You are part of a private alpha test of Verity by Xeit AI. Verity is a layer that improves the reasoning of any large language model (LLM) — like a turbocharger for an engine. At the moment, she runs on Anthropic's Haiku 4.5 (without extended reasoning), but she works on any model: frontier or local, large or small.

Use her as you would any LLM — for productivity, research, curiosity, anything. Verity is free during the alpha, so use her as much as you like.

During the alpha, each user has a daily usage budget of 1,000,000 counted tokens. The budget resets for everyone at midnight UTC.

Before you begin, here is our agreement on your data — by continuing, you accept it. Your conversations, uploaded documents, and usage are retained and used to develop and improve Verity. Inside our systems, your activity lives under an internal identifier rather than your name — we collect only the email address needed to sign you in — and before any of your data is used for development, it is de-identified: identifiers are replaced with one-way hashes, and your email and name never enter development datasets. Your conversations and documents are processed by Anthropic's Claude models to generate responses, and uploaded documents are mirrored to Anthropic's servers for that purpose, subject to Anthropic's API data policies. We never sell or share your data, we do not build profiles of you, and nothing you write trains the underlying models. You may ask us to delete your account at any time, which removes your data from our systems and from subsequent development datasets. Full details: Privacy notice.

You will find a space for your conversations and portfolios by clicking [. . .] in the upper left. Use the thumbs-up and thumbs-down on any response to send us feedback, and you can reach us from the bottom of your space.

Otherwise, have fun and enjoy Verity!

Best,
Ashton, Maf, and Zack