Tell me what just broke.
In plain English. In seconds.
StackSift watches your logs, fires alerts when something cascades, and explains the root cause with AI — built for solo devs and small teams, priced for them too.
[ERROR] 2026-03-27 02:14:33 NullReferenceException at UserService.cs:142 [WARN] 2026-03-27 02:14:33 Redis connection timeout after 30000ms [ERROR] 2026-03-27 02:14:34 Database query exceeded timeout (pool exhausted) [ERROR] 2026-03-27 02:14:34 NullReferenceException at UserService.cs:142 [ERROR] 2026-03-27 02:14:34 NullReferenceException at UserService.cs:142 ... 395 more lines ...
45 minutes to 3 hours. Average time to root cause. At 2 AM. On a Tuesday.
How it works
- 01Connect
Add the Serilog sink (.NET) or Winston transport (Node.js) to your app in ~5 lines. Or POST JSON directly with curl — full API reference in the docs.
- 02Stream
Logs flow in, your dashboard updates live, and alerts fire on the patterns you care about.
- 03Resolve
Click Analyze with AI on any incident. ~8 seconds later you have a root cause and a fix.
90 seconds. The whole pitch.
A real local stack. Real OpenAI call. No edits, no cuts.
›Transcript
Four things, done well.
See errors the moment they happen — no F5, no SSH-tail.
SignalR + Redis backplane
Threshold and pattern rules. No PromQL.
LogBatchConsumer rule evaluation
gpt-4o-mini turns 400 log lines into one paragraph.
RAG pipeline · ai-rag-architecture.md
Every resolved incident teaches the AI about your system.
pgvector HNSW cosine top-3
Why not just use Datadog?
Because you’re not their customer. We’re not theirs either.
Pricing that fits in your Stripe statement.
No credit card to start. The SDKs are open source. Cancel any time.
- Unlimited projects
- 90-day retention
- Unlimited AI analyses
- Unlimited users + roles
We’re a 3-person team. Your money goes into making this better, not into a sales department.
Trust, by design.
Application data, databases, and log indexes run on Hetzner in Germany. Log data is isolated per organisation.
Send logs through open-source SDKs or plain HTTP. You do not install a privileged daemon or give StackSift shell access.
Authentication is handled through OIDC + PKCE, and browser sessions use HTTP-only cookies rather than JavaScript-readable tokens.
Questions, before you start.
How is this different from Datadog?+
Cheaper, simpler, and the AI summary is built in — not an add-on.
StackSift is designed for one engineer or a small team. Datadog is designed around larger SRE organizations.
How do I send logs to StackSift?+
Use the official Serilog sink for .NET, the Winston transport for Node.js, or direct HTTP POST from any language.
All three options hit the same /api/v1/logs/ingest endpoint and keep setup small.
What happens if I exceed the AI-analysis cap?+
The Analyze button returns a graceful upgrade nudge. Log ingest, search, and alerts keep working.
You can upgrade in-app or wait until the month resets.
Do I need to change my application code?+
Yes, lightly. You add the StackSift sink or transport to your existing logger.
We do not run an agent on your servers and we do not tail files from disk.
How is my data isolated from other customers?+
Every entity carries an OrganizationId enforced at the repository level.
Elasticsearch indices, S3 keys, and SignalR groups are scoped per organization or project.
Can I delete my data?+
Yes. Deleting your account starts a 30-day grace period with a one-time restore token.
After that, we hard-delete organisation rows, Elasticsearch indices, and S3 objects, except records with legal retention floors.