Straight answers about running AI in a small company.

Managed AI is a managed service where an outside team runs a company’s AI workspace for it. ScaleSight AI does that for US companies of roughly 20 to 500 people where nobody owns AI as a job title: one secure workspace, allowances set by team under a ceiling the client sets, a report on what work is going to AI, and new workflows every month.

You need one approver at the client. You do not need an expert. Somebody signs off on the ceiling, approves the allowance changes ScaleSight AI recommends, and is the named contact for questions. ScaleSight AI does the configuration, the reporting, the building, and the training. Nobody at the client has to evaluate models, learn the tooling, or give up a quarter to run a pilot.

ScaleSight AI puts the hours in the scope before a client signs: who is needed from your side, for what, and for how long. In general a client provides a list of people and roles, one administrator contact, decisions on the ceiling and the allowances, and access decisions for any connected systems. ScaleSight AI does the configuration and runs the training. If a client cannot name an administrator, say so early, because that one decision cannot be outsourced.

Copilot, ChatGPT, and Claude are products a company licenses and then owns the job of driving. ScaleSight AI is a team that runs the program: allowances set by team and department, a report on what kinds of work AI is being used for, review points for anything with money on the other end, and new workflows built and delivered every month. ScaleSight AI is also not tied to any one of them, so a price rise or a retired model does not become your rebuild, and each job goes to whichever model does it best. ScaleSight AI works alongside an existing IT provider rather than replacing them. The two questions worth asking your provider are who sets the team budgets, and who builds the workflows after launch.

A blank chat box gives an employee no starting point and no reason to break an existing habit, so the licenses sit unused. ScaleSight AI delivers ready-made templates, agents, and workflows for jobs people already do every week, keeps delivering them every month rather than only at go-live, and runs short recorded training at each handover.

Traction is ScaleSight AI’s word for AI that actually reaches the work. ScaleSight AI treats it as four things at once: control of the budget, by team and in total; collaboration in one shared workspace instead of separate personal accounts; governance, meaning a record of what AI is used for and who checks the output; and acceleration, meaning new workflows arriving every month after launch. It is unrelated to the EOS book of the same name.

ScaleSight AI prices managed AI per seat plus usage. Every seat carries a monthly usage allowance, and the client sets a ceiling on total spend. It is not flat and it is not unlimited. ScaleSight AI publishes monthly budget ranges by seat count rather than a per-seat rate, because the accounts are the cheap part of this: what a client pays for is the allowances being set and watched, the record of what AI is being used for, and a team building new work every month. A budget range is something a company can plan against. A per-seat rate only invites a comparison against buying accounts direct, which is a different purchase. See the ranges in the next answer. Ask for a written quote and you get the seat rate, the allowance, and the ceiling on one page. AI Build Projects are quoted separately.

ScaleSight AI engagements start at $250 a month, and above that the budget follows the seat count. Plan on these monthly ranges: up to 10 seats, $250 to $500; 10 to 25 seats, $500 to $1,175; 25 to 100 seats, $1,175 to $3,800. The $250 minimum is where a team of five or fewer lands. Seats above the first ten are budgeted at lower rates, so the budget grows more slowly than the team does. These are seats rather than headcount, because a client seats the people whose work happens on a screen: a 60-person manufacturer commonly seats 20 to 30 and plans against the middle range rather than the top one. Where a company lands inside its range depends on how large the allowances need to be and how much setup the first month carries. These are planning figures rather than a quote, and AI Build Projects are budgeted separately.

Three things drive the cost of the ScaleSight AI managed service: how many people you seat, how large each team’s monthly usage allowance is, and how much of the work runs on cheaper models. Seats are the predictable part. Usage is the part that moves, and it moves inside the ceiling you set. ScaleSight AI routes routine work to lower-cost models, which is one of the levers that keeps a month’s work inside its allowance.

Usage is metered by how much text the models process, so a short draft or a summary uses a small part of a monthly allowance and a long document analysis uses more. ScaleSight AI sizes each team’s first allowance from the work it does, shows usage as a share of that allowance in real time, and adjusts it after the first full month of real data. A client does not have to guess the right number on day one.

Yes. Seats are per person, but budget is not. ScaleSight AI sets a monthly usage allowance for each team and department at setup, so sales writing proposals all day carries a larger allowance than finance closing the month. ScaleSight AI reviews those allowances against real use every month and brings the client the changes worth making. Every allowance sits under one overall ceiling the client sets for total AI spend, so allocation is a management decision rather than a technical one.

