otherapi_key

Faraday

Faraday lets you embed AI in workflows throughout your stack—to make your favorite tools perform even better

Verdict

The Faraday MCP lets your team manage customer prediction and segmentation resources directly from Switchy. By @mentioning Faraday in a Space, you can archive cohorts, datasets, outcomes, persona sets, and other prediction infrastructure without leaving the conversation. Data scientists and marketing ops teams get the most value — they can audit what's running, pause stale models, and clean up pipelines while discussing strategy with stakeholders. The MCP exposes 50 tools covering the full lifecycle of Faraday's prediction resources. You'll need a Faraday API key with write permissions, and archiving actions are immediate (no undo via Switchy).

Common use cases

Archive outdated prediction models during sprint planning
Audit active cohorts before a campaign launch
Clean up unused datasets after a migration
Pause persona sets while refreshing source data
Document which streams feed production dashboards

Integration

Vendor: Faraday
Category: other
Auth: API_KEY
Tools: 50
Composio slug: faraday

Tools

Archive Cohort
Tool to archive a cohort in Faraday. Use when you need to stop updates to a cohort while keeping it visible.
Archive Connection
Tool to archive a connection in Faraday. Use when you need to stop updates to a connection while keeping it visible.
Archive Dataset
Tool to archive a dataset in Faraday. Use when you need to archive a dataset and optionally its downstream resources. Archiving stops updates to the resource but keeps it visible.
Archive Outcome
Tool to archive an outcome in Faraday. Use when you need to stop updates to an outcome while keeping it visible.
Archive Persona Set
Tool to archive a persona set in Faraday. Use when you need to stop updates to a persona set while keeping it visible.
Archive Place
Tool to archive a place in Faraday. Use when you need to stop updates to a geospatial area or address set while keeping it visible.
Archive Scope
Tool to archive a scope in Faraday. Use when you need to stop updates to a scope while keeping it visible.
Archive Stream
Tool to archive a stream in Faraday. Use when you need to archive a stream and optionally its downstream resources. Archiving stops updates to the resource but keeps it visible.
Archive Target
Tool to archive a target in Faraday. Use when you need to archive a target and optionally its downstream resources. Archiving stops updates to the resource but keeps it visible.
Archive Trait
Tool to archive a trait in Faraday. Use when you need to stop updates to a trait while keeping it visible.
Create Account
Tool to create a new Faraday account with specified name and branding. Use when you need to provision a new account in Faraday.
Create Cohorts
Tool to create a new cohort in Faraday. Cohorts are groups of people that share certain traits or behaviors. Use when you need to define a new cohort from your datasets or from Faraday's US Consumer Graph. At least one of stream_name, trait
Create Dataset
Tool to create a new dataset in Faraday. Use when you need to define a new data source for analysis and predictions.
Create Outcome
Tool to create a new outcome. Outcomes are predictions about customer behavior such as likelihood to convert, churn, or purchase specific products. Use when you need to create a new predictive model.
Create Persona Set
Tool to create a new persona set. Persona sets segment cohorts into distinct customer personas based on shared characteristics and behaviors. Use when you need to generate customer personas for a cohort.
Create Place
Tool to create a new place in Faraday. Places are geospatial areas or sets of addresses that can be used as spatial filters when defining other resources. Use when you need to define a new geographic area for analysis or targeting.
Create Scope
Tool to create a new scope in Faraday. Scopes define the population and payload for predictions, specifying which cohorts to target and what outcomes to include. Use when you need to define output data instructions for a new scope.
Create Stream
Tool to create a new stream or find an existing stream by name. Use when you need to initialize a stream for event data collection.
Create Target
Tool to create a target in Faraday. Use when you need to set up a deployment that exports predictions and cohort data to external destinations like CRMs, ad platforms, or data warehouses.
Create Target Preview
Tool to start a preview delivery for a target. Use when you need to initiate a preview of target data before full deployment.
Create Trait
Tool to create a new trait in Faraday. Use when you need to define a new fact about a person or geographic entity that can be used in modeling, analysis, and reporting.
Create Webhook Endpoint
Tool to create a new webhook endpoint. Use when you need to register a webhook URL to receive notifications when resources are ready or encounter errors.
Delete Account
destructive
