Login Book a demo

Changelog

What’s new in Ollive

Release notes by email, when we ship. Privacy policy

Release 10

Ask Ollive in natural language

Ask about your agents from Codex, Claude, Gemini, Cursor and other AI coding assistants, or build Ollive into your own tools.

NewAPI & MCP

Ask your AI coding assistant about your agents

Add https://mcp.ollive.ai/mcp to Codex, Claude Code, Cursor or Gemini CLI (or the Claude and ChatGPT apps), approve access, and you're set. No API key needed.

Then just ask: which rules are my agents breaking? Why was this conversation flagged? Did things improve after the last release? Your assistant only sees what you can see in Ollive.

Read the docs

NewAPI & MCP

Build on the Ollive API

Wire Ollive into your own stack: connect platforms, turn on monitoring, pull findings and metrics into your dashboards, and send conversations for checking.

Admins can create API keys from the new Admin section, each scoped to a single workspace.

Read the docs

Other improvements

  • SecurityAccount & access

    API keys are shown once, expire automatically, and any admin can revoke them.

  • SecurityAPI & MCP

    The approval screen shows where an app will send you, and warns you about apps we can't verify.

Release 9

Home opens on what needs attention

Sign in and you'll see straight away which agents are breaking rules, and which rules come up most.

ImprovedHome

A morning briefing on your agents

Home charts flagged conversations over time, shows the agents and rules that come up most, and lists the agents waiting on your review. Charts go back up to 180 days, and time ranges now default to All time.

Other improvements

  • ImprovedGeneral

    We've tidied up our wording: a run is one conversation, a finding is a broken rule, a rule set is a regulation or policy turned into checks, and evidence shows why something was flagged.

Release 8

Spend less time sorting findings

Findings are the rules your agents have broken. Now you can filter them, skim them and export them to Excel.

ImprovedFindings

Filters, summaries and Excel export

Narrow the list by agent, rule set, severity or date. Each finding has a one-line summary and shows how many conversations and agents it affects, and opening one takes you straight to where things went wrong.

Need to share it? Export to Excel, one row per affected conversation, each linking back to Ollive.

Other improvements

  • ImprovedFindings

    Evidence is easier to scan, with long sections collapsed.

  • ImprovedIntegrations

    The connect screen now spells out how we handle your data: read-only access, encrypted credentials, and we never train models on it.

Release 7

Broken rules, flagged in Slack

Get a heads-up in Slack whenever a monitored agent breaks a rule.

NewMonitoring

Slack alerts

Connect Slack from Integrations, pick a channel and send a test message. Then turn on alerts for any agent.

Each alert tells you which rule was broken, how serious it is and where the rule comes from, with a link straight to the agent.

Release 6

Seven new rule sets, plus your own

Seven more regulations and frameworks to check your agents against, from HIPAA to the OWASP LLM Top 10. Need something custom? We'll build it.

Platform PacksRule sets

HIPAA, TCPA, FDA, MITRE ATLAS and OWASP

New in the library: HIPAA, TCPA for AI voice calls, three FDA rule sets (drug promotion, biologics safety reporting, and product complaints and quality), MITRE ATLAS and the OWASP Top 10 for LLM Applications. Switch on the ones that apply to you.

NewRule sets

Your policies and contracts, as rules

Missing a regulation? Ask and we'll add it. You can also send us your internal policies, or upload a customer contract, and we'll turn them into rules your agents are checked against.

Other improvements

  • ImprovedIntegrations

    Retell calls are checked as soon as they end.

  • ImprovedMonitoring

    Long conversations are now checked in full.

Release 5

Every broken rule, traced back to the law

When an agent breaks a rule, Ollive opens a finding, a case file with the evidence and the law behind it. Now they all live in one place.

NewFindings

One page for every broken rule

See every rule your agents have broken, across all of them. If the same rule is broken in 128 conversations, that's one finding, not 128 alerts.

NewFindings

Evidence behind every flag

Each finding shows the clause it comes from, links to the source, and points to the moment in the conversation where things went wrong.

Results stay tied to the rule as it was when the check ran, so your history doesn't shift when a rule is updated.

Other improvements

  • ImprovedAgents

    The Agents page now groups agents by platform and shows how many conversations each one had in the period you pick.

Release 4

Live agents, checked against the EU AI Act

Turn on monitoring and Ollive checks every live conversation against the regulations you choose, starting with the EU AI Act.

NewMonitoring

Rule-by-rule monitoring

Ollive reviews every conversation your agent has, not a sample, against the EU AI Act, one rule at a time. Anything flagged links straight to the full transcript.

Platform PacksRule sets

Set up in three steps

Connect your platform, pick the regulations and frameworks that apply to you, and choose which agents to watch. Not sure what a regulation covers? The library breaks each one down into the rules Ollive checks.

Other improvements

  • FixedAgents

    Agents built with LangGraph and other graph frameworks now show up under their own names, past runs included.

Release 3

Connect the platforms your agents already use

Ollive now connects to Braintrust, Langfuse, LangSmith, Cekura and Retell, so you can start from the traces you already have.

NewIntegrations

Five new integrations

Connect Langfuse, LangSmith or Braintrust for agents you build yourself, or Retell and Cekura for voice agents. Ollive reads each conversation step by step, tool calls included.

Release 2

Failed tests, grouped by what went wrong

See an agent's pass rate at a glance, and the problems behind its failed tests.

ImprovedAgents

From pass rate to the exact conversation

Instead of a long list of failed tests, you get a short list of problems to fix. Open a problem to see the tests behind it, then open any test to read the conversation.

Other improvements

  • NewAgents

    Using a platform we don't connect to yet? You can now add those agents by hand.

  • ImprovedGeneral

    The app has a new look and simpler navigation.

Release 1

Test your AI agents before they go live

Catch what your voice agent gets wrong before a real caller does.

NewAgents

A dress rehearsal for voice agents

Import a voice agent from Retell, tell us how it's used, and send it for testing. Failed tests show up first on the agent's page, and after a fix you can re-run them without setting anything up again.

Other improvements

  • NewHome

    Home now flags the agents that need work before launch.

Day one

Ollive came into being

Take on bigger AI deployments
without taking on unlimited AI risk.

Book a demo