We compared Jev, TypeSafe’s model for structured decisions, with GPT, Claude, and DeepSeek using MLflow. On 30 human-labeled QA examples, Jev matched the best agreement at lower cost and latency: ✅ 30/30 agreement with human labels ⚡ 369 ms median latency 💰 $0.0247 estimated cost per 1,000 judgments Our new blog walks through building a Jev scorer in MLflow, measuring quality, cost, and latency, and deciding when to use another judge. Try the same comparison on your own dataset. 📖 Full blog: mlflow.org/blog/jev-llm-judg… #MLflow #OpenSource

A multi-agent app can return 200 OK and still be wrong. Request dashboards miss bad tool calls, empty MCP data, and stale-prompt token burn. In this video, @LegareKerrison (@RedHat) covers MLflow tracing, LLM-as-judge evals, and production setup. 🎥 Watch: piped.video/watch?v=iZX6d0Od… #MLflow #LLMOps

Use the MLflow MCP server to query tracking data from code or an assistant. 👇 🔹 Async Python FastMCP client, scoped to genai or traditional ml tools 🔹 Configure Claude, VS Code, or Cursor, then ask in plain English 🔹 ~20–30 read/write tools; runs locally or next to tracking Watch the full tutorial: piped.video/E0tFK9Ah22I #MLflow #MCP #GenAI

The trace explorer in MLflow 3.16.0 has been rebuilt from the ground up and is now the default traces experience. 🔹 Traces table: Tighter row density, a cleaner header, and smoother navigation 🔹 Span-tree explorer: Redesigned for faster drill-down 🔹 Custom columns: Reorder columns and add fields backed by any trace tag or metadata 🔹 Session grouping: Multi-turn conversations folded into the traces view alongside your other traces Learn more in MLflow 3.16.0: mlflow.org/releases/3.16.0/ #MLflow #GenAI #LLMOps #OpenSource

2 weeks until Open Lakehouse + AI Meetup in Paris 🇫🇷 Building with MLflow? Don't miss @omnigent_ai: A Meta Harness for AI Agents. Aravind Segu and Edwin He will show how to capture @opentelemetry traces for observability in MLflow across multi-harness agent workflows. 🗓️ Sept 23 | 6–9:30 PM CEST 📍 La Fondation 👉 RSVP: usergroups.databricks.com/ev… #MLflow #AIAgents #Paris

What’s New in MLflow: September 2026 Roundup nitter.net/i/broadcasts/1kKzDPqQY…

Reminder: MLflow 3.16 deep dive is tomorrow 👇 We'll cover: 🔹 Trace UI/UX Overhaul 🔹 Custom trace view via A2UI 🔹 MCP Registry 🔹 Trace analysis & automatic agent improvement flywheel 📅 Wednesday, Sept 9 🕓 4:00 PM PT 🎟️ RSVP: luma.com/jwvjuz71 #mlflow #opensource #oss #llmops

Custom Trace Views in MLflow 3.16.0 👇 Describe a layout in plain English; Assistant builds it. No config files or custom code. Save and reuse views per experiment so the team shares the same lens. 🔗 3.16.0 release notes: mlflow.org/releases/3.16.0/ #MLflow #GenAI #LLMOps

A workflow can move from Cursor to Claude Code to Pi to Codex by copy-pasting prompts and outputs. Each harness logs differently, which leaves a hole when you try to debug, audit, or trust the run.👇 This blog shows how @omnigent_ai unifies that interface and sends standardized traces to MLflow: agent turns, tool calls, per-turn tokens, and session metadata, with no app code changes. 🔗 mlflow.org/blog/omnigent-mlf… #MLflow #Omnigent

MLflow 3.16.0 is now available! 🎉 Major new features include: 🎨 Custom Trace Views: describe a layout in plain English; Assistant builds it (no config files or custom code) 🔭 Redesigned traces now default: denser rows, cleaner nav, redesigned span tree, custom columns, session grouping 🔗 Span Links: record span relationships in the SDK; Links tab jumps to the destination span Check out the release notes for more 👉 github.com/mlflow/mlflow/rel… #MLflow #GenAI #LLMOps #OpenSource

