First pilot: open
The named-logo slot for the first Tasfi Guard production pilot. Replaced with the partner logo when the first pilot ships.
Pilot program
Builders shipping Islamic AI. Source-grounded verification, on receipts.
Built by Muslims, for the Ummah. Tasfi Guard sits behind your answer pipeline and verifies generated Islamic content before a Muslim user sees it. Pilots are paid, scoped, and produce a privacy safe reliability report you can share with your team, your investors, and your scholars.
Second pilot: open
The named-logo slot for the second pilot. Slots are kept honest. We do not pad this page with imaginary partners.
Third pilot: open
The named-logo slot for the third pilot. Three live pilots is the trigger to revisit the institutional sales motion.
Who Tasfi pilots are for
Builders first. Institutions later.
The current pilot window targets builders. Institution outreach reopens when the named triggers in our methodology fire.
In scope
Islamic education platforms, Quran apps with AI features, Muslim publishers shipping AI-assisted content, masjid software vendors adding answer flows, AI teams adding Islamic content features to a broader product.
Out of scope for this window
Seminaries, halal certifiers, regulated Muslim-facing platforms, school districts, hospital systems. We will take meetings. The pilot list reopens for these buyers when the trigger conditions fire.
Not eligible
Consumer chatbot pivots that want Tasfi to act as a fatwa surface. Halal certification or ingredient claim products. Anything that would have us replace a scholar, not verify a generated answer.
What a pilot includes
Three deliverables, on a 30 to 60 day window.
Integration
Account-owned API key, integration of /v1/verify or /v1/verify/batch into your pre-display path, walkthrough of risk flags and recommended actions, source bundle pin recorded against your account.
POST /v1/verify
Live verification
Tasfi Guard runs against your real generated content. Metadata-only audit ledger accumulates. Risk flags route to your team. Scholar escalation goes through the review packet workflow.
GET /v1/usage
Reliability report
Privacy-safe report scoped to your account. Source bundle identity, pass and fail counts by risk category, false-pass evidence by claim type, named recommendations for your next release.
GET /v1/pilot-report
Scholar escalation
For high-stakes fiqh or aqeedah outputs, route the escalation packet to the Tasfi scholar review workflow. Metadata only, no submitted text stored by default.
POST /v1/review-packets
Privacy contract
What you ship to Tasfi, and what Tasfi keeps.
Pilots are metadata-only by default. Submitted question text, answer text, and raw citation text are not stored. Every verification leaves a signed Trust Receipt with a SHA-256 input hash and the bundle identity. Read the Trust Receipt schema for the field-by-field guarantee.
Private content retention only happens under a written review agreement with explicit scope. The default is no retention.
Pilot verification call
curl -s https://tasfi.app/v1/verify \
-H 'authorization: Bearer <pilot-key>' \
-H 'content-type: application/json' \
-d '{"question":"...","answer":"...","citations":[...],"domain":"education"}'What we will not say
Boundary claims we never make in a pilot.
Not fatwa
Tasfi Guard does not issue rulings, certifications, or scholarly opinions. It verifies sources and flags risk.
Not halal certification
Tasfi does not certify products, ingredients, services, or vendors. The Maktaba lifecycle is about source rights and provenance, not commercial certification.
Not scholar replacement
High-stakes religious content routes to the scholar review workflow. The Guard verifier never claims to substitute for qualified human review.
How to start
Tell us what you ship. We will reply within 48 hours.
Describe your product, where Islamic content surfaces, and your current review process. We reply with fit assessment, a scoped pilot proposal, and a path to a signed agreement.
Prefer email? founder@tasfi.app