Ask Brooklyn AI

Artificial intelligence is the hottest topic in software right now, and also the one that makes risk teams nervous, because using it usually means sending sensitive supplier and contract data somewhere else. Brooklyn takes a different route. There is no central AI module bolted on the side. Instead, AI is woven into the workflows and processes you already use, to speed people up and surface insight where it is needed.

How does Brooklyn actually use AI?

AI shows up across the platform in a few distinct ways:

• Insight automation, reading data from files or records and pushing insight up to the user
• Virtual colleagues that automate knowledge and answer questions
• A Chat to My series, where you can chat to a document or a dataset to pull out exactly what you need
• My AI Prompt Designer, so you can edit any of these prompts and tailor them to your business, including changing the language of responses
• My Secure AI Assistant, a large language model embedded in the platform and tuned to act like a supplier relationship and contract manager

Security is the point that matters most. The assistant runs inside the ecosystem Brooklyn is deployed in, through a service account, and none of your data leaves your account or its jurisdiction, so GDPR and data residency are respected.

Can it take work off a busy team mailbox?

Larger supplier and contract management teams often run a shared mailbox that fields hundreds of simple questions a month: when is the next supplier meeting, when does a contract end, who looks after this agreement. All of that already lives in Brooklyn. Virtual agents answer those questions for non specialist colleagues directly, without tying up the team. Brooklyn aims to remove between 50 and 75 per cent of those questions within the first six months, moving towards 90 per cent over a 24 month period.

Can it draft and capture information for you?

The assistant is context aware of the page you are on, and can produce shareable output. It can draft a set of slides to present a new contract, complete with a milestone table. Working from a tracked risk, it can write the event up in your executive risk committee format in around fifteen seconds, ready to copy into Word.

It also removes retyping. Turn on a field such as contract start date and Brooklyn generates what to look for, scans the contract, fills the metadata and asks you only to confirm it. On large free text fields, it can draft content like a contingency plan from everything on the page, so you edit rather than start from a blank box.

How accurate is it, and who stays in control?

Brooklyn is candid about performance. Across responses, roughly five per cent come back incorrect, and those almost always trace back to missing data or a prompt that needs tuning. Around ten per cent are cases where a human would add nuance the AI missed. About 80 per cent are acceptable as they are, and around five per cent are better than a subject matter expert would manage, a share that grows as the models improve.

The philosophy is that AI is strongest at scale, speed and consistency, so it takes the summarising, finding and drafting, and leaves the complex specialist judgement to people. Admin users get a no code console to edit the prompts behind any of this, so you keep ultimate control over what the AI returns.