📣 3 weeks until Open Lakehouse + AI Meetup in Paris 🇫🇷 Join the open source and AI communities for technical talks, networking, and light bites. 🎉 Building with MLflow? Don't miss @omnigent_ai: A Meta Harness for AI Agents. Aravind Segu and Edwin He will show how to capture @opentelemetry traces for observability in MLflow across multi-harness agent workflows. Also on the agenda: @unitycatalog_io data architectures and @huggingface on deploying AI agents at scale. 🗓️ Sept 23 | 6–9:30 PM CEST | Paris 👉 RSVP: usergroups.databricks.com/ev… #MLflow #AIAgents

📣 Hey Paris: Join us on Sept 23 for the next Open Lakehouse + AI meetup! Aravind Segu and Edwin He will cover why to use a meta-harness: collaborate on your work, exercise control, and choose your coding agent harnesses. The talk includes a demo of Omnigent’s workflow, including @opentelemetry traces in tools like @MLflow. 🗓️ Wed, Sept 23 | 6:00–9:30 PM GMT+2 📍 La Fondation, Paris 🎟️ Register: usergroups.databricks.com/ev… #Paris #Omnigent #OpenSource

When an agent answers, traditional monitoring can still treat the workflow as a black box. You know the API returned in 5 seconds, not which tool ran or what the model saw. This @RedHat blog shows how MLflow connects that answer to the model calls, tools, and context behind it, including on Red Hat OpenShift AI. 👉 Read more: developers.redhat.com/articl… #MLflow #AIObservability

Jules Damji (@databricks) walks through what breaks when you switch coding harnesses: sharing context, carrying policies, and seeing whether the agents did the right thing. @omnigent_ai sits on top as a meta-harness; session-scoped traces go to MLflow. 📽️ Watch the tutorial: piped.video/vvJTyd-egsY #Omnigent #MLflow #OpenSource @2twitme

3 weeks out: Open Lakehouse + AI Meetup is coming to Paris! 🇫🇷 Agenda: 🔹 @unitycatalog_io (You Can Go Your Own Way....Go Your Own Way) 🔹 @omnigent_ai — A Meta Harness for AI Agents 🔹 AI Agents at Every Scale: Personal, Team, and Data Workflows Then networking, light bites & swag. 📅 Wed, Sept 23 | 6:00–9:30 PM CEST 📍 La Fondation | Paris, FR 🎟️Register: usergroups.databricks.com/ev… #OpenLakehouse #AI

Most teams iterate on agents by trying a question, fixing what looks wrong, and repeating. Each issue gets fixed once, then a later change quietly brings it back. At AGNTCon + MCPCon North America, MLflow core maintainer Yuki Watanabe (Tech Lead, @databricks) treats every quality issue as a regression test: a suite scored automatically, so regressions show up before production. The loop in MLflow: failing trace → a suite that grows with the agent. 🗓️ Thursday, Oct 22 | 12:40–1:05 PM PDT 📍 San Jose McEnery Convention Center 🔗 events.linuxfoundation.org/a… #MLflow #AGNTCon #MCPCon #AIAgents

Curious about what’s packed into MLflow 3.16? Join the MLflow Community for a live deep dive into all the latest features, including: 🔹 Trace UI/UX Overhaul 🔹 Custom trace view via A2UI 🔹 MCP Registry 🔹 Trace analysis & automatic agent improvement flywheel 🔹 Live Q&A with the team 📅 Wed, Sept 9 🕓 4:00 PM PT RSVP 👉 luma.com/jwvjuz71 #mlflow #opensource #oss

In this clip from Mastering MLflow for GenAI (Notebook 1.7), @2twitme covers MLflow’s built-in scorers (about 60+), including relevance to the query, correctness, and guidelines. Full tutorial 👉 piped.video/watch?v=WvTqW6gr… Notebook 👉 github.com/dmatrix/mlflow-ge… #MLflow #GenAI #LLMOps

Prompt optimization with DSPy and MLflow 👇 MLflow Ambassador Azad Djan builds a PubMedQA classifier in three lines of DSPy, lets MIPROv2 search the instructions and demos, and logs per-class F1 and full prediction tables in MLflow. The optimizer beat DSPy's own zero-shot prompt. A careful hand-written prompt is a closer call. And what actually improved was hedging less often, not reading abstracts better. Learn more: azaddjan.com/2026/08/20/dont… #MLflow #DSPy #LLMOps