selvanegra systems
SMEs already up and running
Connects to what you already use and takes work off every area. Management gains time and sees earlier what is going off track.
Talk to usWe solve real inefficiencies in your company with AI, from an app for one key process up to the capacity of a large team.
We come from running companies.
We build agents through the eyes of the user and the business owner.
What it does
We find where your company loses time or money and fix it with AI. We work at the scale of the problem: sometimes an app for one key process is enough, sometimes it takes a team of agents across several areas. Always tailored to your company.
For example: a technical office reviews building permit applications. Each reviewer has a quota and files pile up. A tailored app sorts the documents, flags what is missing and drafts the report. The reviewer checks and signs.
For example: a small architecture and construction firm cannot keep up with quotes, invoices, progress billing and deadlines, and hiring another qualified person is not viable. The agents prepare that work. The firm reviews and approves.
A coordinator and agents for each area. The capacity of a large team, within reach of a small company.
Always on the same foundation
So even a small solution is secure from day one and can grow later.
When a problem repeats across a whole sector, an app born in one company can serve many. We take that step with selvanegra ventures. Each company's data stays with that company.
Weekly status, decision papers, action plan
13-week cash forecast, collections, budget against actuals
Pipeline, quotes, customer follow-up, content
Open orders, suppliers, price trends
Stock, deliveries, transport
Production, incidents, maintenance
Org chart, roles, onboarding. Never rates individuals.
Milestones, risks, changes with cost and time impact
Rules, data protection, legal deadlines
Research, interviews, business model
You choose the areas. We fit the system to your sector, your processes, your rules and your pace.
selvanegra systems
Connects to what you already use and takes work off every area. Management gains time and sees earlier what is going off track.
Talk to usselvanegra ventures
We use it to build businesses and see them through until they stand on their own: market research, business model, sales and administration, from day one and with a small team.
View venturesselvanegra consulting
When we run an area at a client, the system prepares the daily work. An experienced person can take on more with the same quality.
View consultingExamples
Every sector has its own. These are typical cases and how we would solve them.
Immigration law firm
Today
Every day client documents arrive by email and WhatsApp, and someone sorts them by hand.
With AI
An app files each one in its case, watches deadlines and expiry dates and drafts the next filing. The lawyer reviews and submits.
Wholesale distribution
Today
Orders come in by email, phone and PDF and are typed into the ERP one by one.
With AI
The agent enters each order in the ERP as a draft, checks stock and price and flags anything that does not match. Sales confirms.
Custom manufacturing
Today
Every custom quote takes days because similar jobs have to be found and costs worked out by hand.
With AI
The agent reads the request, finds similar jobs in the archive and proposes price and lead time. The manager adjusts and sends.
Farming cooperative
Today
Deliveries from hundreds of members, quality grades and prices are matched in spreadsheets before every settlement.
With AI
The agents match delivery notes, grades and prices and prepare each settlement with the differences flagged. Management reviews and approves.
Restaurants
Today
Supplier invoices pile up and nobody notices in time when the cost of a dish goes up.
With AI
The agent records every invoice, flags price rises and recalculates the cost of each dish. You decide prices and menu.
Property management
Today
Dozens of emails from residents every week, open incidents and minutes still to write.
With AI
The agent sorts each email, opens the incident and drafts the reply and the minutes. The manager reviews and sends.
Installation company
Today
Technicians fill in paper job sheets and the invoice goes out weeks after the job.
With AI
The technician dictates the job sheet on the phone. The agent makes it ready to sign, books the material used and prepares the invoice. The office issues it.
Haulage
Today
Signed delivery notes, tolls and fuel are reconciled by hand before each client is invoiced.
With AI
The agent reads the delivery notes, matches them with the routes and prepares each client's invoice with extras. Admin reviews and issues it.
What we build
For one process, an app with one agent is enough. For more, we set up a team: a coordinator shares out the work among agents for each area and every morning shows you what needs your decision.
Good morning. Five points.
Your point of contact. Hands out the work, follows up on open items and each morning sums up in a few points what to decide, what is off track and what is due.
One per area: finance, sales, purchasing, logistics. Each works to a written brief, like a head of department, and knows the figures and deadlines of its area.
Draft emails and quotes, turn voice notes and meetings into minutes, calculate key figures and file documents.
04Checks the others with fixed rules. Compares every figure with your books, stops what does not add up and records it.
Connects to what you already use
We start read only. Your software remains the source of the data, and you do not have to switch programs.
We tried it at home first: our group works every day with its own agent team. Where needed, we also build custom management software.
How we work
The agent reads, drafts, calculates and proposes. Paying, signing or sending something to a client is always done by someone on your team. The agent has no access to do it.
Agent
Person
Step by step
First we talk to the people who do the work every day. Then the agent helps with a single task. When it works well, it takes on more. If it gets worse, it goes back a step.
Local first
The agents work with AI models installed on a server in your company or on ours, within the EU. If a task needs an external model, you decide. Sensitive data does not leave.
Security
Clear, verifiable and without getting in the way of daily work. Organised along the NIST Cybersecurity Framework.
What each agent may do is written down. Only management changes it.
A list of all agents and of what each one can touch.
The bank, read-only. Payments and IBAN changes are confirmed through a second channel.
If data leaves that should not, an alarm goes off.
Every agent can be stopped at once. There is a written emergency plan.
If something goes wrong, we return to an earlier state.
Log
Every step an agent takes is recorded with date and time. Nobody can change the log afterwards, not even us.
········································The chain matches: every entry keeps the fingerprint of the one before.
Contact
Tell us. We will explain how we would solve it and where we would start.