{"id":97928,"title":"IncidentFox. The AI SRE who never drops context.","tagline":"Auto-learns your stack, builds its own integrations, and gets smarter after every incident.","body":"Hey everyone 👋\n\nWe're [Long](https://www.linkedin.com/in/long-yi-b221b2183/) and [Jimmy](https://www.linkedin.com/in/chiehmin-wei/) — co-founders of IncidentFox.\n\n**TL;DR**\n\nAI SRE tools fail without deep context about _your_ systems — and that context lives in integrations nobody has time to build. IncidentFox auto-discovers what each team needs, generates the integrations, and ships with 300+ tools built in. Setup takes less than a day, not months.\n\n👉 Try it: \u003chttps://incidentfox.ai\u003e\n\n👉 Open source (Apache 2.0): \u003chttps://github.com/incidentfox/incidentfox\u003e\n\n👉 Launch video: \u003chttps://youtu.be/TaTpN0JwNYE\u003e\n\n**❌ The Problem: Without the right integrations, your AI has no context**\n\nEvery AI SRE tool connects to Slack, reads your Confluence, queries your Datadog. That part is solved.\n\nHere's what isn't: when the AI actually needs to debug something, it doesn't have the right tools.\n\nYour payments team runs a custom Kafka pipeline with internal dashboards. Your infra team uses a homegrown deployment system. Your ML team has proprietary model serving. Each team's stack is different — and the AI has no way to query any of it.\n\nThe traditional fix? Hand-build integrations (MCP servers) for every team.\n\nBut this creates a new problem:\n\n* Who decides what integrations each team needs? You might know your own team's stack — but when you're debugging another team's service at 3 AM, you don't know theirs.\n* So every team needs to sacrifice an engineer to build and maintain their own integrations. That's expensive, slow, and doesn't scale.\n* Most teams never get around to it. The AI stays half-useful.\n\n**Integration is the bottleneck.** Not the AI model. Not the monitoring data. The integrations.\n\n**🦊 IncidentFox: We auto-build the integrations for you**\n\nIncidentFox is an AI SRE that lives where your team already works — Slack, Microsoft Teams, or Google Chat. It doesn't just connect to your existing tools — it figures out what tools each team needs and builds them automatically.\n\n![uploaded image](/media/?type=post\u0026id=97928\u0026key=user_uploads/1146198/102ebfb0-a3c9-45c5-b993-89e39b94e13b)\n\n**1. Auto-discovers and generates integrations**\n\nIncidentFox analyzes your codebase, infrastructure, and incident history to identify gaps — then auto-generates the tools to fill them. No engineer needs to hand-build MCP servers. No team needs to sacrifice headcount on integration work.\n\nA new team onboards? IncidentFox studies their stack and proposes tools specific to their services — with human approval before anything goes live.\n\n**2. Per-team configuration — because every team is different**\n\nYour payments team and your ML team don't use the same stack. Why would they use the same AI SRE config?\n\nEach team gets their own:\n\n* Tools — enable only what's relevant; disable what isn't\n* Prompts — fully open source and exposed to engineers. Inject your domain knowledge directly\n* Knowledge base — learned from that team's incidents, runbooks, and services\n\nOne team's \"config drift\" is another team's \"model drift.\" IncidentFox understands the difference.\n\n**3. Continuously evaluates and self-improves**\n\nAfter every incident, IncidentFox:\n\n* Detects gaps — \"I couldn't query service X's health endpoint\"\n* Auto-generates the missing tool\n* Evaluates its own investigation quality against the actual resolution\n* Updates prompts and knowledge — with human review\n\nIt gets measurably better every week. Not because you tuned it — because it tuned itself.