ScaleSight AI flags which team is close to its allowance while the month is still running and brings the client one recommendation: raise that team, move unused allowance across from another team, or leave the limit where it is. The client decides. Total spend still cannot pass the ceiling the client set, so the conversation happens in advance rather than on an invoice.

ScaleSight AI puts three things in place: one company account instead of scattered personal subscriptions, a usage allowance on every seat, and a total ceiling the company sets. ScaleSight AI also routes routine work to lower-cost models so each allowance goes further. Spend still varies with use. It varies inside a limit the client chose, and it is visible by team and by department while the month is still running.

No. ScaleSight AI seats are per person and cost money per person, so a client seats the people whose work happens on a screen. In a manufacturer or a field services business that is usually the office: sales, estimating, purchasing, finance, HR, and service coordination. ScaleSight AI recommends going a little wider than the people who asked loudest, because useful patterns often come from people who would never have asked, and then reports monthly which seats went unused so a client can drop them.

ScaleSight AI builds that list during setup: the personal ChatGPT, Claude, and similar subscriptions people are expensing, plus any team plans sitting on a department card. What a client saves depends on how many of those exist. That list usually does not exist until somebody goes looking for it, so if you already have one you are ahead of the exercise. The client decides what to cancel, and the total lands against the new bill.

ScaleSight AI runs on a SOC 2 Type II audited platform, hosted in AWS US regions, isolating each customer tenant, does not train on customer data, and runs outbound calls to model providers under near zero retention settings. ScaleSight AI administers access centrally, so AI accounts are created and removed alongside the rest of employee onboarding and offboarding. ScaleSight AI sends the written security terms before a client signs, and states plainly which certifications are held by the underlying platform and which by ScaleSight AI as a company. ScaleSight AI is part of the Koine Cyber group.

No. Customer data in the workspace ScaleSight AI operates is not used to train AI models, and outbound calls to model providers run under near zero retention settings. ScaleSight AI puts its current data terms in writing before a client signs, so the answer holds up when a client contract or a cyber insurance renewal asks for it.

Shadow AI is employees using AI tools the company never approved, usually free or personal accounts paid for on a personal card. It matters because company and client material ends up in accounts the business cannot see, audit, or shut off when that person leaves. Banning the tools moves the activity to phones and home laptops, where there is no record at all. ScaleSight AI runs a sanctioned workspace that is easier to use than the free tool, and reports what the company is actually using AI for.

ScaleSight AI moves the activity into a workspace the company controls, then writes the policy around what people are actually doing. Blocking AI outright pushes the same behavior onto personal phones, where nothing is logged. Once the work is in one place, ScaleSight AI reports what is being handled, sets rules for who can reach which sources, and defines where a person signs off.

It will, so ScaleSight AI builds review points into anything with money or a customer on the other end. Quotes, estimates, pricing, client-facing documents, and anything with a number in it get a named person who checks before it goes out. ScaleSight AI writes down, per workflow, what the AI does, what the person checks, and what the AI is not allowed to do unsupervised. AI that never needs checking does not exist today, and any provider implying otherwise is selling something else.

Ask before you assume it does. ScaleSight AI hosts in AWS US regions with tenant isolation, which is not the same thing as an ITAR or CMMC compliant environment, and ScaleSight AI will not claim that it is. If you hold customer prints or drawings under NDA, tell ScaleSight AI which material is restricted and it will be kept out of scope. If your compliance boundary is stricter than what ScaleSight AI meets, ScaleSight AI will say so.

ScaleSight AI reports the kinds of work going through the workspace it runs, by team and by department, and reviews that report with the client. Personal and free accounts report nothing back to a business, so the reporting only covers work that has moved into the workspace, which is the practical argument for moving it in. The report shows categories of work and usage rather than a transcript of every chat.

Colleagues see what a person puts in a shared space: shared prompts, shared projects, and published templates. Work a person keeps private stays private to them. Administrators see usage and cost by team, and the categories of work by department, which is what a budget and a policy require. ScaleSight AI writes this down in plain language during setup, so a client can explain it to their team without guessing.

ScaleSight AI keeps shared prompts, templates, and workflows in a workspace the company owns, so a resignation removes a seat and leaves the working method behind. A personal ChatGPT account works the other way: the account belongs to the individual, and the useful prompts leave when they do. Anything a person kept private is still theirs, which is why ScaleSight AI’s setup includes showing people how to publish the work worth keeping.

