Safety as design, not feature.

On most dating apps, safety is a tier. On Saath, it is the floor. Here are the specific design decisions we made before the app opens, and why each one is structural rather than promotional.

The category problem.

Most dating apps treat safety the way airlines treat legroom. Standard exists, but the better version is upsold. The block button is free; the "see who tried to message you before they were filtered out" feature costs nine dollars a month. Photo moderation is real but vague. Reports go into a queue with no published response time. The block list is invisible to anyone but you, so the same bad actor cycles through accounts.

This is not a small problem. It is the central reason women in particular treat new dating apps as suspect by default. The marketing pages say safety is our priority. The product, in dozens of small decisions, says safety is a revenue lever. The gap between the two is what people learn to read in their first week on a new app.

We do not want that gap to exist on Saath. Below are the specific decisions that make it so.

What is free on Saath that is paid elsewhere.

Block, report, hide, unmatch. All four are free for every user. Block is one tap and immediate. Reported users go into a real moderation queue (more on that below). Hide removes you from the other person's discovery feed without notifying them. Unmatch ends the match silently from both sides with no "they unmatched you" surface.

Cultural filters. Community, language, religion, sect, ethnicity, city, age, gender, and dating intention. All free, all the time. We will not gate these behind Plus because gating them effectively makes Plus a tax on dating someone specifically Bengali, or Sikh, or Pakistani-American. A premium tier on cultural identity is not what we are willing to build.

Saath Spark. Our daily one-on-one conversation pool. Always free for everyone, never a Plus entitlement. Connection is not a feature we monetize.

The intro system, as anti-spam infrastructure.

The single biggest source of low-quality interactions on dating apps is bulk-low-effort messaging. The fix on Saath is not a filter or an AI classifier. It is a structural cap on how much message volume a single account can generate:

• Two intros per day on the free tier, accumulating to a small bank of up to four if you skip days. Twelve per day on Plus, which is meaningfully more but still finite.
• Sixty messages per hour per match, enforced server-side at the database. You cannot send a hundred messages in an hour to someone who is not responding. The system stops you.
• Eight likes per day on the free tier. Twenty on Plus. The cap is the point. Eight thoughtful likes beats forty distracted ones, and it is a meaningful change in the discovery feed quality for everyone else.

Caps are unfashionable in dating-app product design. The conventional wisdom is that more swipes equals more retention equals more revenue. We think the conventional wisdom is wrong, and we think the conversational quality on Saath will prove it.

Human photo moderation.

Every profile photo goes through a moderation queue before the profile appears in discovery. Not an AI classifier with a hidden confidence threshold. Real humans, looking at every photo. The trade-off is that the first profile photo takes longer to approve. We think that is the right trade.

Reported photos are re-reviewed by a different reviewer. Photos that fail review are not silently rejected. The user is told what failed and given a clear path to upload a new one.

The no-caste position, as a safety decision.

We have written about why Saath does not have a caste filter at length. The shorter version, in the context of safety: caste filtering on dating apps is a structural harm to women in particular. It is the mechanism by which family-mediated dating reproduces caste-based exclusion at the population scale. Removing it is not a feature decision. It is a safety decision.

We enforce it at the database layer. A PostgreSQL trigger scans bios, prompts, and intro messages for caste-coded language and rejects the insert with a clear error. The phrase list is mirrored in client-side code so people see the warning before they submit, and the two lists are kept in sync by a test that fails CI if they diverge. We did not want to ship this as a vague promise. It is a code path.

What the moderation queue actually does.

When a report comes in, three things happen.

One: the reported user does not know they were reported. Hide rules apply automatically while the report is in review.

Two: the report goes to a human reviewer with the relevant context: the profile, the conversation snippet if applicable, the reporter's stated reason, prior reports against this account.

Three: a decision is made within hours, not days. Account actions range from warning to temporary suspension to permanent ban depending on severity and history. Repeat reports against the same account trigger faster escalation. Underage reports trigger automatic temporary suspension while we verify.

The published target is hours, not a generic we will get back to you. We will be public about whether we are hitting it.

What is deliberately missing.

The things Saath will not ship, for reasons of trust:

• Unlimited swiping. The cap exists for a reason. It would be commercially better to remove it. We are not going to.
• Paid boosts that make free users invisible. Free users always see the same discovery feed as paying users. We do not bury one set of users to surface another.
• Read receipts behind a paywall. If we ship read receipts, they are bidirectional and free. We will not sell one side of a conversation visibility into the other.
• Leaderboards, popularity scores, or ranked tiers of users. No user is a slot to be filled or a tier to be promoted into.
• Sentiment-based intro screening. Intros are between two people. We are not going to score them for tone with an AI model in the loop. The cap on intros is the right place to discipline volume.

The phrase under the phrase.

When we say safety is foundational, what we mean underneath is that the boring product decisions get to be the loud ones. The intros cap matters more than any badge. The block-is-free policy matters more than any landing-page hero. The human photo moderation queue matters more than any blog post about safety, including this one.

We are publishing this because we think the gap between what dating apps say about safety and what they build is the single largest reason new apps are met with skepticism. Closing that gap on Saath is the work, not the marketing.