What Riffly will never do.
Fake testimonials. Stock photos labeled as customers. Fabricated quotes.
You'll see real recruiters with their real LinkedIn profiles linked, or you'll see no testimonials at all. The homepage doesn't have a "trusted by" logo strip until there are real customers willing to be named. The press kit doesn't have inflated numbers waiting to be deflated by a TechCrunch story.
This is the commitment we'd most like to be held to. If you ever spot a quote on Riffly's site without a verifiable LinkedIn profile attached, email support@rifflylabs.com and we'll either prove it's real or remove it.
Inflated user counts. "10,000 recruiters use Riffly" before there are 10,000 recruiters using Riffly.
If a number appears on this site (paying customers, drafts generated, reply rates), it's the actual number. Round to the nearest realistic figure if useful, but don't multiply by anything.
The temptation here is real. Numbers drive conversion; small numbers look like a small product. The brand cost of being caught with a fabricated number is much larger than the conversion lift.
Automated sending. Auto-connect. "Set it and forget it" cold outreach.
Riffly drafts. You read every draft. You decide what to send. You click LinkedIn's own Send button. There is no scenario where a Riffly subscription causes a message to go out without a human reading it first.
The autonomous-outbound category (AI SDRs, automated LinkedIn sequences) collapsed in 2025 for good reasons, hallucinations, irrelevance, account bans. We're not chasing it because we don't think it works for recruiting and we don't want to ship a product that quietly hurts our customers.
Customer data sold to third parties. Candidate data shared with anyone but the model that's drafting.
Anthropic gets the candidate's profile snapshot at draft time so Claude can write the message. Stripe gets your billing details. Supabase hosts your account. That's the full list of who sees what. The security page has the sub-processor details.
Candidate data is not sold, traded, fed into training datasets, or used for advertising. Same goes for your account data.
Friction-engineered cancellation. Retention dark patterns. "Are you sure?" loops.
Cancel from the Stripe billing portal in two clicks. No retention call required. No "but what if we gave you 30% off?" pop-up. The money it would buy us isn't worth what it costs the brand.
What Riffly will do.
Publish what we know about reply rates, even when the numbers are unflattering.
Once we have a meaningful cohort (~1,000 paying users), we'll publish quarterly aggregate reply-rate data, by tone, role, industry. If it turns out one tone gets 35% replies and another gets 12%, both numbers ship. This is the asset that makes Riffly's brand credible to procurement-minded buyers; we don't get to keep it credible by hiding the bad numbers.
Tell you when LinkedIn's behavior makes a feature riskier.
The volume guardrail in the popup, the cadence advisory on saved searches, the structural-vs-behavioral distinction in /securityall of those exist because we'd rather lose a marginal upsell than have a customer's LinkedIn account restricted because we didn't tell them what we know.
Document material privacy events publicly.
If there's a data incident (breach, accidental exposure, sub-processor compromise), we'll write about it on /security and email affected users within 72 hours. Most companies write the public note only after the lawyers approve every word. We'll write it within the time legally required and update if facts change.
Keep prices honest as costs change.
If LLM costs go up and we have to raise prices, we'll raise prices and explain the math. If LLM costs go down and we can lower them, we'll lower them. We won't quietly raise the cap on what counts as "200 drafts a month" or quietly drop a feature out of Pro, both of those are price increases dressed as something else.
Honor the price you signed up at.
If we raise prices for new subscribers, your existing subscription stays at the price you signed up at, as long as it stays continuously active. The "grandfather your early users when you raise prices" principle is the right one, we won't quietly migrate existing accounts to higher tiers without explicit consent, and we won't drop a feature out of your tier as a backdoor price increase.
Why this matters.
The category Riffly competes in is full of trust failures right now. The AI SDR companies got caught faking customer logos. The automation tools collapsed when LinkedIn changed enforcement. The personality-prediction tools coast on perceived accuracy nobody can verify.
Riffly is built on the bet that recruiters who've been burned once will pay a small premium for a product whose claims can be checked. The five "never" commitments above are the ones that, if violated, would invalidate that bet. They're not marketing, they're the deal.
If you ever catch us drifting on any of these, write to support@rifflylabs.com. We'll take it seriously.