We tell you which parts of your operation are worth automating with machine learning and language models, then design, build and deliver the application into production. Fixed scope, fixed price, agreed before work starts. You own the code.
Free, 30 minutes, no preparation. You leave knowing whether your case is clear enough for a proposal, needs a one-week assessment first, or is not a fit.
Prefer email? hello@xval.ai
Two kinds of work. Advice on what to build, buy or leave alone; and the build itself, delivered running in your systems.
1 week, €2,500, credited against a build that starts within 90 days. We go through your operation and your data and come back with a ranked list of what is worth automating, what to buy instead of build, what to leave alone, and a fixed-price proposal for the first build.
A pilot that stalled, a tool that was bought and not adopted, or a system another supplier built. We tell you whether it can reach production, what it would take, and whether it is worth it.
Half-day to two-day sessions for the people who will use AI at work: what it can and cannot do, how to use it safely with company data, and how to spot the next task worth automating.
Invoices, contracts, claims, supplier files, emails, tickets. Language models read them, extract what matters, classify, route and answer, with a person in the loop where the decision matters.
How many units will sell, how many people are needed on Saturday, which of 500 applications to look at first, which customers are about to leave. Models trained on your history, answering every week automatically.
Thousands of projects, records, manuals or tickets in your systems. Staff ask in plain language and get the right record, version or answer back, with the source attached.
Classification, extraction, recommendations or an assistant inside the software you sell. We work in your repository, follow your conventions and ship behind your feature flags, with evaluation data your team can rerun.
One clear, scoped workflow goes straight to a fixed-price proposal. Several candidates, an open question, or a stalled pilot go through the assessment first. When the agreed build needs a pipeline or dashboard to run, it is included in the fixed scope.
Not every project needs a model trained on years of history. Many first builds use language models to read, write or search what the company already has. One clear operational need, one working system.
Orders arrive as PDFs, scans and emails. AI reads each one and creates the order in the ERP. A person checks only the ambiguous ones.
Each new supplier sends a stack of documents. AI checks them against the required list and the regulation, and fills in the file for the quality team.
Photos come back from every collection round. AI spots the anomalies and writes a standard report with the evidence attached.
Descriptions written and translated from the data already in the ERP, in the brand’s tone. The merchandising lead approves before anything goes live.
Thousands of projects, versions and proofs in a bespoke tool. Staff ask in plain language and get the right record, version or history back.
A categorisation service in the company’s own codebase, with an evaluation set the team reruns on every release. Customers see categories; engineers see the metric.
One call, then either a proposal or a 1 week assessment, then a fixed-price build with a fixed end date. Assessments are €2,500, credited once against a build that starts within 90 days. Most builds are €5,000–€25,000 depending on data and integration scope. Maintenance is optional.
On the call you describe the task and the data; we say whether it is feasible. If the case is clear, you get a written proposal within a business day. If it is not, we run a 1 week assessment: data audit, ranked candidates, build-or-buy per candidate, a go/no-go and the proposal. Nothing is built before you sign off.
We connect to your data, design the architecture, build the application and deliver it where your team works: your ERP or line-of-business system, your product, a service your systems call, or a dashboard. The build ends with results against the acceptance test and a handover of everything we made.
We host, monitor and keep the system accurate: drift checks, retraining with new data, fixes and new features as your business changes. Or your team runs it; the choice is yours.
A chat tool gives you an answer. An application gives you the same answer tomorrow, inside your systems, with a number that says how often it is right. The difference is engineering.
We decide which part of the problem is a classifier, a forecast model, a retrieval layer or a language model, and where a person stays in the loop. The cheapest component that passes the acceptance test wins; a trained model is used where it beats a prompt, not by default.
Before the build: a baseline (what your team achieves today), a held-out evaluation set, and the metric in plain numbers. Every release is scored against it. You keep the evaluation set and can rerun it on any future version, ours or anyone else’s.
Delivered into the system where the work happens, with tests and documentation. In your repository and conventions where you have one. Implementation is AI-assisted; the engineer who owns the architecture and the acceptance test reviews and tests every change before release.
Monitoring, drift checks and retraining as a monthly service, or a documented handover (code, tests, runbooks, evaluation set) so your team or another supplier can run the system without us.
Verify: LinkedIn · Google Scholar
In a free 30-minute call we look at the task and the data behind it. You leave with one of three answers: a fixed-price proposal for a clear build, a one-week assessment when the question is still open, or a straight “not a fit”.
30 minutes with Pedro Marcelino, founder. No sales hand-off.
Prefer email? hello@xval.ai
