What does AI governance mean here?
It means deciding, in writing, what an AI system is allowed to do, and who is responsible for its work. Before any system we build goes live, we answer four questions with you: what it can read, what it can do on its own, who checks its work, and what gets recorded. You approve the answers before anything is built.
The details are on the What we do page. Governance sets how the system works. It does not replace your firm's own professional, legal, or regulatory obligations.
How do you decide where AI would actually help?
We map the work before we choose any technology. During design we document each step, what it costs, and where the errors come from. AI goes only where it removes measurable time or risk and a person can still check the result.
Some steps need a process fix or a standard automation instead, and the design says which.
What does the 80/20 principle mean? Is the build only 20 percent of the price?
No. The 80/20 split describes where the thinking goes: most of what decides whether a project succeeds happens before anything is built. It is our consulting philosophy, not a split of hours, fees, or responsibilities, and actual effort varies by engagement.
The design phase is a flat $1,500. The build is priced separately, from the savings documented in design, with a $3,500 minimum, and the price is fixed in writing before the build begins.
Will it make things up?
It is set up to answer only from documents you gave it. It does not search the internet, and it is not asked to fill gaps from general knowledge. Every answer shows you the source, so checking it takes seconds. When it does not have the information, it is designed to say so rather than fill the gap.
No system is perfect, which is why a person approves everything before it leaves your office. That is not a disclaimer. It is the design.
Where does my data go? Is it training somebody's AI?
Your files stay yours. We do not use them to train public AI models, and the agreement says so. Access is controlled by role, the same way your file server is, so people see what they are supposed to see and nothing more.
We put the specifics in writing in the agreement, including where the data sits and who can reach it. If a vendor will not tell you that in writing, that tells you something.
For how this compares with a general AI chat tool, see why a custom build is safer.
Is this a setup to tell me to cut staff?
No. SageLogic is not in the business of helping firms eliminate people. We are in the business of putting tools in front of the people you already have so they can work more effectively inside the roles they already hold.
If your problem is too many people, we are the wrong call.
What we build takes the repetitive part off your team so their hours go to the work that requires a licensed, experienced human. The goal is a smaller backlog and better work from the team you already have. If a headcount reduction is your goal, say so in the first conversation and I will tell you that we are not the right firm.
What if it gets something wrong and it goes to a client?
It should not reach a client without a person reading it, and the systems we build do not send anything on their own. That is the point of the human approval step.
We also test for this before launch by deliberately trying to make it fail, and we train your staff specifically on how to recognize a wrong answer. The failure mode we design against is a confident wrong answer that nobody checked. We design against it rather than just warning about it.
What if AI is not the right answer?
Then the design says so. Sometimes a cleaner process, a simple automation, or better use of the software you already own solves the problem.
The redesigned workflow and the design are yours to keep and use whether or not you build with us.
What does it cost?
Design is a flat $1,500. It covers the Deconstruct and Redesign stages: mapping the current workflow, redesigning the process, documenting the savings, integration planning, and the system design. The redesigned workflow and the design are yours to keep and use whether or not you build with us. Your design fee is fully credited toward implementation when you move forward with SageLogic.
Implementation starts at $3,500. Custom AI solutions are priced according to project complexity, technical requirements, integrations, and anticipated business value, and the fee is fixed in writing before the build begins. If the documented savings do not support the $3,500 minimum, the design will say the build is likely not worth doing.
The implementation fee is charged once. Savings figures are estimates produced by the analysis, not a guarantee of results. No charge per person and no charge per use. After launch, we recommend at least three months of monitoring and support so we can adjust the system and correct any issues that surface. Monitoring and ongoing support are tailored to each firm, so the terms vary and are agreed before the build begins.
The number depends on how many workflows we are building, how much of your existing material we can use as-is, what it has to connect to, and how much training you need. We give you the figure after we have looked at your actual work, not before.
If it is out of range for your firm, we will tell you in the first conversation rather than after you have spent time on it.
What happens when the project ends?
Once the build is paid for, the system, the documentation, and the training materials stay with your firm for its own use, and your team is trained to operate it day to day.
We recommend at least three months of monitoring after launch, with SageLogic available to adjust the system and correct issues as your team uses it. Beyond that, ongoing support can cover monitoring, troubleshooting, maintenance, and improvements as your work and your software change. The length and scope differ for every firm and are agreed with you in advance.
Do I need an IT department?
No. Most firms we would work with do not have one. We build around the tools you already use and we train the people who will actually operate it. If you have an IT provider, we will work with them.
Do we need to replace our software?
Replacing your software is not part of our approach. We start from the software you already use and connect to it where it allows. Which connections are possible is confirmed during design, before you commit to a build. If a connection is not possible, the design says so and names the alternative.
Can our IT team or employees build this themselves?
Internal implementation can make sense when a company has the necessary AI expertise, process-design capability, governance, available development capacity, and someone responsible for maintaining the system after launch.
The challenge is that AI implementation involves more than selecting software or creating prompts. It requires understanding how the work actually happens, determining where AI should and should not be used, establishing controls, testing the system against real work, training users, and maintaining it over time.
SageLogic does not replace your employees or IT team. We work with them. Their knowledge of your business, systems, and day-to-day operations is critical. SageLogic provides the specialized workflow design, AI implementation, governance, testing, and deployment expertise needed to turn that knowledge into a controlled operational system.
How long before I see anything?
Two to four weeks to map your work, redesign the process, and give you a plan. Four to eight weeks to build, depending on scope. Your written plan states the actual timeline.
How will we know whether it works?
During design we document what the work costs today: hours per cycle, cycles per year, and what errors and delays cost, from your own records. You sign off on that figure before the build fee is set.
That gives you a written baseline to compare against once the system is running. Before launch, the system is tested against your real work, and during the monitoring period after launch we review how it is performing with your team.
What if my team hates it?
That is the most common way these projects fail, and it usually happens because a tool got dropped on people who were not consulted. We start by sitting with the people doing the job, which is partly to design the right thing and partly so it is not a surprise when it arrives.
If your team will not use it, it does not matter how good it is. We would rather find that out in week one.
Why should I believe you? You are a new firm.
SageLogic is new. I am not.
I bring fifteen years of environmental consulting experience, so I know operational headaches firsthand. I spent more than a decade managing projects, writing reports, and working through regulatory demands. I have used AI tools since they became broadly available, and I have built my working practices around security, governance, and data integrity. I do not build theoretical solutions. I build systems to solve problems I have lived through.
Here's what I would ask a firm in my position, and what you should ask me: make me explain exactly what I am building before you pay for the build. Make me tell you what it will not do. Make me put the fixed price in writing. Make me show you the rules before you approve them. Every one of those is built into how we work, and you can hold me to all of them.
My industry is not on your list. Does that matter?
Less than you would think. The pattern is nearly always the same: repetitive documents, knowledge stuck in a few heads, and a mistake that costs real money. If that describes your firm, the industry matters mostly for the vocabulary and the rules, both of which we learn in the first stage.
If your work is highly specialized and regulated in a way we cannot learn quickly enough, we will tell you.
Which AI is this? Do I have to use ChatGPT?
We pick the tool to fit the job and we tell you what we are using and why. You are not buying a subscription to a brand name. You are buying a customized system built around your work, and the underlying tool can be changed later without rebuilding everything.
The system often runs on the same leading AI models you may have tried in a chat tool. The difference is what is built around the model: approved sources, access by role, a named approver, and a record of everything it does.