Briard video demo transcript

Watch the Briard video demo. Updated October 12, 2026.

Recorded October 12, 2026 with fictional examples and footage of the current interface. Existing narration and footage are retained where the instructions still fit.

• The proof document is fictional. Private upload and scan results are simulated; no customer storage or scanning service is used.
• Team invitation delivery, email confirmation, account joining and member changes use fictional local responses; no real email or account is created.
• Discovery connections, check results and fleet examples are simulated; no customer account is connected.
• No real tool approval, payment, support message or device deployment is performed. Fleet features need separate enablement and approved setup; these examples do not establish real coverage or runtime enforcement.

00:00:00 — Start with one clear next step

Welcome to Briard! If you manage compliance or security, you probably spend a lot of time asking the same questions. Which AI tools are people using? Who is responsible? And what still needs a decision? Let's see how Briard brings that work together. You don't need a technical background to follow along. We're using a fictional company and sample information throughout this demonstration. Here on Home, Jane sees one clear next step: provide proof for an AI tool. The explanation tells her why it matters, and the button takes her straight to the work. Start here each day, check the place you're reviewing, and focus on the next useful action.

00:00:46 — Know which AI tools you have

First, let's open My AI tools. This is your shared list of the AI tools your organization has recorded. Choose a tool on the left to see its record. In this sample, we have tools used in a secure project, the main work environment, and clinical operations. That context matters. A tool used for everyday office work may need a different review from one used with sensitive information. The guided view keeps the next question easy to find. When you need more detail, open See full record. You can review the profile, answers, supporting proof, and things still to fix. Your team has one place to return to, instead of rebuilding the story from emails and spreadsheets.

00:01:33 — Add a tool in four short steps

What if a tool isn't on the list yet? Open More, then Add an AI tool. We'll add a fictional Contract Review Copilot. First, enter the name and provider. Next, explain its planned use in everyday language. Here, it helps draft contract summaries, and a person reviews the result. Then identify the kinds of information involved. You don't need to paste in a contract or share someone's private details. Finally, review the summary and create the record. That's the starting point: a named tool, a clear purpose, and a record your team can work on together. Adding the record doesn't automatically approve the tool. It gives you a practical place to complete the review.

00:02:21 — Set up a shared workspace

Invite your coworkers into the same workspace so your team can share the tool records, answers, and proof. Open Account and settings, then choose Manage under Members and roles. Jane is the workspace Owner. An Owner or Admin can invite a teammate using their work email and the right role. Choose Admin for someone responsible for workspace settings, Contributor for someone who adds work, or Viewer for someone who reads. Briard creates a private seven-day invitation. The invitation, email confirmation, and join in this recording are simulated; no real email or account is created. The recipient uses the invited email and signs in or creates their account. Their refreshed session opens the invited workspace. Back in the member list, the team shares one workspace, with each person's role visible. To change a joined member's access, select their new role and choose Save role. Removing a member requires confirmation. Owner access and your own access stay protected. Owners and Admins also need an authenticator code to protect workspace management. Keep a responsible administrator in place, and give each person the access their work requires.

00:03:39 — Support each answer with the right proof

A saved answer states what your team knows. Supporting material lets a reviewer check it. Here, Microsoft 365 Copilot is an internal productivity pilot. This fictional purpose and allowed-use brief supports that purpose answer. Open What can I use as proof. A purpose record explains the job, provider and setup. Data and safeguard records explain permitted information and required checks. Completed test or approval records show what was checked or decided. Include the tool, workplace, owner, current version and dates. From the question, choose Add a file instead. Check Proof supports, set when this material needs another review, and select the file. Use an approved file type within the displayed limit. Choose Add and check proof, then check the result. Scan passed means the file passed safety and integrity checks. It does not prove the answer is correct or approve the tool. This recording uses a simulated upload and scan. A separate review procedure supports whether a person must check important AI work. An approved HTTPS link is another option when a reviewer can access its source. Match each item to its own question. Use approved, non-regulated material only. Keep real customer records, regulated data, passwords, prompts and outputs out. If a fact or its support is uncertain, keep it open for the person who can confirm it.

