With Q3 in full swing, we've shipped product updates to speed up your development: Side‑by‑side Playground testing, real‑time alerts (Webhooks & Slack) for prompt changes, usage alerts on Langfuse Cloud, and one‑click remote experiments for non‑technical teammates.

Q3 Roadmap Highlights & Townhall

Earlier this month, we held our Q3 planning and a community Townhall. Here are some of the highlights we will be working on in Q3:

• Agent observability: unified agent graphs, richer tracing (tool‑call span types), built‑in agent evaluations across conversation/response/turn/trajectory.
• Evaluation: rule‑based evaluators; LLM‑as‑judge at multiple levels with debugging & cost tracking; comparison dashboards (correlation, confusion matrix, overlap); SDK abstraction for dataset runs.
• Data, prompts & platform: multimodal support for Datasets & Playground; I/O schema validation; tool/function support in Prompt Experiments; platform webhooks; in‑app routing.

→ View the public roadmap → Watch the Town Hall Recording

Playground Side-by-Side Comparison

Open multiple chat windows in parallel to compare prompts, models, variable inputs, or tools, and restore the test state.

This enables faster A/B testing, fewer open tabs, and shorter feedback cycles.

→ Learn More

Slack & Webhooks for Prompt Changes

We added Webhooks and a Slack integration to get real-time notifications for new prompt versions, changes to deployment labels or tags, and other edits.

We’re expanding webhook triggers to cover more Langfuse events soon.

→ Learn More

Trigger Remote Custom Experiments

You can now trigger external Dataset Runs directly from the Langfuse UI. This feature enables non-technical users to run tests on Datasets without requiring code changes or manual setup.

→ Trigger a remote Dataset Run

New docs and integration pages

We reorganized our docs and integration guides to make discovery easier across observability, prompts, and evals.

→ Explore our new documentation and integration pages

More July Releases

• Usage Alerts (Cloud): Set a threshold for monthly events; we'll notify you once per billing cycle when you cross it.
• Sessions in the Annotation Queue: Integrate Sessions into your human annotation workflow.
• n8n Node: Fetch Prompts from Langfuse Prompt Management into your n8n workflow.
• Prompt Full-Text Search: Search through all your prompts to help find and organize large prompt collections.
• LiveKit Integration: Trace your LiveKit agents to see latencies, costs and use it together with other Langfuse features.

---