Tool to delete an account by its UUID. Use when you need to permanently remove an account from Faraday. The deletion is scheduled asynchronously and returns a 202 status on success.
Delete All Orphaned Traits
destructive
Tool to delete all orphaned traits from Faraday. Use when you need to clean up traits that are not associated with any source dataset.
Delete Cohort
destructive
Tool to permanently delete a cohort from Faraday. Use when you need to remove a cohort that is no longer needed.
Delete Connection
destructive
Tool to delete a Faraday connection by its UUID. Use when you need to permanently remove a connection.
Delete Dataset
destructive
Tool to delete a dataset from Faraday. Use when you need to remove a dataset permanently. The operation returns success when the dataset is successfully deleted.
Delete Outcome
destructive
Tool to delete an outcome by its UUID. Use when you need to remove an outcome from Faraday.
Delete Persona Set
destructive
Tool to delete a persona set. Use when you need to permanently remove a persona set from Faraday. Returns success status and HTTP code 204 on successful deletion.
Delete Place
destructive
Tool to delete a place by its UUID. Use when you need to remove a place from the system.
Delete Scope
destructive
Tool to delete a scope by its UUID. Use when you need to permanently remove a scope from the system. Returns 204 status code on successful deletion.
Delete Stream
destructive
Tool to delete a stream by its UUID or name. Use when you need to remove a stream that is no longer needed.
Delete Target
destructive
Tool to delete a target by its UUID. Use when you need to remove a target from Faraday.
Delete Upload
destructive
Permanently deletes a file from a Faraday dataset directory. Use this when you need to remove an uploaded file from a specific directory. This is a destructive operation that cannot be undone - the file will be permanently removed from the
Delete Webhook Endpoint
destructive
Tool to delete a webhook endpoint by its UUID. Use when you need to remove a webhook endpoint from Faraday. The operation returns 204 No Content on success.
Force Update Cohort
Tool to trigger a rerun for a cohort. Use when you need to force rebuild a cohort's data.
Force Update Connection
Tool to trigger a rerun for a Faraday connection. Use when you need to force refresh connection data.
Force Update Dataset
Tool to trigger a rerun for a dataset. Use when you need to force a dataset to update immediately rather than waiting for its normal schedule.
Force Update Outcome
Tool to trigger a rerun for an outcome resource. Use when you need to force rebuild an outcome, causing it to reprocess and update its predictions.
Force Update Persona Set
Tool to trigger a rerun for a persona set resource. Use when you need to force rebuild a persona set, causing it to reprocess and update its personas.
Force Update Place
Tool to trigger a rerun for a place. Use when you need to force rebuild a place's data.
Force Update Scope
Tool to trigger a rerun for a scope resource. Use when you need to force rebuild a scope, causing it to reprocess and update its data.
Force Update Stream
Tool to trigger a rerun for a stream. Use when you need to force rebuild a stream's data.
Force Update Target
Tool to trigger a rerun for a target. Use when you need to force rebuild a target's data.
Force Update Trait
Tool to trigger a rerun for a trait. Use when you need to force rebuild a trait's data.
Get Account
Tool to retrieve detailed information about a Faraday account by its UUID. Use when you need to fetch account details, status, branding, or billing information.
Get Accounts Billing
Tool to get billing information about your account. Use when you need to retrieve invoices, payments, or usage information for a specific account.
Get Cohort
Tool to retrieve a specific cohort by ID. Use when you need to fetch details about a specific cohort.
Get Cohort Membership Analysis
Tool to get cohort membership counts over time for a specific cohort. Use when you need to analyze how cohort membership has changed over days, weeks, and months.
List Accounts
Tool to list all Faraday accounts accessible by the current user. Use when you need to retrieve account information, check account status, or get account IDs for other operations.

Setup

Setup guide

11. Open your Switchy workspace and navigate to Settings > Integrations > Browse MCP Servers. 2. Search for Faraday and click Connect. 3. You'll be prompted to enter your Faraday API key — generate one in your Faraday account under API Settings with at least write access to cohorts, datasets, and outcomes. 4. Paste the key into Switchy and click Authorize. 5. Switchy confirms the connection and shows which tools are now available. 6. Open any Space and type '@Faraday list my cohorts' to test — the MCP should return your active cohorts. 7. To archive a resource, use '@Faraday archive cohort [cohort_id]' and confirm the action when prompted. 8. Check your Faraday dashboard to verify the resource now shows as archived.

