Simon

Simon is Antartida's platform to test, validate and monitor AI agents directly on Salesforce.

Explore Simon

WHAT IS SIMON

The testing layer for AI agents

Simon helps teams test conversational agents before they reach production. It lets you create realistic test scenarios, run them at scale, evaluate responses automatically, and understand exactly what happened when something fails. From testing individual conversations to validating hundreds of variations, Simon gives teams a repeatable way to know whether an agent is ready to ship.

Same deadline.Same team.Ten times the coverage.

Simon automates the execution, evaluation and documentation of QA — so teams can test more without extending the timeline.

Swipe

QA cycle
Regression
Failure diagnosis
Validation
Findings
Who finds the bugs
QA without Simon
Weeks
Redone by hand, from scratch
Manual, reconstructing in Salesforce
“We think it works”
✕Scattered across chats and loose docs
✕The client
QA with Simon
Days
One click, full suite
Automatic: response, flow or subagent
Evidence, traceability and metrics
✓Documented and centralized
✓Simon

See it working

Watch Simon run a suite

The impact

More coverage. Less manual work.

Test more scenarios, repeat validations and diagnose failures without scaling QA effort at the same rate.

Implementation

Ship validated agents, with evidence of what was tested.

Quality

Repeat the same tests across versions, languages, tones and edge cases.

Support

Diagnose failures faster with full context and proactive alerts.

Clients

Measure confidence before production and reduce unexpected failures.

Velocity

70% less testing time and 3.3× faster testing operations.

How Simon solves it

Everything you need to validate an AI agent.

From the first test case to the final verdict, Simon gives your team a repeatable way to simulate, evaluate, diagnose and monitor agent behavior.

Swipe

Smart simulations
01

Smart simulations

Create granular test cases and run them repeatedly to measure consistency.

Automatic judge
02

Automatic judge

Evaluate every response against defined rules and get a clear verdict — passed or failed, and why.

Context-aware paraphrasing
03

Context-aware paraphrasing

Test the same scenario in formal language, casual language, regional slang, with typos and different ways of expressing the same intent.

Real-time metrics
04

Real-time metrics

Track success rate, failed validations and compare performance across agent versions.

Notion and Slack integration
05

Notion and Slack integration

Monitor agent behavior after deployment and receive alerts when something breaks.

How it works

The cycle, in eleven steps.

A continuous improvement loop: test → diagnose → fix → validate → deploy.

Swipe

Connect

Steps 1–2
  • Connect your Agentforce agent to Simon.
  • Choose the judge: Salesforce + Flex Credits or an external LLM.

Design

Steps 3–4
  • Define use cases and simulations.
  • Set up validations and success criteria.

Run

Steps 5–6
  • Run the suites automatically.
  • Simon judges every response.

Diagnose & fix

Steps 7–9
  • Review results, traceability and metrics.
  • Create tickets for the errors.
  • Adjust the agent in Salesforce.
Step 10

Re-run everything and confirm the fix.

Step 11 · Result

Deploy with confidence.

Validated agents, with evidence of every test.

Less time. More coverage. More control.

Simon turns conversational agent QA into a fast, scalable, measurable and traceable process, so you can reach production with more confidence.

Meet Simon