How we work
We start with the work that costs you the most, prove an agent on your real tasks with your people in control, and expand only when the numbers say so.
Free 30-minute call. You get a written plan, whether or not you work with us.
Customer Agent
Website chat and text · version 1
Each phase ends with something in writing and a decision you make, so you always know what’s next and why.
We learn how your business runs, where the hours go and which tools you use.
A 30-minute call, free
You getA written plan in about 3 business days: your top opportunities, ranked by value
We map the workflow with the people who do it, and agree the agent’s rules and limits.
Typically 1 to 2 weeks
You getA one-page agent spec: its job, its access, its limits and the success measure
We connect your tools, build the agent and test it on your real past work.
Typically 1 to 2 weeks
You getA test report: the agent’s work next to your team’s, on real past cases
The agent goes live with your team approving its work. We review it daily, then weekly.
Typically 2 to 4 weeks
You getA pilot report against the success measure, with a go, adjust or stop call
We monitor, tune and maintain the agent, and expand what it does only when you approve.
Ongoing, reviewed monthly
You getA monthly report in hours and dollars, and a quarterly roadmap review
Time to value
A first agent typically goes live 2 to 4 weeks after kickoff, with your team approving its work.
Inquiries that arrive after hours get answered the day the agent goes live.
Savings grow as routine work goes through without review. Practice-management, field-service and ERP projects take longer.
Timelines depend on scope and access. You get a dated plan before work starts.
Every month
Once an agent is running, you get one page a month: what changed since launch, what the agent got wrong and who caught it, and what we’d change next. It takes about 30 minutes to review.
It starts with the free assessment. See a sample assessment summary.
Gablecrest Roofing & Exteriors
Fictional business · Customer Agent on web chat and text · month 3 since launch
Against the baseline you signed off
| Measure | Before launch | This month |
|---|---|---|
| First reply to a new lead | Over an hour for 4 in 10 | Under a minute, every lead |
| Inspections booked | 200 a month | 208 a month |
| Office time on calls and texts | All by hand | 30 hours back |
| Value this month | – | $7,975 |
8 more inspections × 25% close = 2 jobs × $11,000 × 34% = $7,480 · 30 h × $33 × 50% = $495
Caught and fixed
Proposed for next month
FIG. 2 Illustrative monthly report. Gablecrest Roofing & Exteriors is a fictional 85-person roofing company; outcome figures are the base case of its worked example, by month.
Not a quote or a guarantee. Estimates exclude the cost of the engagement.
Every engagement is scoped and priced in writing after the free assessment. There’s no long-term contract to start.
Prove it on one job.
What’s included
No obligation to continue. If the pilot misses its measure, we say so plainly and you decide what’s next.
We run it. You get the results.
What’s included
Issues reported to you the same business day. Support hours and response times are written into your agreement.
Your AI team, across the business.
What’s included
Starts with a pilot on one job, so you see results before anything expands.
How pricing works
Scope, price and terms are set in writing before any work starts. New work outside the scope is quoted the same way.
Third-party costs, like phone minutes or a software seat, are listed in your proposal at what we pay.
Every proposal puts our fee next to the modeled savings, so you can weigh the payback before you commit.
AI Consulting can also be booked as a standalone project. Want a yardstick before we talk? The savings calculator models five levers, from team time to pricing and vendors, and shows a break-even budget: the most the agents could cost each month and still pay off.
The practices we follow on every engagement, in plain words. Send your security questionnaire and we’ll answer it in writing.
Each agent gets its own credentials and the minimum access its job requires (least privilege), read-only wherever possible. We review access regularly and remove it when it’s no longer needed.
We build on business-grade AI services whose terms exclude training on your data. Before anything connects, you get a list of every outside service that will touch it.
Only the people building and running your agents can see your data, under a signed confidentiality agreement. We review agent work inside your systems, not in copies.
We connect the minimum a workflow requires, avoid copying data out of your systems and don’t keep what we don’t need.
Payment details, health information and personal records are masked, excluded or kept in the systems built for them. We agree the rules with whoever handles compliance for you before anything connects.
Passwords and keys are kept in a secure vault and never shared over email or chat.
Every agent action is logged: what happened, when and why. The audit log is yours to read at any time.
If something goes wrong, we pause the affected agent, tell you the same business day, fix the cause and give you a written account of what changed.