What teammates see: by default, memories from Faraday are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.

Works well with

Top models

Compatibility data appears once enough Spaces have used this MCP together with a given model.

How Switchy teams use it

Not enough Spaces yet to publish anonymised usage stats (we require ≥ 50 Spaces per week).

Starter prompts

Audit Active Cohorts

@Faraday show me all active cohorts in our account, grouped by creation date, and flag any that haven't been updated in the last 90 days.

Open in a Space →

Archive Stale Dataset

@Faraday archive the dataset with ID 'ds_12345' and include all downstream resources that depend on it.

Open in a Space →

Review Outcome Definitions

@Faraday list all outcomes we've defined, including their target variables and the datasets they reference.

Open in a Space →

Pause Persona Set Updates

@Faraday archive the persona set named 'Q1_High_Value_Segments' and confirm it's no longer receiving updates.

Open in a Space →

Map Stream Dependencies

@Faraday show me all streams connected to our production cohorts and highlight any that are archived or inactive.

Open in a Space →

Example outputs

Illustrative - representative of the model's voice and quality, not literal recordings.

Prompt

@Faraday show me all active cohorts and their current sizes, then tell me which ones haven't been updated in the last 90 days

Output

Based on your Faraday account, you have 12 active cohorts. Here's a summary: 'High-Value Customers' (47,203 records, last updated 14 days ago), 'Lapsed Subscribers' (8,941 records, last updated 6 days ago), 'Q1 Campaign Targets' (22,156 records, last updated 127 days ago), and 9 others. Three cohorts haven't been refreshed in over 90 days: 'Q1 Campaign Targets', 'Holiday 2023 Prospects' (last updated 134 days ago), and 'Beta Testers' (last updated 201 days ago). These stale cohorts may need archiving or a manual refresh depending on your current campaigns.

Notes

This example demonstrates read-heavy data inspection across multiple Faraday resources. The MCP can surface metadata about cohorts, datasets, and other predictive modeling artifacts. Requires API key with read permissions on cohort resources. Actual cohort names and sizes will vary by your account state.

Prompt

@Faraday archive the 'Q1 Campaign Targets' cohort and the 'Holiday 2023 Prospects' cohort since those campaigns are over

Output

I've archived both cohorts in your Faraday account. 'Q1 Campaign Targets' (22,156 records) and 'Holiday 2023 Prospects' (11,847 records) are now marked as archived. They remain visible in your account for historical reference but will no longer receive automatic updates or consume active processing resources. If you need to reactivate either cohort for a future campaign, you'll need to unarchive them through the Faraday dashboard or API.

Notes

This showcases Faraday's resource lifecycle management. Archiving is a soft-delete that preserves data visibility while stopping compute overhead. The MCP can batch multiple archive operations in one prompt. Note that archiving a cohort doesn't cascade to dependent resources unless explicitly requested—downstream predictions or persona sets remain active.

Prompt

@Faraday list all my datasets and outcomes, then recommend which ones I should archive based on their last-used dates and whether they're referenced by active cohorts

Output

You have 8 datasets and 5 outcomes in your Faraday account. After cross-referencing usage: 'Purchase History 2022' dataset (last used 289 days ago, not referenced by any active cohort) and 'Email Engagement Outcome' (last used 176 days ago, not referenced) are candidates for archiving. 'Customer Demographics' dataset is 412 days old but still powers 3 active cohorts, so keep it. 'Churn Prediction Outcome' hasn't been used in 203 days and has no active dependencies—safe to archive. I recommend archiving the 3 unused resources to reduce clutter and focus on current modeling workflows.

Notes

This example pairs Faraday's metadata with AI reasoning to surface actionable cleanup recommendations. The MCP provides raw resource data; the AI synthesizes usage patterns and dependency graphs. Requires read access to datasets, outcomes, and cohorts. Recommendations are heuristic—verify business context before archiving anything tied to compliance or historical reporting.

