Molly + AI agents
AI that ships work, not vibes
Molly sits inside the workspace with real capabilities: she queries your data, renders live charts, creates and updates records, drafts replies, runs code, and speaks. AI agents extend that to autonomous teammates with their own knowledge and tools.
Ask
Questions in, artifacts out
Ask 'which brand generated the most revenue this quarter?' and Molly runs a real query against your data and answers with a live chart — switchable between visualization types, downloadable, and openable in the app. Tables, files and images are first-class artifact types too.
- Schema-aware: Molly knows your objects and fields
- Live chart artifacts with 57 renderers
- Multiple artifacts per answer
- Never-silent fallback — you always get a response
Act
The same hands as your team
Molly performs actions through the platform's action registry — the exact same operations the UI and API use, gated by the same permissions. Create records, log tasks, draft campaigns, update pipeline stages. Every write shows a confirmation card first; nothing mutates behind your back.
- 165 actions across 19 feature areas
- Respects your role and field-level permissions
- Confirm-each-write with defer/resume
- Persistent conversations with history
Models
Bring the model you trust
klerc is model-agnostic: run Molly on OpenAI, Anthropic Claude, Google Gemini, or anything on OpenRouter — picked per organization, switchable anytime. Voice notes are transcribed with Whisper; replies can be spoken aloud with natural TTS. Code execution runs in a sandbox and returns images.
- OpenAI, Anthropic, Google, OpenRouter
- Per-org model picker
- Voice in (Whisper) and voice out (TTS)
- Sandboxed code execution with chart output
Model menu
AI agents
Teammates you configure, not hire
Each agent gets a name, instructions, a model, its own vector-store knowledge base (upload files or inherit from your org KB), and a tool belt you select per agent. Preview-chat to test before activating. Email and phone channels for agents — inbound mail, SMS and voice IVR — are on the roadmap.
- Per-agent knowledge with selective scope
- Tool selection per agent
- Draft / active / paused lifecycle
- Preview chat for safe testing
AI across the platform
Inbox copilot
Reply drafts from conversation context; opt-in auto-reply per account.
AI captions
Caption generation in the publishing composer.
Report suggestions
Generated, validated SAQL report ideas you can accept in a click.
KB summaries
Automatic tl;dr and tags on every ingested document.
NL queries
Plain-English translation to SOQL / SOSL / SAQL in the search bar.
Chart recommendations
The right visualization suggested from data shape.
Works better together
Related modules
Run your whole business from one tab.
Join the early access — bring your team, connect your channels, and keep every workflow in one place.