AI
Sift AI Context
The Sift AI Context page is where admins teach Sift's AI about the organization it's working for. Everything Sift does — from synthesizing records to ranking relevance, suggesting replies, and auto-closing noise — is grounded in the context configured here. Filling this out thoughtfully is the single highest-leverage thing an admin can do to improve the quality of every downstream AI feature.
The page has two parts: a brand identity form that captures the qualitative description of your business, and an auto-close strategy card that controls how aggressively Sift filters out irrelevant content.
Brand identity
The top section is a single form that captures everything Sift's AI needs to know about your business. Saving the form updates the org's customer profile, which is what every AI prompt in the product reads from.
The fields are:
Industry - A single-select dropdown of common industries. Selecting "Other" reveals an additional text field where you can specify a custom industry or space.
Competitors - A free-text field for listing competitor names so Sift can recognize when customers compare you to alternatives.
About Your Business - A multi-line textarea describing what your company does and what community content is worth monitoring. This is the AI's primary mental model of your brand.
What's Worth Acting On - A textarea describing what makes a post or message worth your team's attention. This drives Sift's actionability scoring across the entire workspace.
When to Auto-Close - A textarea describing the rules for when a conversation should be automatically closed versus kept open for your team.
Topics to Ignore - A managed list of tags or topic patterns the AI should explicitly de-prioritize as noise.
Auto-close strategy
The lower section is a single card titled Auto-Close Strategy that controls how aggressively Sift's AI auto-closes actions based on relevance and actionability. DMs are never auto-closed, regardless of strategy.
You pick one of four strategies:
Disabled
No auto-close. All actions require manual review.
Conservative
Only closes actions when the AI has very high confidence that the content is clearly irrelevant.
Standard (Default)
Closes low-relevance and low-actionability content when the AI has high confidence.
Aggressive
Closes more aggressively with lower confidence requirements.
Auto-Reply Bots
The Auto-Reply Bots page lists the conversational bots configured for your workspace. An auto-reply bot is a flow-based AI agent that handles incoming messages on your behalf - for example acknowledging a complaint, asking for a customer's order number, or routing a question to the right queue. Each bot has a visual flow that admins build and edit in the dedicated bot builder.
The page is a single table listing every auto-reply bot in the workspace, with columns:
Name - The bot's display name.
Created - The date the bot was first created.
Last Updated - The date of the most recent edit.
Rows are draggable. Drag a row up or down to change the priority order in which bots are evaluated against incoming messages - the first matching bot wins, so order matters when multiple bots could apply to the same message.
Creating a bot
The Create button in the top-right of the page opens the bot builder, a dedicated visual editor where you assemble a bot's logic - its trigger criteria, the message tree it walks through, the prompts it uses, and the actions it can take. Editing an existing bot opens the same builder, pre-populated with the bot's current configuration.
MCP Connections
The MCP Connections page lets you connect any Model Context Protocol-compatible AI client - Claude Code, Claude Desktop, ChatGPT, Cursor, Windsurf, Codex CLI, and others - to your Sift workspace. Once connected, the AI client can use Sift's analytics and investigation capabilities directly from inside its own interface, so analysts, engineers, and operators can ask questions like "How many open actions this week?" without leaving their preferred tool.
The page begins with a one-line introduction - Connect any MCP-compatible AI client to access Sift's analytics and investigation capabilities - followed by your workspace's unique MCP Server URL in a copyable code block. This URL is the only piece of information any client needs to connect; every setup guide below is just a wrapper around pasting it into the right place.
Setup guides
Below the URL is a card per supported client. Each card has the client's logo and name on the left, a numbered list of steps in the middle, copyable inline commands where appropriate, and a Docs link to the client's official MCP documentation.
The guides currently shipped on the page are:
Claude Code
Run claude mcp add - transport http siftgpt <URL>, then /mcp and authenticate.
Claude Desktop
Settings → Connectors → Add custom connector → paste the URL and sign in.
ChatGPT
Enable Developer Mode, Settings → Connectors → Add custom connector → paste the URL → enable per chat.
Cursor
Settings → MCP → Add new MCP Server (Streamable HTTP) → paste the URL → authenticate.
Windsurf
MCPs icon → Add Streamable HTTP server → paste the URL → authenticate.
Codex CLI
codex mcp add siftgpt --url <URL>, then codex mcp login siftgpt.
Each client follows its own OAuth flow when connecting. After pasting the MCP URL, the client redirects you through Sift's sign-in to authorize access; on success, the client confirms the connection and Sift's tools become available inside its interface. Every request the client makes is scoped to the user that authenticated, so MCP-based access respects the same permissions as the Sift app.
Last updated