canonical: https://jentic.com/apis/faraday.ai/faraday # Faraday API Faraday is a customer prediction platform whose API exposes 80 endpoints to manage accounts, datasets, cohorts, outcomes, persona sets, scopes, targets, traits, connections, streams, recommenders, places, and webhooks. The surface lets a team load customer data, define what to predict, and consume scored cohorts back into downstream systems. It is aimed at marketing and growth teams that want lift, churn, or conversion predictions without standing up a data science stack. ## For AI agents Manage datasets, cohorts, outcomes, and recommenders to produce customer predictions on the Faraday platform. Useful for churn, lift, and conversion modelling flows. ## Scope Does not handle marketing campaign delivery, ad bidding, or warehouse query execution - use for customer prediction modelling and scored audience generation only. ## Capabilities - Create datasets that hold customer records for prediction - Define cohorts that segment customers for modelling - Set up outcomes that describe what the model should predict - Build target audiences scored by an outcome - Manage traits and attributes used as model features - Configure connections and streams to load data from external systems - Publish recommenders that surface next-best-product or content scores ## Use cases ### Churn Prediction Pipeline A subscription business loads customer data into a Faraday dataset, defines a churn outcome, and scores a target audience to surface at-risk customers. The flow uses /accounts and /attributes for setup, then datasets, outcomes, and targets to materialise the prediction. Reduces what would otherwise be a multi-week data science build to a configuration exercise on the Faraday API. Example prompt: POST a dataset describing customer events, POST an outcome predicting churn, then POST a target referencing the outcome and read the scored audience back through the targets endpoint. ### Lift Modelling for Campaigns A growth team builds a lift model over a control and treatment cohort to estimate the incremental effect of a marketing campaign. The Cohorts and Outcomes endpoints define the experimental groups and the model output, and the Targets endpoint produces the scored list. Works for channel teams that want lift estimates without a custom modelling stack. Example prompt: Create control and treatment cohorts via /cohorts, define a lift outcome via /outcomes, then create a target that scores customers and read the ranked list. ### Next-Best Recommender An e-commerce site uses Faraday's recommenders endpoints to publish next-best-product scores for each customer. The flow combines a dataset of past purchases, a recommender configuration, and an output stream that pushes scores to the merchandising system. Suited to mid-market retailers that want personalised recommendations without rolling their own ranking model. Example prompt: POST a recommender configuration referencing the products dataset and a customer cohort, then read scored recommendations through the targets and streams endpoints. ### Data Connection Setup A data engineer wires Faraday to the warehouse by creating a connection and one or more streams that ingest customer events. The /connections and /streams endpoints handle the load side of the platform, and downstream datasets reference the streams as their source. Designed for one-time integration work that then feeds many downstream models. Example prompt: POST a connection record describing the warehouse, POST stream definitions for the customer and event tables, then create datasets that reference the streams. ### AI Agent Growth Assistant An AI agent assisting a growth team uses Jentic to drive Faraday - answering questions like 'who is most likely to churn next month' or 'create a target for the holiday campaign'. Jentic exposes the 80 Faraday operations with their input schemas, so the agent can navigate datasets, outcomes, and targets without reading the developer reference. Useful for embedded marketing copilots. Example prompt: Call jentic.search with 'create a Faraday target for churn', load the /targets operation, and execute it with the churn outcome id and the customer cohort, returning the resulting target id. ## Key endpoints | Method | Path | Description | | --- | --- | --- | | GET | `/accounts/current` | Read the current account | | GET | `/accounts/current/usage` | Read current account usage | | GET | `/attributes` | List attributes | | POST | `/attributes` | Create an attribute | | GET | `/accounts` | List accounts | | POST | `/accounts` | Create an account | ## Key resources - **Accounts** — Read account info and usage on the workspace - **Attributes** — List and manage attributes used as model features - **Datasets** — Hold customer data used by models - **Cohorts** — Segment customers for modelling - **Outcomes** — Define what a model should predict - **Targets** — Produce scored audiences from an outcome - **Recommenders** — Configure next-best item or content scoring - **Connections and Streams** — Load data from external systems ## Why Jentic - **Setup:** Wiring the Faraday API by hand means minting its bearer token, targeting the api.faraday.ai/v1 host, and navigating an 80-endpoint prediction surface yourself. Through Jentic you install once, import the Faraday API from the API Directory, store the token once, and your agent calls it. - **Permission scoping:** Faraday's core routes here act on collections with the details in the request body, so you limit the agent to the operations it needs, such as reading account usage, listing attributes, or creating a target. You choose those operations, so account creation is not included unless you add it. - **Credential handling:** Your Faraday bearer token is stored once, encrypted, by your own Jentic One instance and attached to the Authorization header at execution time. It never enters the agent's prompt, logs, or context. - **Discovery method:** Agents search Jentic by intent such as 'create a Faraday target' or 'list Faraday outcomes', and Jentic returns the matching operation with its input schema so the agent calls the right endpoint without browsing the reference docs. ## Related APIs - **Amplitude** — Product analytics with built-in cohorts and predictive segments - **Mixpanel** — Product analytics platform with cohort and behavioural segmentation - **Segment** — Customer data pipeline that often loads events into Faraday - **Snowflake** — Warehouse that often serves as the source for Faraday connections ## FAQ ### What authentication does the Faraday API use? Faraday uses HTTP bearer auth - a token is passed in the Authorization header. Through Jentic the token is stored encrypted in the vault and attached to outgoing requests so the raw value never enters the agent's context. ### Can I create a churn prediction with the Faraday API? Yes. Create a dataset of customer events, define an outcome describing churn, then create a target that scores the customer cohort against the outcome. The resulting target exposes scored customers through the targets endpoint. ### What are the rate limits for the Faraday API? Per-second rate limits are not declared in the spec. Faraday governs throughput by plan - read `/accounts/current/usage` for the workspace's current usage and remaining capacity rather than relying on per-call quotas. ### How do I list available attributes through Jentic? Run pip install jentic, call jentic.search with 'list Faraday attributes', and execute the returned /attributes operation. The bearer token is replayed from your Jentic One instance on each call. ### Does the Faraday API support recommenders? Yes. The recommenders resource publishes next-best-item or content scores for each customer in a cohort. Configure the recommender, attach a dataset of historical interactions, and read scores back through the targets and streams endpoints. ### How do I load customer data into Faraday? Create a connection that points at the source system, define streams for the relevant tables, and reference the streams from a dataset. The /connections and /streams endpoints handle the load side and the dataset endpoint binds the loaded data into modelling. ### Can I limit what my agent is allowed to do with the Faraday API? Yes. Because Jentic One is self-hosted, you run your own instance and your own rules decide which Faraday operations and credentials the agent may use. You can allow just the endpoints a task needs, such as reading account usage, listing attributes, or creating a target, while leaving out anything else. Since you choose the operations explicitly, actions like creating an account are not available to the agent unless you add them.