AI agents with full context of your Ashby data that can act on your behalf

Recruiting teams carry a lot of context: candidate sentiment across a pipeline, interview feedback on a long-open role, notes from a debrief that inform an offer. That context lives in Ashby, but pulling it together often means navigating across profiles, scrolling through feedback forms, and piecing it together across tabs.

Today, we're releasing Ashby Assistant, a chat-based agent that lets you ask questions in plain language and take action across Ashby from a single interface.

Ashby Assistant is available in open beta with opt-in access.

Ask Ashby anything across your hiring data

The assistant has access to your candidates, interview plans, jobs, applications, interviews, feedback forms, and interview transcripts. Your existing Ashby permissions apply, so it can only read and act on information you can already access.

For example, you can ask for a pipeline update on an open Product Engineer role that includes candidate sentiment, current stage, and concerns raised in interviews. Every answer includes citations back to the underlying data in Ashby, so you can trace each claim and click through to the source.

You can ask follow-up questions in the same conversation to dig deeper or refine a result.

Take action directly from the chat

Ashby Assistant can also take action on your behalf. From the same conversation, you can ask it to:

• Draft and send a candidate email
• Add a note to a candidate profile
• Advance a candidate to the next stage
• Consider a candidate for a job
• Create a new candidate
• Fill out a form

Every action prompts you for review before it runs. If you ask for an offer congratulations email that references a candidate's feedback highlights, you'll see the full draft with the option to refine, regenerate, or confirm before anything is sent.

Build Custom Agents for repeatable work

Great prompts take time to get right, and often need more context than you want to rewrite every time. Custom agents let you capture that work once and reuse it. To create a custom agent, click "New agent" in the Ashby Assistant interface, then build from scratch or start from a template.

Write plain-language instructions that define its goal, what inputs it should ask for, which sources it should use, how to format outputs, and what actions it's allowed to take. Once saved, run it by @mentioning it in any chat or selecting it from your agent list.

If you want every candidate debrief to check against your team's operating principles, you can build that format once in a custom agent and reuse it whenever you need it.

Share agents across your organization

Personal agents are for your own work, like capturing how you like to draft a particular type of email.

Shared agents are for workflows you want consistent across the team, like a debrief format every hiring manager uses.

Org admins can designate an agent as shared when saving it, or convert an existing personal agent into a shared agent. Org admins can also edit shared agents at any time. Visibility is role-based, so the right teams see the right tools.

Permissions and citations

All agent actions run as the user, so every read and write respects that user's existing job, team, and object-level permissions. An agent can't help someone see or change something they couldn't already access in Ashby. Elevated access users and org admins can create and use agents; limited access users cannot. Every response includes citations back to the underlying records in Ashby.

Release and pricing details

Ashby Assistant and custom agents are available today as an opt-in open beta for all customers. They are free during the beta period. Once generally available, they will consume your existing allotment of AI Credits.

Getting started

To turn it on, go to Admin and enable Ashby Assistant in your opt-in settings. Once enabled, you'll see "Ashby AI" in the top menu. Visit our help center for more details, or book a demo to see it in action.