Use-case deep-dives

Customer segmentation cleanup after campaign

When Faraday MCP helps marketing ops retire stale cohorts

A 6-person growth team runs monthly acquisition campaigns and builds new Faraday cohorts for each test. After three months they have 40+ cohorts cluttering the workspace and slowing down model selection. The Faraday MCP is the right call here because it exposes archive operations for cohorts, datasets, outcomes, and persona sets—letting an AI agent bulk-retire old segments without deleting the audit trail. The team can prompt Switchy to "archive all cohorts from Q1 campaigns" and the agent handles the API calls in one pass. This works cleanly if your Faraday account has under 200 total resources; beyond that threshold you hit rate limits and need to batch the requests manually. If you're archiving more than twice a quarter, build this into a scheduled workflow instead of ad-hoc prompts.

Geospatial targeting handoff between analyst and ops

When this MCP bridges the gap between place creation and deployment

A retail analytics team defines new store trade areas in Faraday every week, then hands off the place IDs to a media buyer who activates them in ad platforms. The handoff breaks when the analyst forgets to archive test places or the buyer can't tell which places are live. The Faraday MCP solves this by giving both roles a shared AI interface: the analyst prompts Switchy to "show me all active places created this week" and the buyer prompts "archive places tagged 'test-only'". The 50-tool scope includes place and scope management, so the agent can read, filter, and archive without switching contexts. This setup works for teams running under 10 place updates per week; higher velocity needs a dedicated orchestration layer because the MCP doesn't expose bulk operations or webhooks. If your workflow is purely read-only, skip the MCP and use Faraday's dashboard.

Data pipeline audit before model retrain

When Faraday MCP helps data teams verify upstream dependencies

A 3-person data science team retrains prediction models monthly and needs to confirm that all input datasets and streams are current before kicking off the job. Manually checking 15+ resources in the Faraday UI takes 20 minutes and introduces errors. The Faraday MCP is a clean fit because it exposes dataset and stream metadata through natural language: the team prompts Switchy "list all datasets updated in the last 7 days and flag any archived streams" and gets a structured answer in seconds. The agent can also archive outdated datasets as part of the same conversation, keeping the workspace tidy. This works if your pipeline has under 30 dependencies; larger graphs need a dedicated orchestration tool because the MCP doesn't support dependency traversal or impact analysis. If you're only checking status once a month, the manual workflow is faster than setting up the integration.

Frequently asked

What does the Faraday MCP do in Switchy?

It connects your team to Faraday's customer prediction platform, letting you manage cohorts, datasets, outcomes, persona sets, and geospatial scopes from Switchy's AI workspace. You can archive resources, query predictions, and coordinate data pipelines without switching tabs. The MCP exposes 50 tools covering Faraday's core resource types—cohorts, connections, streams, outcomes, and places.

Do I need a Faraday API key to connect this MCP?

Yes. The Faraday MCP uses API key authentication, so you'll need to generate a key from your Faraday account settings before connecting it in Switchy. Whoever connects the MCP should have admin or developer permissions in Faraday to ensure the key has full access to cohorts, datasets, and outcomes your team needs to manage.

Can the Faraday MCP create new cohorts or just archive them?

The representative tools shown focus on archiving resources—cohorts, datasets, outcomes, persona sets, scopes, streams, and connections. Faraday's API typically supports creation and updates too, so the full 50-tool set likely includes those operations. Check the complete tool list in Switchy after connecting to see create and update capabilities alongside the archive functions.

How is this different from using Faraday's web dashboard?

The MCP lets you script and automate Faraday operations from Switchy's AI chat—archive stale cohorts in bulk, query outcome definitions, or coordinate data refreshes—without logging into Faraday's UI. It's faster for repetitive tasks and lets non-technical teammates trigger predictions or manage datasets using plain English prompts instead of navigating Faraday's interface.

Who on the team should connect the Faraday MCP?

Whoever owns your Faraday account and has API key access—usually a data lead, growth engineer, or marketing ops person. They'll generate the API key and paste it into Switchy once. After that, anyone on your Switchy workspace can use the MCP to query cohorts or archive resources, but the key's permissions control what actions are allowed.