AI Platform
Drag Agent
Early AccessDrag Agent reads every email, classifies by intent, pulls context from your tools, takes real action, and resolves, or escalates when it should.
Email in. Resolution out.
Four steps, under 30 seconds. The agent classifies intent, retrieves context from your tools, takes real action, and resolves, or escalates to your team.
23s
Average resolution
94%
Confidence threshold
0
Unsourced answers
09:41:02 CLASSIFY intent=billing, entity=invoice #4821
09:41:03 RETRIEVE stripe.invoices.get(4821) → $249.00 paid
09:41:04 RETRIEVE kb.search('refund policy') → 30-day window
09:41:05 ACT stripe.refunds.create(amount=249.00) → rf_8xK2
09:41:06 DRAFT confidence=0.96, sources=2, tokens=142
09:41:07 SEND thread_id=t_9f3a → resolved
Four steps to resolution.
Here is exactly what the agent does with every inbound email.
Classify
Reads the email, identifies intent (billing, bug, how-to, sales) and extracts key entities.
Retrieve
Pulls context from your knowledge base, CRM, ticketing system, and previous conversations.
Act
Takes real action: issues a refund in Stripe, updates a ticket in Zendesk, logs a note in HubSpot.
Resolve
Drafts a response, cites its sources, and sends, or escalates to your team if confidence is low.
Connected to every tool your team uses.
The agent connects to the tools you already use, reading context, taking action, and listening for messages across every channel.
Reads from
Knowledge bases, docs, and data your agent uses to answer accurately.
Notion
Google WorkspaceConfluenceHelp centerPDFsURLsTakes action in
CRMs, billing, and ticketing tools your agent writes to.
Stripe
HubSpot
Zendesk
IntercomJiraSalesforceListens on
Channels your agent monitors for incoming messages.
Gmail
Slack
WhatsAppLinearPowered by Intelligence — your connected knowledge layer.
Knows when to act. Knows when to ask.
Every resolution carries a confidence score. The agent only sends autonomously when it is sure. Otherwise, it escalates to your team with full context.
Agent resolves autonomously
When the agent has high confidence in its answer and the actions it took, it sends the response directly. No human in the loop.
94% confidence
Escalates to your team
When confidence is below threshold, the agent surfaces the email to your team with a draft, reasoning, and all retrieved context.
62% confidence
You can see what it's doing. Always.
The agent exposes its state in real time. Every thread shows exactly where the agent is in its pipeline.
Idle
Waiting for a new email to arrive.
Thinking
Classifying intent and planning next steps.
Working
Retrieving context and taking action.
Done
Resolved and response sent.
Escalating
Confidence too low, handing off to your team.
Every action logged. Every source cited.
Trust is non-negotiable. The agent keeps a full audit trail, cites every source, and meets enterprise security standards.
Full audit trail
Every action, every API call, every decision, timestamped and searchable.
Source citations
Every response links back to the knowledge base article, CRM record, or doc it drew from.
Enterprise security
SOC 2 and CASA certified. Data encrypted at rest and in transit. No training on your data.
Your shared inbox, accessible from any AI tool.
Connect your inbox to the AI tools your team already uses. Read, reply, assign, and resolve through MCP, API, or Drag Agent.
Drag Agent
Classifies, drafts, and resolves tickets end-to-end.
MCP Server
25 tools for read + write access from any AI client.
REST API
Boards, threads, contacts, automations. Webhooks included.
AI Drafts
Reads the thread, drafts a reply in your team's tone.
Auto-classification
Tagged by intent, sentiment, and urgency automatically.
Knowledge base
Answers grounded in your docs, emails, and tools.
See AI in your inbox.
Drafts, classification, summaries, co-pilot, all included. No extra cost, no per-agent pricing.
200,000+ users · 4.7 on G2 · CASA & SOC 2 certified
