Your embedded fractional AI partner
Business process automation and AI training for a small business. A founder stays on after the build goes live.
Preferred AI Partners builds the systems that take repetitive work off your team: invoices, quotes, order exceptions, lead routing and CRM updates. The 30-minute AI assessment names where to start. After that, you choose how many fixes we build at once, and we scope the engagement to your ranked list.
Use cases, a ranked list of fixes, then working systems
Three steps, each with a timing and something you keep at the end. You can stop after any one of them.
- 01Thirty minutes
The AI audit call
One of our founders asks where the hours go in your week. You leave with your top AI use cases, or a plain no, and a recap email the same day.
- 02Weeks 1 to 2
Find
Two weeks with your team, then the written operations map. It ranks the fixes worth making, with the effort each one takes and what it should return.
- 03From week 3
The build
We start at the top of the map and connect what you already own. We train your team to run what we built, and we stay on to keep it working.
What every engagement includes
- One of our founders on every call
- The systems you already own connected first
- Regular reports on what each build is saving you
- Exceptions routed to a person, not to everyone
- Written procedures and a handover session
- Training on what we built and on the AI tools
What we take on in your business
An engagement can combine two or three of these. The AI audit call names which one your business needs first.
AI strategy and roadmap
Deciding what is worth doing, before anything gets built
- Where AI would pay for itself in your operation, and where it would not
- A ranked shortlist with the effort and payback of each
- A first project picked because it is the one most likely to work
Business process improvement
Fixing how the work runs before automating any of it
- Where your team’s time goes each week, process by process
- Steps removed, merged, or put in a different order
- Sometimes a redesign and no software at all. We will say so
Workflow and process automation
The routine part runs itself and a person handles the rest
- Routine cases handled start to finish with nobody touching them
- Exception routing, so only the odd case reaches a desk
- Documentation, so your team can change it without us
Lead automation, routing, and nurturing
Every inquiry reaches the right person while it still matters
- Captured and qualified as the inquiry arrives
- Routed and assigned by rules you set
- Followed up on a schedule when a prospect is not ready yet
- Picked back up when a lead goes quiet
CRM, integrations, and internal tools
The software you already own, agreeing with itself
- A CRM that reflects what is happening, because the systems around it write to it
- Agents that carry a task end to end and hand off where they should
- A tool built around your process, when no product fits the way you work
Training, handover, and ongoing optimization
You end up owning this, not renting it
- Handover on every build, deeper training when you want it
- Your team can tell a good use of AI from an expensive one
- After launch we stay on and fix what did not land
- Then we go looking for the next thing worth doing
AI automation for invoices, quotes, orders and leads
Each kind of work has a system below that takes it over. They are grouped by the capability they come under, so you can see where each one fits.
Workflow and process automation
Back office automation
Invoices go out, approvals get chased and the CRM gets updated without anyone typing it in twice.
- Invoicing
- Scheduling
- Approvals
- CRM updates
- Data entry
Proposal and quoting
Quotes and estimates built from your own pricing and templates, ready for a person to check and send.
- Quotes
- Estimates
- Proposals
- Price lists
- Follow up on unsigned quotes
Order and exception handling
Routine orders move from the storefront to fulfillment on their own. The short shipment, the failed address and the return outside policy get routed to a person.
- Order entry
- Fulfillment handoff
- Returns
- Backorders
- Address and payment failures
Lead automation, routing, and nurturing
Lead capture and routing
An inquiry that lands after hours is logged, scored by your rules and waiting in the right rep’s queue. What you already know about them is attached to it.
- Web forms
- Shared inboxes
- Missed calls
- Referrals
- Trade show lists
Follow up and nurture
The quote that went unanswered gets its follow up on schedule, in your words, and the reply lands with the person who sent it. Nobody keeps a list.
- Quote follow ups
- Reactivation
- Reminders
- Review requests
- Renewal notices
CRM, integrations, and internal tools
Document processing
Statements, invoices, forms and inspection reports get read, sorted and filed. The figures land in the system that needs them.
- Supplier invoices
- Bank statements
- Applications
- Inspection reports
- Contracts
Internal knowledge search
How do we handle a return past ninety days? The answer comes from your own procedures, in plain words, with the page it came from shown beside it.
- Procedures
- Policies
- Past quotes and jobs
- Product specs
- Contracts
Reporting and dashboards
One place to see sales, cash and open orders. The systems themselves keep it current, instead of a person each Monday.
- Sales
- Cash
- Open orders
- Job margins
- Hours
- The weekly numbers
Custom internal tools
The job you run from a spreadsheet, a pricing calculator or an approval queue, rebuilt as a small tool with rules. It connects to the systems you already own.
- Job boards
- Scheduling tools
- Pricing calculators
- Approval queues
- Customer portals
Training, handover, and ongoing optimization
AI training for your team
The person who runs a system can change a rule, add a step and read the log without calling us. The rest of the team learns to get real work out of ChatGPT and Claude.
- Handover sessions
- Written procedures
- Tool workshops
- An AI use policy
Start with the assessment. Choose your pace after
The 30-minute AI assessment names the fixes worth making. Then you choose how many we build at once, and we scope the engagement to your list.
Get the first fix live
1 build runningWe build the top fix on your list, train your team on it, then start the next.
Start your 30-minute AI assessmentWhat you get:
- The next build, taken from your list
- Every live system checked and kept running
- One training session for your team each month
- A monthly report on what each system saves you
- Most common starting point
Fix two things at once
2 builds runningWe build two fixes side by side and coach whoever runs them day to day.
Start your 30-minute AI assessmentEverything in the first, plus:
- A second build running beside the first
- Two training sessions a month
- Coaching for whoever runs the systems day to day
Work through your list fast
3 builds runningWe build three fixes at once, with weekly training and one of our founders in your monthly review.
Start your 30-minute AI assessmentEverything in the second, plus:
- Three builds at once
- Weekly training and coaching
- One of our founders in your monthly operations review
Scoped to your list, after the assessment
Your engagement is shaped by the ranked list of fixes we find together, not by a rate card. You decide whether to go on once you have seen it. You can stop after any step.
Questions about our AI services
How is this priced?
After the 30-minute AI assessment and Find, you get a ranked list of fixes. We scope the ongoing engagement to that list and to the pace you choose: one, two or three builds at once. Nothing is priced from a rate card on the site. You see scope in writing before anything starts.
Do we have to commit to anything ongoing?
No. Find stands on its own, and so does one build after it. The ongoing engagement is for companies that want someone owning this month to month. You decide after you have seen the operations map.
What if Find turns up nothing worth doing?
Then we say so, and the map says why. You have a written answer on where AI does and does not pay in your business, and there is no next step to buy.
How do you decide how many builds run at once?
From the map. If three processes each show a clear return and none depends on another, three can run side by side. If one has to land before the next makes sense, one at a time is the right size, and we will say so.
Do we have to buy all ten of these?
No. We start with the two or three problems worth solving first. The list is what we can do, not a menu to work through.
Can you work with the systems we already have?
Yes, and we try that first. We look for the cost in the gaps between systems before we blame the systems. Connecting what you own costs less than replacing it, and it disrupts less.
What if the answer is not AI?
Then we say so. Sometimes the fix is removing a step, changing who does what, or a plain integration with no model anywhere near it. Business process improvement is on this list for that reason.
How do we know it worked?
You see it in the work first: the task that took your team hours now runs without them. We also note how each process ran before we changed it. You get a report against that on a regular schedule, for as long as we are engaged.
The questions people type into a search box, answered one term at a time in the glossary





