✦ Feason

Christian context layer · Retrieval for AI

Christian AI Grounding API
Sources your model can cite.

Christian AI grounding gives a language model relevant Scripture and theological sources before it generates an answer. Feason returns a prompt-ready context block plus separate citation, provenance, token-usage, and tradition-coverage metadata.

Get an API keySee an API response
POST /contextualize endpointhttps://www.feason.com/api/context/v1/contextualize

How it works

A Christian context layer built for inspection.

Ground AI answers in inspectable Scripture, creeds, councils, historical theology, and named Christian traditions with token-budgeted context and verifiable citations.

Prompt-ready context

Receive bounded Markdown designed to sit immediately before the user’s question in a model request.

Verifiable citations

Keep citation objects separate from generated prose and verify quotations against immutable corpus releases.

Visible coverage limits

Requested, represented, and missing traditions stay explicit so retrieval gaps cannot masquerade as consensus.

API example

Concrete JSON.
Traceable evidence.

The REST response mirrors the same versioned Christian source corpus available through Feason’s MCP server. Every evidence item retains its passage ID, canonical reference, source provenance, and verification status.

See all six API operations →
RequestPOST /contextualize
curl https://www.feason.com/api/context/v1/contextualize \
  -H "Authorization: Bearer $FEASON_CONTEXT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "How did early Christians speak about grace?",
    "traditions": ["catholic", "orthodox", "reformed"],
    "maxTokens": 2000
  }'
Response shapeapplication/json
{
  "corpusVersion": "theology-v7.0.0",
  "context": "# Feason Context\n...",
  "citations": [{ "label": "E1", "passageId": "..." }],
  "coverage": {
    "representedTraditions": ["catholic", "orthodox"],
    "missingTraditions": ["reformed"]
  },
  "usage": { "estimatedTokens": 1842, "truncated": false }
}

A retrieval layer for Christian AI applications

Feason Context sits between your application and its language model. Your service sends a theological question, Feason retrieves relevant passages from a curated corpus, and the contextualize endpoint formats the strongest evidence within the requested source and token limits. The model remains interchangeable because the context contract is provider-neutral.

Beyond Bible verse retrieval

Christian questions often require more than a list of verses. The corpus connects Scripture with creeds, councils, patristic writings, historical works, confessions, theological terms, and documented tradition positions. Each category stays visible, allowing an application to distinguish primary text from later interpretation.

Grounded answers without synthetic consensus

Feason does not ask a model to average Christian traditions into one anonymous position. Disputed claims remain attributed, areas of agreement and disagreement remain separate, and missing coverage is reported as a corpus limitation—not as proof that a tradition has no teaching.

Use cases

What developers can build.

Feason supplies the evidence layer. Your application remains responsible for presentation, pastoral boundaries, and the conclusions it draws.

Christian chat assistants

Place inspectable evidence behind conversational answers without locking the product to one model provider.

Bible study applications

Connect a question to Scripture, historical context, theological terms, and named interpretive positions.

Research and publishing

Build source-backed briefs, editorial research tools, and citation-aware drafting workflows.

Questions

Frequently asked.

Short answers to the implementation and methodology questions teams usually ask first.
What is Christian AI grounding?

Christian AI grounding supplies a model with retrieved Scripture and theological sources before it answers. Feason packages exact passages, provenance, citations, tradition coverage, and caveats into a bounded context block that can be placed directly in a model prompt.

Is this the same as training a Christian language model?

No. Grounding happens at request time and does not require retraining a model. Your application retrieves a versioned evidence package from Feason and gives that package to the model alongside the user’s question.

Can I restrict context to particular Christian traditions?

Yes. The contextualize and evidence-packet operations accept requested traditions. The response reports which requested traditions are represented and which are missing from the retrieved evidence.

How are citations verified?

Every included passage has a stable passage ID and provenance. The citation verification endpoint can compare a passage or quotation against an immutable corpus release and report exact, normalized, hash-only, or mismatch status.

Feason Context · REST API 1.2

Build with Christian sources your users can inspect.

Create a scoped keyConnect through MCP