Product Roadmap
TokenGuard is moving fast. 13 of 20 planned features are already live. This page is updated as we ship.
🏆 Our Unique Competitive Edge
Two capabilities that no other observability platform in this space currently offers:
- ✓Cost Waste Engine with AI recommendations — TokenGuard doesn't just show you spend. It proactively identifies what's wasting money and tells you how to fix it.
- ✓Spend enforcement (block before overspend) — Set hard limits. TokenGuard throttles ingest when you're about to go over budget. Competitors only alert after the fact.
Core Observability
Record every LLM call with full token counts, latency, and estimated cost.
Python and Node.js SDKs that wrap any OpenAI-compatible client with one line of code.
Step-by-step execution timeline with prompt/response inspector.
Group multiple traces into a single session to track multi-turn conversations end-to-end.
Send spans from any OTel-compatible SDK directly to TokenGuard.
Cost Intelligence
AI-powered waste detection that identifies oversized prompts, retry loops, and model mismatches — and tells you how to fix them.
Set monthly trace or cost budgets. TokenGuard blocks ingest when limits are hit — not just alerts after the fact.
Daily estimated spend chart on the dashboard overview.
Cache identical prompts to eliminate redundant LLM spend (cuts 20–40% for high-repetition workloads).
Alerts
Trigger email notifications when daily cost, error rate, or per-request cost crosses a threshold.
POST to any URL when an alert fires — works with Discord, Zapier, and custom APIs.
Rich Slack notifications with metric details, threshold, and a direct link to the dashboard.
Data & Evaluation
Export production traces as JSONL for fine-tuning or as CSV for analysis.
Annotate and curate production traces into labelled datasets for regression testing.
Auto-grade agent outputs using LLM-as-judge or custom scoring functions.
Track prompt templates across versions and roll back to any previous version.
Team & Access Control
Pro includes 3 seats with +$10/seat add-ons. Enterprise has unlimited seats.
Assign fine-grained roles to team members per project.
Single sign-on for enterprise teams.
Infrastructure
Deploy TokenGuard on your own infrastructure for data-residency-sensitive workloads.
Have a feature request?
Tell us what you need. We prioritize based on real user demand.
Request a feature →