Today we're shipping one of the features I most wanted to get into the product: Detection Shields.
A lot of the conversation-AI category is pointed in one direction — making people more persuasive. We pointed this one the other way. Detection Shields exist to help you see when a conversation is being steered by tactics you didn't consent to, so you can respond from awareness instead of reacting from confusion.
Why We Put This Near the Center of the Product
When you operationalize a product that sits inside real conversations, you have to decide what your AI is allowed to help with and what it isn't. For us, the answer was simple: we don't build tools that help people manipulate others, and we do build tools that help people recognize when manipulation is being aimed at them. Detection Shields are the execution of that principle.
The reality is that manipulation tactics aren't rare, exotic things that only happen in extreme relationships. They show up in sales calls, performance reviews, contract negotiations, family group chats, customer service escalations, landlord disputes, HR mediations. The pattern is usually the same: one party has studied influence, and the other party is showing up to a conversation they didn't realize was tactical. Detection Shields close that gap.
Who This Is For (Hint: It's Not Just Personal)
When people hear "gaslighting detection," they often think of high-drama personal scenarios. Those matter. But the Shields were designed to be useful everywhere a conversation has stakes:
- Ops and HR leaders running mediations, terminations, or escalations where composure and pattern-recognition matter.
- Sales and account teams on the receiving end of procurement playbooks — pressure tactics, false urgency, invented competing bids.
- Legal and paralegal professionals prepping clients who have to hold their ground in depositions, mediations, or hearings.
- Anyone navigating a hard personal conversation where they want a second set of ears that won't be charmed, rushed, or guilt-tripped.
How Detection Shields Work
Detection Shields layer on top of any conversation mode. When active, the AI watches the live transcript for specific tactical patterns and surfaces them as alerts — the tactic name, a severity signal, and a suggested counter-response. The current lineup:
- Gaslighting Shield — reality-distortion language, memory denial, and "you're too sensitive" deflections. (Robin Stern's work on gaslighting in The Gaslight Effect informed a lot of how we framed these patterns.)
- Narcissist Defense — love-bombing / devaluation cycles, projection, and DARVO (Deny, Attack, Reverse Victim and Offender).
- Manipulation Detection — guilt trips, false urgency, emotional blackmail, information withholding, artificial scarcity.
- Lie Detection — inconsistencies between statements, too-good-to-be-true claims, evasive answers, pressure-to-decide patterns.
- Body Language Coach — coaches you on how you're projecting and, when you describe what you're seeing, helps you read the other side.
Shields run alongside the other safety primitives in the product — Safe Word, Witness Mode, and the Live Dynamics HUD — so you don't have to choose between getting help in the moment and staying in control of the session.
What Detection Shields Are Not
They're not a diagnosis of the person on the other end of the line. Detecting a tactic in one sentence is not the same as calling someone a narcissist, a liar, or a bad actor. The Shield tells you "that specific phrasing matches this pattern — here's how to respond if it is what it looks like." You stay the decision-maker. The AI is a second pair of ears, not a verdict.
That distinction matters to us, and it's baked into how the alerts are written.
Available Now
Detection Shields are included in Pro and Elite. You can activate any combination before a session. If you only ever use the product for Detection Shields and nothing else, that's a win by me — it's one of the clearest ways we can live up to the idea that technology should level the playing field instead of widen the gap.
— Tavish Hower, VP of Operations, Tactical Talk