AI agents for hospitalshotelsfood producersshipping fleetsenergy assets
AI agents that read your data, explain what moved and why, and prepare the fix for your team to approve. Built for operators in healthcare, hospitality and manufacturing across the Gulf and Europe.
Margin is 16.5%, down 2.1 points this week. Every other site is flat.
1Main supplier up 8% on 3 Nov, against a fixed-price contract.
2Two deliveries invoiced, never received.
3Overtime at double the norm on Tuesday and Wednesday.
Proposed
Query the supplier, citing the contract (drafted). Hold the two invoices. Rebalance the Tuesday roster.
The problem
Dashboards show averages. The money hides in the exceptions nobody opens. The branch that drifted. The supplier that crept up. The claim never resubmitted. Finding them is a full-time job. Fixing them is another.
How it works
01
One trusted picture of every branch, product, shift or patient. Built from the systems you already run. Refreshed daily.
02
The agent finds what drove each change and says it plainly. Which supplier. Which shift. Which policy. You trust the diagnosis before you act.
03
It prepares the fix: a supplier query, a corrected claim, a new rota. Inside your systems, within your limits. Your team approves. The agent carries it out and logs every step.
What a manager receives
Not a dashboard. A short note each morning: what moved, why, and what the agent proposes.
Operations brief · Tuesday
Illustrative example
See
Food cost at the Jeddah branch rose 2.1 points this week. Every other branch is flat.
Why
Three drivers. Poultry supplier up 8% on 3 November, against a fixed-price contract. Two deliveries invoiced but never received. Tuesday and Wednesday waste at double the branch norm.
Proposed
Query the supplier, citing the contract (drafted). Hold the two unmatched invoices. Cut Tuesday prep by 15% for two weeks, then review.
Under the hood
Your data stays in your warehouse. The agent reads it there, reasons with Claude or OpenAI, and acts only through rules you can read. Every rule fits on one screen.
agent: revenue-leakage watches: pos.transactions, erp.invoices baseline: 8 weeks, by branch flag_when: variance > 2% may_do: draft_supplier_query hold_invoice (up to 5,000 SAR) needs_approval: everything else log: always▍
Reads, never copies
The agent queries your warehouse in place. Nothing leaves your region.
Rules you can read
Plain text, versioned like code. Change one and the log shows who and when.
Limits that hold
Spend caps and approval thresholds are enforced in code, not in a prompt.
The film 1:30
In production
Where it pays off
Pick your sector. Four places profit leaks, and what the agent does about each. No new systems. Just something that opens every case.
Hospitals: Revenue cycle, staffing and patient flow.
Four places it pays off
01
Rejections pile up and expire. The agent groups them by cause, drafts the appeal and flags the coding fix.
02
Procedures billed below what was done. The agent matches notes to codes and proposes the correction.
03
Rotas are set a month ahead. Census moves daily. The agent forecasts by ward and proposes the shift changes.
04
Discharges slip to the afternoon. The agent finds where the delay sits, pharmacy, transport or paperwork, and flags it before noon.
Hotels: Bookings, outlets and rotas.
Four places it pays off
01
OTA commissions and rate mismatches go unchecked. The agent reconciles bookings to invoices and flags the gaps.
02
Food cost drifts by restaurant and bar. The agent reconciles sales to stock daily and names the item.
03
Housekeeping and front desk staffed to habit. The agent forecasts by day and proposes the rota.
04
The same issue on the same floor or shift. The agent groups them and proposes the fix.
Food producers: Planning, purchasing and getting paid.
Four places it pays off
01
Plans built on last year. The agent forecasts by SKU and customer and flags the lines that will run short or long.
02
Prices move. Contracts do not. The agent checks every invoice against contract and drafts the query.
03
Complaints and returns logged in three places. The agent links them to batch and line and raises the pattern.
04
Supermarkets and distributors pay late. The agent tracks every invoice against terms and sends the right reminder.
Shipping fleets: Crewing, procurement and compliance.
Four places it pays off
01
Parts bought at port prices. The agent compares quotes to history and flags the outliers.
02
Disbursement accounts arrive late and wrong. The agent matches them to port calls and raises the differences.
03
Contracts, visas and flights tracked by hand. The agent keeps the calendar and flags the costly late change.
04
Reporting assembled from spreadsheets. The agent pulls the data from your systems and drafts the filing.
Energy assets: From inspection to work order.
Four places it pays off
01
Drone and sensor findings sit in reports. The agent turns them into prioritised work orders.
02
A string or turbine produces below its neighbours. The agent spots it, names the likely cause and proposes the check.
03
O&M invoices arrive without evidence. The agent matches them to completed work orders before payment.
04
Filings and curtailment reports on a calendar. The agent assembles the evidence and drafts the filing.
Behind every sector, the same seven levers: revenue leakage, buying price, labour, waste, receivables, penalties, and the exceptions nobody opens. Insight without action changes nothing.
Solutions
Healthcare
Pilots openDenied claims resubmitted in days, not quarters. Staffing matched to census. Patient journeys fixed where they stall. A 90-day pilot, scoped with your CIO and revenue cycle lead.
Talk to us about a pilotWhere we are going next
Track record
What we learned: nobody needs another dashboard. They need to know why the number moved, and a fix ready to approve.
5 markets
Saudi Arabia, UAE, Egypt, Jordan, Kuwait
3 sectors
Restaurants, retail, hospitality
4 languages
Arabic, English, Greek, French
[George's full name]
Co-founder, commercial. Led analytics and AI adoption for operators across the Gulf. Co-founded and sold an F&B analytics company.
[LinkedIn]Anastasios Anastasiadis
Co-founder, CTO. Built the data and AI platform behind that company. Leads the Aigentyq platform.
[LinkedIn]Security and trust
Data stays in your region
Hosted in-region on Snowflake. Covered by EU GDPR, KSA PDPL and UAE PDPL.
You set the limits
Which systems, which actions, which amounts need a human. Nothing moves outside them.
Every action is logged
What the agent saw, what it concluded, who approved, what changed. Auditable end to end.
Best-in-class models, your choice
Reasoning runs on Claude and OpenAI under enterprise terms. No training on your data. No surprises on the invoice either.
Straight talk
Logos of companies we have not worked with.
Testimonials we did not earn.
A price you learn after three meetings. We tell you in the first call, in writing.
Your data used to train anything.
A contract that traps you. Finish the pilot, keep every output, leave if you want.
A revolution. We fix leaks and give your team its week back. That is enough.
Thirty minutes. We will tell you plainly whether an agent would pay for itself in your business.
Book a discovery call