When an engagement ends, we remove our access, hand over the documentation and delete working copies of your data on request.
Every agent starts with people approving its work. It takes on more only when its track record supports it, and only when you say so.
The agent prepares the work. A person sends, books or changes it.
A person acts
The agent acts once a person approves each item.
Your OK on each item
The agent handles routine work on its own. People review a sample and every exception.
You review samples
The agent handles routine work end to end, inside hard limits. Anything outside them goes to a person.
Exceptions come to you
You can move any agent down a level, or pause it, at any time.
Choose what needs a yes: by category, dollar amount, customer type or anything else.
Spending caps, send limits, discount and refund ceilings, and hours the agent can’t exceed.
Complaints, legal or medical questions, safety issues, key customers and large orders always go to a person.
When an agent passes something to a person, it includes the full context and a summary.
We review samples of agent work weekly at first, then monthly once things are stable, and track accuracy over time.
Pause any agent at any time, and your team picks up where it left off.
Done for you doesn’t mean done without you. Here’s what makes an engagement go well.
Someone who can approve rules and make calls, so the work never waits on a committee.
Admin or delegated access to the systems in scope. We send a checklist and set up least-privilege accounts with you.
FAQs, price lists, policies, templates and examples of work you’re proud of. Rough is fine.
Today’s numbers, like response times, hours spent or cost per lead, so we can measure the change. We’ll help you pull them.
Fast reviews of the agent’s work in the first two weeks make the biggest difference to quality.
People do better with change they understand. We’ll help you explain what’s changing and why.
Something else? Bring it to your free assessment call.
We don’t publish prices because scope drives them, but here’s how pricing works. A pilot is a fixed price for a fixed scope, agreed in writing before any work starts. If you continue, managed agents are a flat monthly fee that covers monitoring, upkeep, changes within scope and AI usage at the volumes we agreed. Third-party costs, like phone minutes or a software seat, are listed in the proposal at cost.
There’s no long-term contract to start, and every proposal puts our fee next to the modeled savings, so you see the payback before you sign. Want a yardstick now? The savings calculator estimates what your team’s time is worth each month: the most an AI program could cost and still pay for itself.
A first agent is typically live 2 to 4 weeks after kickoff. For the first few weeks, your team approves its work in short reviews built into the day, and that’s how it learns your standards. Some benefits, like answering after-hours inquiries, start on day one. Time savings grow as you let routine work through without review, and the full benefit usually takes 2 to 3 months. Projects that connect to practice-management, field-service or ERP systems take longer, and you’ll get a dated plan before work starts.
It will happen occasionally, so we plan for it. Agents answer only from information you approve, anything unclear goes to a person, and early on your team approves the work before it goes out, which is where most mistakes get caught.
If one slips through, we tell you the same business day, correct it with the customer or in the record, pause that kind of task if needed and fix the cause. You’ll see what happened and what changed in the next report. Your agreement spells out who is responsible for what, and every action is logged, so there’s never a question about what the agent did or why.
Not yet, and we won’t imply otherwise. Instead, we answer your security questionnaire in writing, give you a list of every outside service that will touch your data, and work within your IT policies, in accounts you own wherever possible. See how we handle access and data.
The people who design and build your agents. You get one named lead who knows your business and answers your questions directly. There’s no handoff from a sales team to a support queue.
We tell you plainly, explain why and recommend whether to adjust the approach, try a different job or stop. You decide, and you’re under no obligation to continue.
Your data and accounts stay yours, and so does the business knowledge we organize for your agents, like FAQs, price lists and rules. We set things up in accounts you own wherever possible and document how everything works. Ownership terms are spelled out in your agreement.
Your agents run in accounts you own wherever possible, and everything is documented as we go. If we stop working together for any reason, we hand over the documentation and credentials, remove our access and follow the exit steps written into your agreement.
Yes. We coordinate access with your IT provider and keep them informed about what we connect. The Marketing Director works alongside your agency, sharing findings and test results.
Yes. Short, practical sessions on reviewing agent work, changing rules and handling handoffs, plus a one-page guide for each agent.
Free AI assessment
Thirty minutes, no cost. You get a written summary of where AI would pay off first, and what it would take.
What you leave with
By estimated value and effort, specific to how your business runs.
A modeled estimate, with every assumption shown.
The first job to hand an agent, and what it would take.
Yours to keep, whether or not we work together.