Custom AI & automation
Your business.
Your AI.
Find answers. Draft replies. Spot changes.
AI shaped around your information, your tasks and your team.
Spend less time searching technical files, drafting repeat replies and checking product updates, with your team reviewing the results.
What could Sorted IT build for you?
See what a Custom AI solution could do for your business.
Sorted IT builds custom AI solutions and automation. We can explore your existing information and processes, then design a workspace around what would be useful to your team.
Start with a useful task. The information, tools and review steps would be agreed with you.
Prewritten examples. No AI or live connections.
Start with one task. Grow when it makes sense.
Pick your starting point
Sound familiar?
Three fictional examples. Choose one to try.
Step inside the workspace
Less explaining. More exploring.
A full-screen example hub. Find an answer, prepare a reply and review a product update — all in one place.
Sorted IT Custom AI · Demo Workspace
Your information. Your tools. Your review.
INTERACTIVE DEMO · FICTIONAL DATA · NO LIVE CONNECTIONS
Discuss a taskYour custom AI hub
Your work, brought together.
Find. Draft. Monitor. You stay in control.
Your tools
Choose a task to tryLatest in the demo
13 Oct 2026 · fictionalYour team stays in control
Review before the next step.
Inspect each example, then try marking it reviewed. No email or business record is changed.
Demo markers reset when you reload. In a real project, approval rules, review records and permitted actions would be agreed with you.
The right information. Agreed access.
Sources & connections.
Explore the example inputs and possible system connections. Nothing is connected by this demo.
Deployment & access choices
Some solutions may run on customer owned hardware or use an approved AI service, depending on requirements. Sources, permissions, security requirements and sensitive actions need to be scoped and tested with the customer.
A possible solution, at a glance
One workspace. Built around you.
The Brain is the chat layer. The Intelligence Hub brings agreed tools and systems together.
Example design · select a layer to explore
TO BE AGREEDDeployment & connections
A deployment choice to discuss. Some solutions may run on customer owned hardware or use an approved AI service, depending on requirements. The workspace, connections, information access and review rules would be agreed with the customer.
01 · Approved files → answer → staff review
Ask. Find. Check.
Pick a question. Inspect the cited source.
4 fictional technical files
Input · fictional document library
Choose a sample question
How this could work
A real solution could search approved technical files through a tailored interface. Answers can cite sources and acknowledge gaps; accuracy still needs evaluation. Open WebUI is one possible interface choice, not a required technology or a connected system.
02 · Incoming request → context → reviewed draft
From request to reply.
Choose an email. Edit the draft. You decide what gets sent.
Input · fictional incoming email
Show the context used
Your review → your send. Copying only puts text on your clipboard.
How this could work
A real setup could prepare drafts using approved business information. A person reviews and sends them. This page keeps your draft edits in memory only, until it is closed or reloaded.
03 · Agreed sources → comparison → review queue
Spot the change.
Open a finding. Compare the snapshots. Decide what matters.
3 fictional sitemap sources
Prewritten report. No live scan.
Prewritten demonstration report
4 sample pages compared: 1 new product, 1 newly listed part, 1 changed specification and 1 with no change.
3 findings for staff review. No purchases, records or source pages have been changed.
Output · sample findings feed
4 EXAMPLESHow this could work
A real setup could check agreed public websites through their sitemaps, compare product pages over time and flag possible new products or parts. Staff inspect the original source and decide any follow-up. The client’s source list, permissions, frequency and monitoring rules would be agreed during the project. All .example addresses here are fictional and are not opened.
Possible custom workflows
What else could your Custom AI help with?
Pick an area. Explore two possible custom workflows.
We could map the task, identify the information and agree where people review the result. Then scope a small, testable version together.
This explanation is prewritten. Your task has not been analysed by an AI model.
An illustration using your own numbers
Put a number on the task.
Your figures. Transparent calculation. No savings promises.
Your current workload
Estimated staff hours spent on this task in a year
Enter your three workload figures to see the estimate.
Potential staff capacity to free up
—Add an optional assistance percentage to explore this.
Potential capacity = annual staff hours × your assistance estimate ÷ 100
An illustration using your numbers and a 52-week year. No savings guarantee, financial forecast or promise to remove the task. Review time and practical constraints still matter.
From one task to a useful solution
Start small. Learn. Build.
An example path, shaped around your team. No fixed schedule.
Understand how the task works today.
Identify useful information and suitable systems.
Build a small, testable version.
Add agreed connections and review steps.
Test it with the people who will use it.
Improve it based on feedback.
Got a task in mind?
Bring one recurring task. Let’s explore what could help.