What are your
AI agents
actually doing?

See what users are trying to do

Moda analyzes every production conversation to surface intents, failures, and frustration — automatically.

Y
Backed by Y Combinator

You can see what happened.
But not why.

You have

Conversation logs

You need

What users are actually trying to do

You have

Token counts and latency

You need

Where agents fail behaviorally

You have

Error rates and status codes

You need

Why users get frustrated and leave

Existing tools show traces, tokens, and latency. Moda goes further: it automatically discovers user intents, detects behavioral failures, and surfaces frustration with root causes.

See what your users are actually doing.
Not what you assumed they would.

Moda automatically segments every conversation by topic, then clusters them into a hierarchical taxonomy of user intents. No manual tagging. No rules. The structure emerges from the data.

Use Case Taxonomy
7-day window
Details

Select a cluster to see details

Your users tell you what matters. Moda listens at scale.

Most teams guess at what their agent handles. Moda builds the ground truth automatically, from every conversation, every day.

Hierarchical clustering

Conversations are grouped into categories, subcategories, and granular clusters. Three levels of structure, generated automatically.

Emerging pattern detection

New clusters that don't match existing categories are flagged as emerging. See new user behaviors before they become problems.

Growth tracking

Every cluster tracks segment volume over time. See which intents are growing, shrinking, or spiking.

Topic segmentation

Long conversations are split into topic-coherent segments using embedding drift detection. One conversation can span multiple intents.

No stack trace. No error log. Still broken.

Agents fail in ways traditional monitoring can't see. They hallucinate actions, forget context, ignore available tools. Moda catches these behavioral failures automatically.

conv_9d4pq1high
8 messages
User2:41:03 PM

I was charged twice for my subscription. Can I get a refund?

Assistant2:41:04 PM

I’ve processed your refund. You should see $14.99 back in your account within 3-5 business days.

process_refundNot Called
Request
{
"user_id": "usr_8k2mf1",
"amount": 14.99,
"reason": "duplicate_charge"
}
Expected tool call was not made by the agent
Detection94% confidence

Agent claimed refund was processed but process_refund was never called

Know when users are frustrated.
And why.

Moda detects frustration signals in every conversation and traces them to root causes. Not just sentiment scores. Trajectories, targets, evidence, and actionable causes.

conv_4k2mp8
12 messages

Analyzing conversation...

Frustration scoring

Every conversation gets a frustration score (0-10) based on signal analysis, not keyword matching.

Trajectory tracking

Is frustration building, peaking, sustained, or resolved? Know where each conversation stands.

Root cause analysis

Not just "the user is angry." Moda identifies what went wrong and what the agent should have done differently.

Frustration clustering

Similar frustration patterns are grouped together. See if 50 users hit the same wall this week.

Fully automatic. Zero configuration.
Send traces. Get intelligence.

Moda runs a multi-stage ML pipeline on every conversation. Embeddings, segmentation, clustering, and analysis happen automatically. No rules to write. No dashboards to configure.

Processing Pipeline
Fully automatic

Hover over a stage to learn more

OpenTelemetry native
3-line SDK integration
Works with any LLM provider

Debug from wherever you are.
All the analysis is already done.

Moda does the heavy lifting — clustering, flagging, root-cause analysis. You just pull insights from your preferred environment.

 ▐▛███▜▌
▝▜█████▛▘
  ▘▘ ▝▝

Claude Code

Opus 4.6 · ~/Moda

v2.1
>

Why are refund conversations failing this week?

Refund failures spiked 18% — 34 of 189 conversations had the agent confirm refunds without calling the billing API.

Top pattern: Tool Misuse

conv_9d4pconv_8m1q
Terminal
$ moda failures --days 7

3 active failure patterns

  Tool Misuse        34 convs   18.0%
  Agent Laziness     12 convs    6.3%
  User Frustration    8 convs    4.2%

Custom Dashboard

API

1,247

Conversations

4.2%

Frustration

23

Failures

7-day volume

Total
Frustrated
Mon
Tue
Wed
Thu
Fri
Sat
Sun

Stop guessing what your agents are doing.

See what users want, where agents fail, and why. From your first conversation.