00:05:10 — Find AI through approved work sources

You can also reduce some of the manual work. Open Shadow AI Discovery, then select the workplace you're reviewing. The source areas show the different places AI can appear, from work accounts to devices, cloud services, and employee reports. We'll choose Work accounts and Microsoft 365. Each check has its own access and coverage. Here, the fictional workspace already has an approved account. Review What access does this use before starting. The connection and result in this recording are simulated; no customer account or provider is contacted. In your workspace, wait for a completed check and review its scope, last-success time, and limitations. Saving a connection alone does not mean it was checked. Now review a finding. Look at its source, confidence, and what the evidence cannot establish. Choose Yes, this is ours, No, not ours, or I'm not sure. Uncertainty stays visible. After confirmation, Add to AI inventory creates a governed record. Your team still supplies the owner, purpose, and permitted information. A finding doesn't approve an AI tool or create a KAIDAN case. Keep unchecked sources visible as gaps, and add anything the approved sources cannot see. When separately enabled for your workspace, Device and model inventory shows reported computers and their apps or models. This expanded workflow is off by default and still needs approved sources and rollout validation. Installation or a model being available does not prove it was used. Check fleet coverage compares completed checks with an approved expected-device register. Here, the percentage stays unknown because that expected set hasn't been confirmed. Review the source's inventory date, missing checks, and identity gaps. Downloading a signed collector is not the same as installing it or completing a check. Your administrator approves that setup separately.

00:07:14 — See what needs your attention

Once the records are in place, What needs me helps you decide where to spend your time. Choose the part of the organization you're reviewing, then read the next action and its explanation. You don't have to work through every detail on every visit. Start with the item that needs a person, follow the button, and bring back the answer or evidence that's missing. This is especially helpful when you're new to the platform: the next step is connected to the record it belongs to. Briard helps organize the follow-through, while your team stays responsible for the decisions.

00:07:53 — Build a policy people can understand

Next, let's open My AI policy. This brings policy work into the same workspace as your AI records. At the top, you can see the draft's status and the next action. Review my answers takes you through straightforward questions, such as the organization name, who owns the policy, and what information employees must not put into an AI tool. The draft is organized into sections, so you can review one topic at a time. Here are the rules for acceptable and prohibited use, followed by a short employee guide. That makes the policy easier to discuss with the people who will actually use it. You can also set a review schedule. The aim is a policy your team can maintain, rather than a document that's written once and forgotten.

00:08:43 — Keep review, approval, and sharing clear

A draft is only the beginning. Approve and publish shows who needs to review the policy before it becomes the approved version. In a signed-in workspace, you choose eligible reviewers and the required roles. This sample doesn't send approval requests. Keeping the draft and published version separate lets your team work on changes while the current approved policy remains in place. Share and track is the next stage, where the approved policy can be distributed and acknowledgments recorded. You can see who has responded without treating a response as proof that someone has learned everything. For a compliance manager, the benefit is visibility: what is being edited, what needs approval, and which version people should be using.

00:09:32 — Prepare a package for a reviewer

When someone asks for an update, open Make a review package. Start with who it's for: a customer, advisor, insurer, or your leadership team. We'll choose Leadership, select the relevant AI tool, and make the package. The new executive package appears in the recent list, with a download available. You're bringing recorded information together for a specific review, rather than starting from an empty document each time. Check the package and any open items before sharing it outside your team. A package helps organize the evidence; it doesn't turn missing proof into approval. The same habit applies whenever a colleague asks about a tool: open the current record, review what's supported, and share the appropriate information.

00:10:22 — Know where incident review begins

One quick note about alerts. Home shows active AI incident alerts by review priority and category, with links into KAIDAN. P1 needs the fastest review. An alert means something needs checking; it isn't proof that harm occurred. Review a selected alert, or open all KAIDAN alerts through the protected handoff. A summary that cannot load is unavailable, not zero. And no open alerts does not mean every AI tool is monitored. Check the source coverage in KAIDAN. We'll cover the deeper investigation in its separate walkthrough.

00:11:01 — Find help as you go

You don't have to remember every step from this video. Learn Briard provides short guided lessons for common tasks, including adding a tool, answering questions, and preparing a package. Open a lesson when you're ready to do that work. The Help page also explains common questions and gives you a route to support. In your signed-in workspace, you can ask for help and follow your request. This sample won't send a message. If you're just getting started, keep your first session small: record one AI tool, answer the first question, and identify the person who can provide any missing proof. You can build from there as your team gets comfortable.

00:11:46 — Take your next step with Briard

That's a practical first day with Briard: see your AI tools, understand what needs attention, keep your policy current, and prepare information for the people who need to review it. You don't need to become a technical specialist to keep that work moving. Ready to begin? Visit briard dot A I and choose Start thirty days free, or sign in to your existing workspace. If you'd like help choosing your first steps, contact our team at support at briard dash A I dot app. Tell us what your team is trying to manage, and we'll help you get started. Start with one tool and one clear next action.