for a 30-day paid pilot. See /pilot.
Who is building this?
Shahin, founder. Two co-founder seats are open: a CTO / distributed systems lead, and a Head of DX / platform lead. See /about.
What if my pain is in support, not engineering?
That's the right place to start. The "is feature X on for client Y" question is the cleanest demo for what Argus replaces. Tell us about it: email a founder.