Ask your assistant. It does the actual work.
If your firm already uses an AI assistant, you can connect it to blankit. Then “pull up Humber Freight’s renewal and tell me what they should actually be paying” is a sentence you say, not an afternoon you spend. It isn’t a chatbot guessing at benefits — it runs the same actuarial engine, on your real book, and everything it does is written to the same audit log.
A conversation, not a copy-paste.
The usual way to use an AI on a renewal is to paste figures into a chat and paste the answer back out. That is where mistakes come from, and the result is nothing you would put in front of a carrier. This skips both halves.
Step 1
You ask, in plain words
In the assistant you already have open. No new app to learn, no screen to find, no particular phrasing to remember.
Step 2
It reaches into blankit
As your firm, and only your firm. It can look up your clients, your renewal history, your saved assumptions — the same records you see when you log in.
Step 3
The real engine runs
Not the assistant’s opinion of the numbers. Pooling stripped out, unreported claims added, experience trended and credibility-blended, the carrier held to their own target loss ratio — line by line.
Step 4
You get something you can send
The analysis saved to the client’s history, the carrier one-pager rendered, the quote started. Waiting for you in blankit when you next open it.
A renewal, start to finish, out loud.
Each of these is one thing you can ask for. Strung together they are the whole job — find the client, read the carrier’s letter, run the numbers, save it, produce the document you send back.
| “Which of our clients renew in the new year?” | Finds the client or clients you mean in your own book. |
| “What assumptions are we using for Humber Freight?” | Reads back their target loss ratios, unreported-claims factors, trend and pooling. |
| “The carrier just told me what they price dental to — record that.” | Writes the figure to the client’s assumptions so every future analysis uses it. |
| “Here’s the renewal letter. Read it.” | Turns the carrier’s PDF into structured figures, line by line — including a renewal split across several files. |
| “What should they actually be paying?” | Runs the full fair-premium analysis and shows you the variance against what the carrier asked for. |
| “Save that to their history.” | Files the analysis against the client, exactly as if you had run it on the dashboard. |
| “Make me the one-pager for the carrier.” | Renders the branded carrier-facing PDF, ready to attach to your challenge. |
| “Start a quote for a January 1 effective date.” | Opens a draft marketing case for the client with the lines you want quoted. |
You decide which of these your assistant is allowed to do. Anything that writes to your account is off until you switch it on.
Three steps. Nothing to install.
No software on anyone’s machine, no IT project, nothing to keep running. There is a guide inside blankit that walks you through it and stays with you while you do it.
Step 1
Create a key in blankit
Firm settings, one button. You tick what the assistant is allowed to do — reading only, or reading and writing.
Step 2
Paste two values into your assistant
blankit shows you a web address and a pair of credentials. Your assistant asks for exactly those. That is the whole handshake.
Step 3
Ask it something
“Which clients do we have?” is a good first test. If it names your clients, you are connected.
A key that reads is not a key that writes.
Your firm only
A key acts as your firm and reaches nothing outside it. No other firm’s clients, renewals or figures are visible to it, ever.
Writing is opt-in
Saving an analysis, changing assumptions and starting a quote are each switched off when you make a key. Leave them off and it can only ever look.
Written down
Every single thing the assistant does lands in your audit log, next to every other change your firm makes. Nothing happens quietly.
Off in one click
Revoke a key from firm settings and it stops working immediately. Access is short-lived by design and renews itself quietly, so cutting it off really does cut it off.
One thing worth being clear about: a key is a key. Anyone holding it can do what it allows, so treat it the way you would treat a password, give it only the permissions the job needs, and revoke it when whoever set it up moves on.
Same engine. Wherever you work.
These are not three products. It is one platform, one methodology and one audit trail, reachable from wherever you happen to be. Use one, or all three.
Rider, in the dashboard
The assistant built into blankit. It knows the screen you are on and the client you are looking at, and it walks you through the work step by step. Nothing to set up.
How the analysis worksblankit in Slack
For the team. Drop a carrier PDF in a channel and the analysis comes back in the thread where everyone can see it. Good when the work is shared.
blankit in SlackYour own AI
For the advisor who already lives in an assistant and wants blankit to be one more thing it can do. Most useful when you are moving between tools all day.
Book a walkthroughThe one paragraph with jargon in it.
blankit is an MCP server — the open standard AI assistants use to connect to outside systems. It speaks MCP over HTTP, with per-firm OAuth credentials and scopes that map one-to-one onto the permissions above. Any assistant that supports remote MCP servers can connect; discovery is published, so most find the token endpoint on their own. Full setup instructions, the endpoints and the tool list are inside the app under Firm settings.
Your documents and data stay hosted in Canada and are processed exactly as they would be on an upload — more on that on the Trust & security page.
Bring your own assistant.
Create a key, paste it in, ask it a question. If it names your clients back to you, the rest of the work is available too.