\n\n**4. 300+ integrations included on day one**\n\nWhile it auto-builds what's missing, you're not starting from zero. Kubernetes, AWS, Grafana, Prometheus, Datadog, Elasticsearch, PagerDuty, GitHub — all built in. Integration time is under a day, not months.\n\n**🧠 Why this matters**\n\n**Integrations** — Traditional: Hand-built per team. Each team sacrifices an engineer. IncidentFox: Auto-discovered and auto-generated. Human approves.\n\n**Multi-team scaling** — Traditional: Breaks — you can't know every team's stack. IncidentFox: Per-team config. Each team's AI knows their stack.\n\n**Domain knowledge** — Traditional: Black box prompts, hope it works. IncidentFox: Open source prompts. Engineers inject and edit freely.\n\n**Over time** — Traditional: Stagnates unless manually updated. IncidentFox: Self-evaluates, finds gaps, improves continuously.\n\n**Setup** — Traditional: Months of custom integration work. IncidentFox: \u0026lt; 1 day. 300+ tools out of the box.\n\n**📊 Results**\n\n* 85–95% reduction in alert noise through intelligent correlation\n* Hours → minutes for incident investigation\n* Zero-config onboarding — Docker in 5 min, production K8s in 30 min\n\n![uploaded image](/media/?type=post\u0026id=97928\u0026key=user_uploads/1146198/844d70ec-e308-472c-9bc4-f212cd954153)\n\n**🔒 Enterprise-ready, open-source**\n\n* Open source — Apache 2.0, no vendor lock-in\n* SOC 2 compliant, SSO/OIDC, RBAC, audit logs\n* Self-hosted, on-prem, or managed SaaS\n* Bring your own LLM keys (OpenAI, Claude, Gemini, etc.)\n\n**👬 The Team**\n\n[**Jimmy (CEO)**](https://www.linkedin.com/in/chiehmin-wei/) — Previously at Roblox, where he built social communication features (in-experience calling for 100M+ DAU). Before that, worked at Meta FAIR on multiparty conversational AI, with published research. Cornell CS. Serial founder — previously CTO at a startup in Outlier Ventures' DeFi accelerator.\n\n[**Long (CTO)** ](https://www.linkedin.com/in/long-yi-b221b2183/)— Previously at Roblox, where he built database infrastructure supporting 100M+ daily active users on the Stateful Infra team. Experienced the chaos of on-call firsthand — which is why we're building this. Brandeis CS + Neuroscience + Business.\n\nWe've lived on both sides: Jimmy built the AI systems, Long was the SRE drowning in incidents. IncidentFox is what happens when you combine both.\n\n![uploaded image](/media/?type=post\u0026id=97928\u0026key=user_uploads/1146198/261f0711-50aa-4a0f-b6c9-9cdc7c853dbc)\n\n\n**🙏 Our Asks**\n\n* Try IncidentFox — self-serve at [incidentfox.ai](https://incidentfox.ai) or star us on [GitHub](https://github.com/incidentfox/incidentfox)\n* Intros to engineering teams spending too much time on on-call\n  * [Book a demo with us](https://calendly.com/d/cxd2-4hb-qgp/30-minute-demo-call-w-incidentfox)\n* Feedback from SREs and infra engineers — [founders@incidentfox.ai](mailto:founders@incidentfox.ai)\n\nThanks for reading ❤️","slug":"PTU-incidentfox-the-ai-sre-who-never-drops-context","created_at":"2026-02-18T01:06:25.804Z","updated_at":"2026-05-25T01:05:38.526Z","total_vote_count":5,"url":"https://www.ycombinator.com/launches/PTU-incidentfox-the-ai-sre-who-never-drops-context","share_image_url":"https://www.ycombinator.com/media/?type=post\u0026id=97928\u0026key=user_uploads/1146198/844d70ec-e308-472c-9bc4-f212cd954153","company":{"id":31165,"name":"IncidentFox","slug":"brownie","url":"https://incidentfox.ai","logo":"https://bookface-images.s3.amazonaws.com/small_logos/43fb964e8e0ca12dcc1000b4c4c2d6be7ffc203e.png","batch":"Winter 2026","industry":"B2B","tags":["AIOps","Artificial Intelligence","Developer Tools"],"search_path":"https://bookface.ycombinator.com/company/31165"}}