Working Group Charter: Ansari 4

Last updated 31 July 2026

IASER is adopting Ansari as an IASER project. Ansari has been answering Islamic questions since 2023 — 140,000+ conversations across 25+ languages, with benchmark-verified accuracy — and moves under IASER for a long-term institutional home: access to funding, a volunteer network, and governance so theological and editorial decisions have real backing. This working group delivers Ansari 4, the first release of that new era.

Group name

Ansari 4

Mission

Make Ansari the most accurate, most helpful, free, open-source Islamic question-answering application in the world — and keep it that way.

Deliverable

A launched Ansari 4: one modern app (a single React Native codebase for web, iOS, and Android) on an open-sourced production backend, living in the github.com/iaser-ai/ansari monorepo, relaunched under the canonical home askansari.ai with a real marketing effort behind it — evaluated against Ansari 3 on JaleesBench before public release.

In scope

• Rebuild the frontend from scratch as one React Native app (web, iOS, Android), built with the AI-assisted Codev approach.
• Open-source the production backend — a return, not a debut: the modern engine that actually serves users, cleaned up and published as the foundation everyone builds on.
• Market the relaunch deliberately; develop a dedicated marketing plan and find someone to lead it.
• Keep Ansari available to other AIs — API, MCP server, and the Ansari Skill — so questions asked inside any assistant come back grounded and cited.
• Safety systems for people in distress: recognize hard moments, respond with care grounded in the faith’s support through hardship, and point to real-world help where warranted.
• Consolidate the brand on askansari.ai: redirect all other owned domains, invest in SEO, differentiate from the unaffiliated “Ansari AI” at ansari.com.

Out of scope

Each of these was considered carefully and deferred deliberately — they collide with the promise we are least willing to break: Ansari stays free. Revisited as model costs fall and IASER funding grows; a donor underwriting any of them (kept free for all users) brings it forward.

• Deep research mode (~$2–3 per report at scale; the Ansari Skill already covers this through tools people use).
• Real-time voice (first in line the moment costs drop below ~1¢/minute — the accessibility case is strong).
• An always-on daily-life assistant (high-frequency by design; the opposite of what a donation-funded free service can absorb).
• Multi-agent “Shura” deliberation (~5× cost per query; we keep the goals — sources shown, scholarly agreement and difference shown — without the swarm).
• Any paid tier. Free is a choice, not a limitation.

Duration, timeline, and milestones

Roughly five months to public launch, in this order: feedback and alignment; assemble the frontend and backend teams; create the Ansari 4 foundation; marketing plan development in parallel.

T₀ = the chartering date (co-leads confirmed). Offsets firm up at the first milestone review.

Resources needed

• Inference and serving costs: a grounded conversation runs on the order of $0.20 (≈100k tokens of retrieved context) at ~290 conversations/day and growing.
• Marketing budget for the relaunch (amount set by the marketing plan).
• From IASER: funding channels (grants and donors), the volunteer network, and governance/advisory backing for theological and editorial decisions.

Audience

Muslims everywhere with a question about their deen — free, in 25+ languages, with no paywall on knowledge; other AI assistants (via API, MCP, and the Ansari Skill) whose users ask Islamic questions; and the developers and volunteers who build on the open codebase.

How we work

The work lives in github.com/iaser-ai/ansari (backend/, frontend/, docs/), built with the Codev methodology. Meeting cadence set by the co-leads at chartering.