Portkey is a strong AI gateway — routing, fallbacks, caching. AgentLens focuses on what happens AFTER the call: deep traces, quality scoring, agent debugging, and compliance. The two coexist nicely.
| Capability | AgentLens | Portkey |
|---|---|---|
| LLM gateway / routing / fallbacks | ✗ | ✓ |
| Semantic caching at edge | ✗ | ✓ |
| Self-hosted (free path) | ✓ open-core, BSL 1.1 | Enterprise contract |
| Multi-step agent waterfalls | ✓ trace_agent + spans | Limited |
| Automatic quality + hallucination scoring | ✓ heuristic, zero config | Manual eval |
| Air-gap mode (single env var) | ✓ tcpdump-verified | Limited |
| EU-hosted managed cloud | ✓ Frankfurt | Multi-region |
| Agent debugger UI | ✓ waterfall view | Limited |
| Compliance workflows (DPA, audit, retention) | ✓ Scale plan | Pro+ |
Based on public Portkey pricing as of May 2026. Verify at https://portkey.ai/pricing.
Both tools share a similar mental model. The SDK surface is different but migration is mechanical.
Mostly complements. Portkey is a gateway; AgentLens is observability. Many teams run both.
Yes. They operate at different layers — Portkey at the request gateway, AgentLens at the application SDK level. Both see the same call.
Heuristic quality scoring out of the box + agent waterfalls + EU compliance + verifiable air-gap. Portkey's tracing is functional but lighter-weight.
15-minute migration call. We'll map your Portkey setup to AgentLens and get you running.