SIPSymposium is a diagnostic platform for VoIP and SIP engineers who need fast, deep, and accurate analysis. It reasons the way an expert does — over your specific data, not generic patterns.
SIPSymposium is an AI-powered VoIP diagnostic platform. You provide raw inputs — SIP traces, SDP exchanges, RTP/RTCP statistics, PCAP captures, logs, or plain-language scenario descriptions — and the platform performs structured, expert-level analysis across all of them.
It understands the SIP stack end-to-end: signaling correctness, RFC compliance, media negotiation, codec symmetry, encryption handshakes, NAT traversal, and timing behavior. When multiple endpoints are involved, it correlates findings between them — surfacing asymmetries and interaction failures that single-endpoint tools miss entirely.
The diagnostic engine is Claude — a large language model by Anthropic. Claude is prompted with deep expert context covering RFC 3261 and extensions, SDP, RTP/RTCP, SRTP and DTLS-SRTP, ICE/STUN/TURN, TLS signaling, NAT traversal, and codec negotiation across all common codecs.
Every analysis request goes to Claude's API via a secure server-side proxy. Claude reasons over the full input — including PCAP-extracted data and uploaded network diagrams — and returns a structured report with severity-ranked findings, technical detail referencing specific headers and RFC sections, and actionable recommendations.
SIPSymposium is a static frontend paired with Supabase for user authentication and analysis history. API calls to Claude are proxied server-side to protect credentials. PCAP files are parsed entirely in the browser — the raw binary never leaves your device. Only the extracted text representation is sent for analysis.
Network topology and call flow diagrams are passed to Claude as base64-encoded images via the multimodal API, so the AI can visually interpret architectural context alongside your trace data.