Managed AI for Columbus Insurance Agencies
By nine on a Tuesday the expiration list has accounts at 90 days out, two of them need remarketing, and the loss runs on one are still pending with the carrier. Three certificate requests are in the queue from the same general contractor, each worded differently. A producer is retyping a coverage explanation he has already written for other contractors this month. Every one of those is a document job, and together they take the day.
Where it starts
The recurring work worth handing over first
Not the impressive demo. The jobs that come up every week and eat somebody’s afternoon, which is where the hours actually come back.
Side-by-side carrier proposal comparison
Put three or four carrier proposals in and get back a grid: limits, deductibles, sublimits and the exclusions that differ between them, with a plain-language note on where the cheaper quote is buying less coverage. The template compares wording as well as price, because a defined term can change what gets paid at claim time. The producer edits the note and takes it into the renewal meeting.
Renewal summaries a client will actually read
Point it at the expiring policy, the loss runs and the new quote. It produces a one-page summary in your agency’s renewal letter voice: what changed, what it costs, what the client needs to decide, and what happens if they do nothing. The template holds your structure, so every account manager sends the same shape of letter and a client who moves between producers sees the same service.
Certificate requests and holder language
COI requests come in worded differently for the same thing. The template reads the request and the contract language behind it, works out what the holder is actually asking for (additional insured, waiver of subrogation, primary and non-contributory, notice of cancellation), and drafts the response, including the harder response that says the policy does not do that and here is what it would take. Issuing still happens in your management system, by a person.
Policy documents into plain language
A producer has twenty minutes to explain a commercial package to an owner who has never read one. Ask the workspace what the policy says about water damage from a sprinkler leak and get the answer with the page and endorsement it came from. The citation is what makes the answer usable, because an answer with no page reference gets checked by hand anyway.
Claims correspondence and follow-up
Claims work is repetitive and easy to let go quiet for a week. Working from the file and the last email thread, the workspace drafts the status update to the insured, the follow-up to the adjuster and the internal note for the file, each in its own register. The account manager edits and sends. The drafting was the part eating the afternoon.
Onboarding a new commercial account
A new commercial account arrives as a pile: expiring dec pages, prior loss runs, a vehicle schedule, a payroll breakdown by class code and a contact list in whatever format the client’s controller uses. The workspace pulls it into your standard onboarding structure, lists what is missing, and drafts the request email for the gaps. The list of gaps is the part you would otherwise build by hand.
Our producers each have their own way of doing things. Does AI spread the good version or freeze what we have?
Agency knowledge sits in individual heads and individual inboxes. The producer who writes contractors knows which carriers will still look at a roofer with three losses, how to word the exclusion conversation, and what the client asks about in year two. That knowledge goes home at five and comes back in the morning. When she retires, it does not come back.
A shared workspace changes where the method lives. When she works out a sequence for summarizing a commercial package renewal, that sequence becomes a template other people run. The account manager who opens it next month gets her questions, her order of explanation, and her structure. Somebody improves it, and the improvement goes back to the whole agency.
Collaboration is one of the two pillars that compound the return, and it is the reason the outcome we sell is a cultural one. In a year, the best AI ideas in your agency should come from your people, not from us. Our job is to make a good method spreadable, report on what is spreading, and then stay out of the way.
How do we use this without creating an E&O problem?
Start with what it does not do. The workspace does not decide whether a claim is covered. That determination comes from the carrier and the policy, and any explanation that reaches a client goes out under a licensed person’s name. What the workspace does is draft the explanation, pull the policy language with the page and endorsement it came from, and hand it to a producer who reads it, corrects it, and signs it. The review step is part of that template, a page a producer works from rather than a habit somebody has to remember.
Then the documents. Applications, loss runs and claim files carry Social Security numbers, dates of birth, driver’s license numbers, EINs and payroll detail by class code. Those files stay inside your own workspace. ScaleSight AI runs on a SOC 2 Type II audited platform, and the audit belongs to the platform vendor rather than to ScaleSight AI as a legal entity. We state it that way because that is the wording that holds up when your E&O carrier or a commercial client’s procurement team asks.
Containing the risk has two pillars. Control is the budget ceiling, covered further down. Governance is this: each month you get a report of the kinds of work AI is being used for across the agency, grouped by category such as renewal summaries, certificate language, claims follow-up, marketing copy and spreadsheet cleanup. It covers work inside the workspace; anything still happening on a personal account outside it is invisible, which is the practical argument for moving it in. When somebody asks what AI is used for at your agency, you answer from that report.
The privacy boundary runs alongside all of it. Anything built on your own files, your own data or the way your agency in particular works stays inside your workspace and goes to nobody else. Your carrier appetite notes, your renewal letter voice, your commercial onboarding checklist: those stay yours.
We could just buy the agency ChatGPT accounts. What breaks?
Buy direct and every template you build lives inside one AI company’s product, which means it carries that company’s pricing and its release schedule. A price change or a retired model version turns into rebuild work on your side. Everything in your ScaleSight AI workspace sits one layer above the model, so we re-point it and the templates keep running. Your account managers open the same template they opened last week.