ScaleSight AI runs one workspace where prompts, projects, and templates are visible to colleagues and reusable by them, so one person’s method for a proposal, a quote, or a call summary becomes something the next person opens. ScaleSight AI’s team also publishes the versions of common jobs that come up across its clients, so nobody on your team is starting from a blank box.

ScaleSight AI’s managed team does, as part of the managed service. Every month it delivers new templates, agents, and workflows into the client’s workspace. Some are built from the client’s own usage report, and some are jobs that are the same in every business, built once and given to every client. ScaleSight AI also re-points and re-tests existing workflows as models and prices change, so what was built in January still holds up later in the year.

Included in the ScaleSight AI managed service: the workspace, allowances and reporting, monthly templates, agents, and workflows that run inside the workspace on material you upload, plus maintenance of what has been built. A separate AI Build Project: anything that connects to another system such as an ERP, CRM, or ticketing tool, anything that writes back into one, and anything that needs custom testing against your own data. ScaleSight AI names which one it is in writing before work starts.

Yes. AI Build Projects is ScaleSight AI’s project work, and it is available on its own or alongside the Managed AI service. ScaleSight AI scopes the work in writing, builds it, tests it against your real past cases, and hands it over documented with the users trained. It has a start date and an end date. Clients who want it maintained afterwards put it on the managed service; clients who do not, own it as it stands.

No. Anything built on your documents, your data, or the specific way your company does a job stays inside your workspace and is never shared with another ScaleSight AI client. What is shared is the generic half: drafting a proposal, summarizing a meeting, writing a job posting. Before anything is built, ScaleSight AI writes down which side of that line the work falls on. If a workflow needs your data to work at all, it is yours alone.

Different work goes to different models, and nobody on your team has to manage that. No single AI company leads at everything: one is ahead on generating an image, another on live web research with current sources, another on reasoning across a long contract, and for high-volume drafting the honest answer is often whichever model is cheapest that month. A company that buys accounts from one AI vendor gets that vendor for all of it, including the tasks it is worst at, or starts paying for a second and third subscription to fill the gaps. ScaleSight AI runs one workspace that reaches multiple providers and routes each job to the model that does it well, which is also how routine work is kept on cheaper models so an allowance lasts the month.

No. Everything ScaleSight AI builds is model agnostic, so templates, agents, and workflows are not tied to a single AI company. The scenario this protects against is specific: an AI company raises its prices sharply, or retires the model your automation was built on, and anyone who built directly on it rebuilds the work. ScaleSight AI re-points the workflow to another model, re-tests it against the cases it was accepted on, and it keeps running. Being able to move is also what keeps any one provider honest on price, and it is why routine work can sit on cheaper models without anyone changing how they work.

The prompts, templates, and documents your people create in the ScaleSight AI workspace belong to your company. Ask for the contract term, the notice period, and the export terms in writing before you sign, and ScaleSight AI will send them. An exit answer belongs in an agreement rather than in marketing copy, so read it there and hold us to it.

The test ScaleSight AI holds itself to is who is finding the work. After a year, ask who found your last three AI use cases: if the answer is ScaleSight AI, you bought some automations, and if the answer is your own people, you have something that keeps going without buying anything else. The observable signs are ordinary rather than dramatic. The person who closes your books sets something up and mentions it after it already works. The question in a meeting stops being whether AI could do something and becomes whether anyone already has. Somebody asks ScaleSight AI to check what they built rather than to build it for them. ScaleSight AI cannot promise a culture, because nobody can sell one, but it can run the conditions that make one possible and report honestly at each quarterly review on whether it is actually happening.

ScaleSight AI is not a fit for a company that already has an AI lead and people who build, because it would be paying for something it already owns. It is not a fit for a company that wants one custom system and no ongoing service, which is a project rather than a managed service. It is not a fit for a company unwilling to fund usage, since usage is metered and priced. It is not a fit for a company too small to justify the $250 a month minimum, which is roughly a team of five or fewer at a desk. And it is not a fit where the compliance boundary is stricter than a SOC 2, US-hosted, tenant-isolated platform.

ScaleSight AI does not publish client names, logos, or case studies. What a buyer can check instead is method: the pricing model stated before the contract, ScaleSight AI’s written security terms, a project scope that names the hours and the review points before anything is built, and a walkthrough on a call using one of your own recurring jobs. Judge the parts that are checkable now rather than a story you cannot verify.

ScaleSight AI is part of the Koine Cyber group. The group also holds managed IT and cybersecurity providers and a specialist recruiting firm, so when a need runs past what ScaleSight AI does there is a coordinated answer rather than a referral into the dark. Each engagement is contracted with the company delivering it.