Use Metatextai from Claude
Add the Metatextai connection to a Space that runs Claude, then use it from every channel in that Space.
Create a Metatextai connection, then control which Spaces can use its approved tools and data without pasting credentials into prompts or threads.
Add the Metatextai connection to a Space that runs Claude, then use it from every channel in that Space.
Add the Metatextai connection to a Space that runs Codex, then use it from every channel in that Space.
Connect your own OpenClaw to Sidekick while type.com provides the conversations, permissions, connections, skills, and automations.
Metatextai focuses on applied natural language processing: generating text, classifying content, and checking messages against safety guardrails. From a type.com thread, Claude can run a classification job, test a guardrail policy, or evaluate a draft reply before it goes out, keeping that review inside the same workspace as the rest of the work.
Connect Metatextai once, then decide which Spaces can use it in threads, skills, automations, and coding work.
Send a batch of messages to a trained Metatextai model and get back labels with confidence scores, useful for routing support tickets or tagging feedback.
Classify these 50 support tickets with our Metatextai model and tell me how many are billing related.
Create a policy guardrail for a specific use case, then evaluate sample messages against it to see what would get flagged before the policy goes live.
Create a guardrail policy blocking medical advice in our support app, then evaluate these five draft replies against it.
Generate a conversational response through chat completions when a quick draft is needed, then hand it to the team for review in the same thread.
Use Metatextai chat completions to draft a reply to this customer complaint, then I'll review it before sending.
Generates a conversational, OpenAI-style response to a prompt, useful for quick drafts a teammate can review.
Runs text through a trained Metatextai model and returns labels with confidence scores for sorting or routing content.
Defines a new guardrail for an application, setting rules that later messages get checked against.
Removes a guardrail policy by ID once it is outdated or no longer needed for an application.
Checks a model's output against a set of policies and returns any violations or suggested corrections.
Metatext exposes AI APIs for language-model routing, generation, guardrail evaluation, and NLP model workflows such as classification, with API-key authentication and HTTPS JSON requests.
Yes. With a Metatextai connection shared to a type.com Space, Claude can classify messages, evaluate them against guardrails, or draft a reply, and the results stay visible to the whole team in the thread.
Metatextai uses an API key. An admin adds it once when the connection is created, then picks which Spaces can use it, so the key is never pasted into a prompt.
Yes. A recurring check, such as evaluating a day's flagged messages against your policies, can run as a type.com automation that reports results to a channel.
Not exactly. type.com uses connectors and connections to give selected Spaces access to approved app tools and data. Some connectors use hosted MCP, while others use OAuth, API keys, service accounts, or custom APIs.
Connect BigML to Claude in type.com to create projects, set up external connectors, and check stored configurations before training a model.
Connect BigPicture.io to Claude in type.com to identify website visitors by IP, find company domains, and pull profiles from a thread.
Connect Bitquery to Claude in type.com to query historical and real-time blockchain data across 40-plus chains from a shared thread.
Connect Bright Data to Claude in type.com to trigger site crawls, browse scrapers, and filter datasets from a shared team thread.