AI models: meet Dawn

Dawn is nancit’s first AI model, built with OpenAI’s GPT-6 Luna and trained for business: what it can do, how it keeps you in control, and its effort levels.

Meet Dawn 1.0, nancit’s first model

Dawn 1.0 is the AI model behind nancit, and the first in our family of models. It launches with nancit Beta 1.0 and is built to do one thing exceptionally well: help you run your business through conversation.

Dawn is built in conjunction with OpenAI’s GPT-6 Luna, one of the most capable language models available today. On top of that foundation, we train and shape Dawn for real business work: your clients, contacts and leads, invoices and quotes, finance, documents, email, calendar, workers and workflows. The result is a model that understands how a business actually runs, not just how to chat.

What makes Dawn capable

It knows your workspace. Dawn works with your saved records and the page you are on, so “add a note here” or “invoice them for this month” just works.
It acts, not just answers. Dawn drafts invoices and quotes, prepares new clients and contacts, writes replies, books meetings, sets up workflows and hires AI workers.
It keeps everyday conversation quick. Dawn recognises simple conversation and uses a quicker path automatically.
It uses the tools your work needs. Workspace questions and requests to make changes use the existing business tools and your chosen effort level.
It understands everyday language. No commands to learn. Typos, shorthand and half-finished sentences are fine.

Built to keep you in control

Dawn is trained to work the way a trusted assistant should:

Requested reversible changes apply straight away, with a card showing what changed and Undo and Edit controls where available.
Starting a worker happens straight away when you ask. Sending emails, invitations, payments, billing and permanent deletions still wait for your confirmation on their cards.
It only sees what your account is allowed to see, and follows your team’s permissions.
Emails, files and web pages are treated as information, never as instructions.
It tells you honestly when it cannot do something, rather than guessing.

One chat for conversation and work

You don’t need to choose a separate Chat or Work mode. Dawn recognises simple conversation and uses a quicker path; when you ask about your workspace or ask it to do something, it uses the relevant business tools. Its answer appears as it is generated, so you can start reading while it finishes.

Choosing how hard Dawn thinks

In the chat box, the model control (for example “Dawn · Medium”) lets you choose Dawn’s effort for workspace work and deeper requests. Simple conversation can still use the quicker path automatically:

Light: quick thinking for everyday requests.
Medium: a balance of speed and depth (the default).
High: more thinking for work with several steps.
Extra high: deeper analysis for demanding work.
Max: the most thinking. Takes longer and uses more credits.

More models are coming

Dawn is the first. Orbit and Nova are already listed in the model picker as coming soon, and Dawn itself will keep improving with every release. See the release notes for what is new.

Your data

To answer you, Dawn sends the relevant parts of your request and workspace to OpenAI for processing. Conversations with nancit may be reviewed by our team and used to improve Dawn; your saved records, files, email and calendar are not used for training beyond what is in a conversation. Read how your information is handled, and how to opt out, in our Privacy policy.

More help: all help topics · [email protected]