Routing is the other half. Answering a question about what one endorsement in a long commercial policy actually does is one kind of job. Drafting renewal reminder emails in bulk at low cost is another. Live web research on a carrier’s current appetite is another, and producing an image for a client newsletter is another again. The strongest model for each of those is a different one, so each job goes to whichever model handles it best. Your account manager opens a template called Summarize a commercial policy, and she never has to know which model ran it.
There is a plainer version of the answer too. Buying direct means paying per person across the whole agency with no allowance ceiling, no report of what the agency is actually using it for, and a template library that lives inside whichever individual’s account happened to build it.
How does the money work, and what stops a surprise invoice?
Managed AI is an ongoing managed service, priced per seat plus usage. A seat is a person: every producer, account manager, CSR and marketer who logs in has one. In an agency nearly every role works at a keyboard, so seat count tends to track headcount closely. Count it that way when you plan.
Each seat carries a monthly usage allowance, and you set a spend ceiling for the agency. Budget is allocated by team and department, so commercial lines, personal lines, benefits and marketing each hold their own allowance. It is never set person by person. An account manager in a heavy remarketing month does not have to ask anybody for permission. Her department’s allowance absorbs it, and a department that runs hot every month shows up in the monthly report before it shows up on an invoice.
Published budget ranges are monthly totals by seat count. Up to 10 seats, $250 to $500. Ten to 25 seats, $500 to $1,175. Twenty-five to 100 seats, $1,175 to $3,800. A $250 monthly minimum applies, and there is no published per-seat rate. Where an agency lands inside its band depends on how hard the workspace gets used, which is what the ceiling is for.
ScaleSight AI belongs to the Koine Cyber group, and its Columbus office at 1733 W Lane Ave is shared with TTS Cyber on the managed IT and cybersecurity side. When a carrier’s vendor questionnaire asks about your network controls instead of your AI use, that is the desk it goes to.
How does the workspace keep getting better after the first month?
The library is what keeps it moving. Anything that looks the same in every business gets built once and handed to every client: turning meeting notes into assigned action items, drafting a job posting, cleaning up a messy spreadsheet export, writing the follow-up nobody wants to write. An insurance agency and a machine shop in Central Ohio both need those, so neither one pays to have them built. Anything built on your own files, your book or your process stays in your workspace and is never shared out.
New templates, agents and workflows ship into your workspace every month out of that shared library. Insurance-shaped work goes in too, once it is generic enough to share safely. A template that turns three carrier proposals into a side-by-side comparison grid carries a method and no client data. Every agency that joins gets it, including the ones that join after it was written, so each client inherits everything built before them.
That is the Acceleration pillar. The library grows month after month, and every workspace picks up what gets added to it. The boundary holds while it grows. Generic methods travel, and your specifics stay where they are.
When you need something only your agency needs, that work is an AI Build Project: scoped work with a start date and an end date, to build one specific workflow, template or agent. An intake agent that reads a submission packet and fills your standard ACORD applications is a build project. It gets built, it ships into your workspace, and it stays there.
Questions
Asked by insurance agencies
Will AI end up giving our clients coverage advice?
Coverage answers stay with a licensed person at your agency, and the carrier and the policy decide what is covered. The workspace drafts the explanation and pulls the supporting policy language with a page and endorsement reference. A producer reviews it, corrects it, and puts their name on it before it reaches a client. That review step is part of the template a producer works from, not something left to individual habit.
Is client PII safe in there? Applications and loss runs are full of it.
Applications, loss runs and claim files stay inside your own agency workspace, and ScaleSight AI runs on a SOC 2 Type II audited platform. The audit belongs to the platform vendor rather than to ScaleSight AI as a legal entity, which is the accurate way to state it when your E&O carrier or a commercial client’s procurement team asks. Anything built on your documents or your process stays in your workspace and is never shared with another client.
What does this cost for an agency our size?
Managed AI is priced per seat plus usage, with published budget ranges as monthly totals by seat count: up to 10 seats, $250 to $500; 10 to 25 seats, $500 to $1,175; 25 to 100 seats, $1,175 to $3,800. A $250 monthly minimum applies. You set the spend ceiling, and budget is allocated by team and department, so commercial lines and personal lines each hold their own allowance.
How is this different from giving everyone a ChatGPT or Copilot account?
Work built directly on one AI company’s model has to be rebuilt when that company changes its pricing or retires the model, while work in a ScaleSight AI workspace gets re-pointed and keeps running. Jobs are also routed to whichever model handles them best, so reading a long policy, drafting renewal emails in bulk and running live web research on a carrier’s appetite do not all have to happen in one place. Your account managers open a template and never have to know which model ran it. Individual accounts also give you no allowance ceiling and no monthly report of what the agency is using AI for.
Do we need to know what we want built before we start?
You do not need a list of projects to start. Managed AI begins from the shared library, so a new workspace arrives with general-purpose templates already in it and new ones ship in every month. When you find something only your agency needs, that becomes an AI Build Project: scoped work with a start date and an end date to build that one workflow, template or agent, which then lives in your workspace and stays yours.
Other industries
We also work with
Tell us what your week actually looks like.
The readiness review starts by finding where AI is already being used in your company, what it is costing, and which recurring work is worth handing over first. You keep the findings either way.