AI automation that knows when to ask.
We connect the tools you already pay for and let AI do the sorting, matching and drafting that eats your week. A person still approves anything that moves money or speaks for your business.
The work AI is good at.
Most of the hours a small business loses aren't hard work. They're the same small judgments, made hundreds of times: which email is urgent, which payment belongs to which invoice, which lead is worth a call. That is exactly the work today's AI models do well, as long as someone has decided what happens when the model isn't sure.
Sorting
Read every inbound email, form or alert and route it: urgent to a person, routine to a queue, noise to the archive.
Matching
Tie a payment to an invoice, a lead to a customer record, a meeting to the right client, and flag what doesn't fit.
Drafting
Prepare the reply, the call notes, the meeting brief or the follow-up, ready for someone to read and send.
Connecting
Move information between the tools you already pay for, so nobody retypes it from one screen into another.
Any model, picked for the job.
We aren't tied to one AI company. Each workflow gets the model that fits the job and the budget: a commercial model where quality matters most, an open-source model where cost or privacy matters more. Open-source models like Llama, Mistral, DeepSeek and Qwen can run on servers you control, so sensitive information never leaves your business.
Commercial models
- GPTOpenAI
- ClaudeAnthropic
- GeminiGoogle
Open-source models
- LlamaMeta
- MistralMistral AI
- DeepSeek
- QwenAlibaba
Where open models run
- OllamaRuns models on your servers
- Hugging FaceOpen model hub
Names and logos belong to their owners. Listing them doesn't mean they endorse us.
We run it on our own business first.
Before we offered this to anyone, we built it for ULTRA. These run every day across the 130 websites we look after and the business behind them. Each one does the reading and the matching, and each one stops and asks a person when the answer isn't clear.
Security alerts
Every security alert from the websites we host is read and labeled. Fixed rules decide what can never be dismissed, so the three that matter are not buried under two thousand that don't.
Payments to invoices
Bank deposits are matched to open invoices. Only matches the model is confident about are recorded; every other one waits for a person.
Leads to the CRM
A qualified lead becomes an account and a deal in the CRM. If the company or contact is missing, it writes nothing and asks, instead of guessing.
Meeting prep
Before each client meeting, a private brief is written from the CRM, the email history and the last call notes. It never sends anything to the client.
The rule that makes it safe.
Anything that only reads, reports or schedules your own time can run on its own. Anything that moves money, speaks for you or changes a record someone else relies on waits for a person. When a workflow can't tell which side of that line it's on, it refuses and asks.
That one rule is the difference between automation you can leave running and automation you have to watch. What AI agents can and can't be trusted with
How it works.
Three steps, priced the way our websites are: a fixed price to build, a monthly fee to keep it running.
1. Workflow review
$250credited toward the build
- How work moves through your business today
- Three to five jobs worth automating, each with a price
- Which approvals stay with a person
2. Build
From $1,500per workflow
- Built and tested on your real data
- Runs in accounts in your name
- Approval steps agreed up front
- Handover notes your team can read
3. Run
$500 to $1,000a month
- Every workflow monitored
- Fixes when a tool or setting changes
- AI models kept current
- Priced by how many workflows we run
Software and AI usage are billed to you at cost, in accounts in your name. What AI automation costs, and why
Custom tools, when a workflow needs a screen.
Some automations need somewhere for people to work: a place for clients to upload documents, a form that asks the right next question, a dashboard for the morning numbers. We build those too, as small web tools tied into the same systems.
Client portals
A private place for clients to upload documents, sign, pay or check status, connected to the systems behind it.
Intake forms
Smart forms that ask the right next question, check answers as they're entered, and land clean data in your CRM.
Internal dashboards
One screen showing the numbers you check every morning, pulled from the tools they live in.
Questions we hear.
How much does AI automation cost?
Most small business workflows cost $1,500 to $5,000 to build, then a monthly fee to run. Ours start with a $250 workflow review, credited toward the build, then from $1,500 per workflow and $500 to $1,000 a month to run. The software and AI usage are billed to you at cost, usually a few dollars to a few hundred a month.
Will the AI send emails or move money on its own?
Not unless you decide it should, and we'll usually advise against it. Our rule, which we follow in our own business: the AI sorts, matches and drafts; a person approves anything that moves money, speaks for you or changes a record someone else relies on.
Which tools do you work with?
The ones you already use: CRMs like Zoho and GoHighLevel, Google Workspace and Microsoft 365, accounting, forms, Slack and your website. We connect them rather than asking you to switch.
Who owns the automations?
You do. They run in accounts in your business's name, and if you ever leave we hand over everything, the same as we do with websites.
What happens when something breaks?
Tools change their settings and AI models get replaced, so automations need looking after. That's what the monthly fee covers: we monitor every workflow and fix it, usually before you notice.
Do you build mobile apps?
Ask us. Most businesses that ask for an app need a portal, a form or a dashboard, which we build regularly. For app-store apps we'll tell you plainly whether we're the right team.
Find the hours AI can give back.
Launch a new project or improve the one you have. Building websites since 2002.
