4 updates
· 4 updates
AI that knows who is asking, and ranking by real skill experience
Hive and your connected AI assistant now follow each person's role, rank anyone you have by how long they have worked with a skill, and say plainly when something is out of reach.
AIPermissionsSourcingSettings
Hivemind runs your hiring end to end: pipelines, assessments, outreach, interviews and offers in one place, with an AI assistant you can ask about any of it. This release is mostly about that assistant.
It now follows the role of whoever is asking, so a teammate can only do through the AI what their role lets them do in the app. It ranks people by how long they have worked with a skill, across a pipeline, a saved list or your whole company, and links every person it names so you can check. When you ask for something it cannot do, it says so before anything else. And you can now make Hivemind your own colour.
This page walks through each feature: the problem it replaces, how it works, and where to find it in the app. Everything here is live in Hivemind today.
AI that knows who is asking
Roles decide what each person on your team can do in Hivemind: who creates pipelines, who sends outreach, who touches billing. An AI assistant connected to Hivemind should follow the same rules, whoever happens to be typing.
Now it does. The assistant checks the role of the person asking and only offers the actions that role allows, using the same rules as the app. Ask it for something outside your role and it tells you plainly that it is not available to you, instead of looking for another way round.
Roles are read live, so changing someone's role or deactivating them reaches the assistant within a minute. The same check covers Hive in the app: publishing a pipeline or an assessment that Hive drafted needs permission to create one.
Where to find it: Settings → Roles, where an admin decides what each role can do. Hive and any connected AI assistant follow it automatically.
Rank by experience with a skill
Rediscovering someone you already have starts with one question: who here has actually worked with this? Searching for the skill tells you who mentions it, not who has years of it.
Now your assistant can rank people by the time recorded against a skill on their resume. Point it at one pipeline, a saved list or every candidate in the company, narrow by status, stage, hired or the date they applied, and it puts the longest experience first.
It is honest about the gaps. People who list the skill without a duration are counted as unknown, never as zero, and it tells you how many have no parsed resume or no matching skill. Every candidate it names comes with a link to their profile, so you can check the claim in one click.
Where to find it: Any AI assistant connected to Hivemind. Try asking it to rank your candidates by experience with a skill you are hiring for.
Answers that say what they cannot do
An assistant that always has an answer is only useful until the day it confidently answers a different question from the one you asked.
Some requests are beyond what candidate search can do today: matching on meaning rather than words, judging fit against a job description, reading interview transcripts. Ask for one and the assistant now says so first, then offers the closest search it can run, clearly labelled as an approximation.
Results carry the same honesty. A search tells you it matched by word rather than meaning, and an answer bounded by dates shows today's date beside the window, so "last month" is never ambiguous.
Where to find it: Any AI assistant connected to Hivemind.
Pick your own accent colour
Hivemind came with a set of preset accent colours. If none of them was yours, you lived with the closest.
Now you can choose any colour, with a picker or a hex code, alongside the presets. Buttons, links, highlights and the sidebar follow it in light and dark mode, and text placed on it stays readable.
It is a personal setting: it changes how Hivemind looks for you, not for the rest of your team.
Where to find it: Settings → Appearance, under Accent color.
Try it yourself
Everything on this page is live in Hivemind today. Open the app and try it on a real pipeline, or book a walkthrough with the team.