Oxyok
AI work layer
for food distributors.
Oxyok helps employees handle messy customer order requests across ERP, WMS, inboxes, customer portals, pricing, routes, and approval rules.
Customer asks for a repeat order, substitution, and next-day delivery
Oxyok checks history, sellable stock, price, route, and account rules
The employee sees risks before anything is promised to the customer
Approved actions can update ERP, WMS, CRM, or the customer reply
Where it starts.
One painful workflow before a broad platform.
The first pilot should automate a real exception path where employees currently move between systems, messages, and approvals.
Workflow 01
Order changes and substitutions
A customer wants to repeat an order, swap a product, or change quantities.
Workflow 02
Availability and delivery promises
The answer depends on sellable stock, warehouse status, route timing, order deadlines, minimums, and customer-specific terms.
Workflow 03
Actions across existing systems
Oxyok does not replace ERP or WMS.
Example pilot workflow
A simple customer message becomes real work across several systems.
The pilot starts from one repeated exception, not from a broad transformation program. The goal is to prove that Oxyok can reduce the manual checking before a customer promise is made.
Customer message
Can we repeat last week's dairy order, swap strawberries for blueberries, and still get it tomorrow?
Order history
Find the previous order, customer location, SKUs, quantities, order guide, and usual delivery pattern.
Sellable stock
Check stock that is actually available to sell, not only a raw inventory number.
Substitution rules
Compare product, pack, price, customer preference, margin, and whether approval is needed.
Delivery feasibility
Check tomorrow's route, order deadline, warehouse status, truck capacity, and delivery minimum.
Commercial constraints
Check customer-specific pricing, credit holds, account notes, and terms before confirming.
Human approval
Prepare the proposed action and customer reply, then stop for approval before writing to systems.
What Oxyok connects
The value is in the work between systems.
Oxyok is not another ERP, WMS, or ecommerce portal. It sits above the tools already running the operation and helps employees act across them with review and audit.
Inbox and customer channels
Email, portal messages, forms, PDFs, order notes, and other places where customer intent appears.
ERP and order history
Customer master, sales orders, price lists, credit status, order guides, invoices, and account terms.
WMS and inventory
Warehouse stock, reservations, picking state, lot or expiry constraints, and sellable availability.
Routes and delivery rules
Route calendars, order deadlines, truck capacity, delivery minimums, and operational exceptions.
CRM and account memory
Sales notes, customer preferences, previous exceptions, complaints, and approval history.
Approvals and audit
Clear review points before customer promises, order changes, substitutions, or system writes.
Oxyok Cloud
Managed runtime for vertical AI systems.
Oxyok runs the AI workflow layer in our managed cloud so vertical products can move from prototype to pilot without the operator managing AI infrastructure.
On-Premise AI System
Vertical AI systems close to your data.
Oxyok deploys the AI system inside your environment for operators that need local models, private infrastructure, stricter data boundaries, or deeper internal-system access.
How a pilot works.
01
Map one workflow
We identify the messages, roles, systems, data fields, rules, approvals, and edge cases behind one repeated customer request.
02
Build the work layer
We connect to the needed systems, prepare drafts and actions, show the reasoning, and keep approval with the responsible employee.
03
Pilot, measure, expand
We measure saved time, avoided mistakes, response speed, and adoption before expanding to more order, support, finance, or procurement workflows.
Pilot outcomes
The first pilot should be measured like an operational workflow, not a chatbot demo.
We pick one high-friction workflow and define success before building. If it works, it becomes the base for more workflows.
Speed
Faster customer answers
Less waiting while employees search order history, availability, pricing, route status, and account notes.
Accuracy
Fewer unsafe promises
Risks are visible before the team confirms a delivery, substitution, price, or order change.
Adoption
Useful inside daily work
The assistant prepares actions inside the existing workflow instead of asking employees to live in another dashboard.
Expansion
Path to broader automation
Once one exception flow works, the same layer can expand to order intake, quotes, support, finance, or procurement.