AI projects don't fail because of AI. They fail because nobody fixed the process underneath.
And almost nobody knows what the AI they already pay for is giving back.
TuOpinas lays the foundations —reference architecture, risk lanes and a review rhythm— so the people who know the business can build at the speed AI allows, and what they build holds up.
Two practice areas, one principle: put things in order before speeding up
Every project starts with the problem, not the tool. Scope, timeline and price are agreed before work begins.
Practice 01 · Mid-sized companies
Processes and AI applied to operations
The AI has already been bought, but the process underneath is still a mess and nobody knows what return it brings.
Fix the process before automating itMap how the operation works today, where the hours go and what data isn't being produced.
Design and integrate the solutionArchitecture, selection and implementation of SaaS tools and ERP, integration between systems that don't talk to each other, and building what the market doesn't offer.
Lead executionTechnology leadership for companies without an in-house technical head: priorities, vendors, budget, and making sure what's delivered actually works.
Practice 02 · Software companies
AI in how the product is built, supported and run
With AI, a small team builds and runs what used to take a large one, as long as the workflow, security and support are solved from the start.
Cost and securityInfrastructure sized to what is actually used, and security reviewed before speeding up.
From design to codeAn AI-driven flow from design to user stories with acceptance criteria, code and automated tests.
Support from day oneAI agents coordinated with human agents, user manuals, and satisfaction and new-feature surveys.
The product in plain sightMonitoring of usage, code and uptime, with alerts before the complaint arrives.
Perspective
Two things almost nobody says out loud
Neither is an opinion: companies report both themselves, with date and source.
Almost nobody knows what the AI they already pay for is giving back
7%
Only 7 in 100 executives have managed to put a number on the return of their AI. Those who know what it costs them are five times more likely to get there.
The question that closes itHow much are you spending on AI each month, counting every area?
You don't automate a broken process
60%
Gartner predicted that through 2026, 60% of AI projects without AI-ready data would be abandoned. It's the only recommendation that asks you to do less, not more.
The question that closes itWhen someone asks for a number from the operation, where does it come from and how long does it take?
“AI doesn't fix a team; it amplifies what's already there.” Strong teams get better. Struggling teams watch their problems grow as fast as their speed.
Anyone who knows the business can already build, and will, with or without permission. The risk isn't that they build; it's that they build without foundations, because AI multiplies speed and disorder at the same rate.
Protect the speed of whoever builds, and concentrate control where the risk is real. Whatever crosses the line is reviewed before release, not after.
Lane
Free
Used only by its author, touches no personal data and doesn't write to a system of record: a personal dashboard, a macro, a prototype.
What's required
Nothing. It's built with the tools in the catalog.
Who reviews
No one.
Lane
Shared
More people use it, or it reads operational data.
What's required
An inventory entry, an owner, credentials kept out of the code, access through a corporate account.
Who reviews
The coordinator, at the regular forum.
Lane
Critical
Touches personal or customer data, writes to a system of record, serves several areas or goes out to a third party.
What's required
Architecture and security review before release, tests, a support plan and a cost center.
Who reviews
The architect and the security lead.
AI tools follow the same lanes. New ones appear every week and the team will try them before anyone else: that is protected, not banned.
Seven pieces, none of them new: lanes, data in a single place, a paved road, buy before you build, mixed teams, a lifecycle and a fixed rhythm. Classic software engineering and systems integration, put at the service of whoever builds.
And one rule that doesn't depend on the tool: AI consumption is measured like any other technology line item, with an owner and a threshold. Until that is measured, everything else is opinion.
How we work
A conversationUnderstand where it really hurts, before talking about tools.
Closed scopeFixed scope, timeline and price before starting, with a return metric the client can recalculate on their own.
Building by lanesWhatever doesn't cross the risk line moves without waiting; what's critical is reviewed before release.
HandoverWhat's built is documented, with an owner and a review date, so the team can carry on without depending on TuOpinas.
No published rates, and no savings percentages promised up front.
For twenty-five years I've been translating how a business runs into a system that works. What changed in 2026 is the price: what used to take a team and six months now fits in weeks.
2021 – today
TuOpinas.ai· Process automation and applied AI for mid-sized companies, and AI-driven development, support and operations for software companies.
2023 – 2026
Artificial Dynamics and GO SHARP· COO of a holding of five AI-powered SaaS products: implementations at ten large companies and a platform rolled out to 40 end clients.
2015 – 2026
Control Plus· Co-founder and CEO of an HR SaaS built from scratch: 20,000 employees managed across more than 15 clients.
2013 – 2015
Smile Pill· Transformation of an operation with 5,000 field employees, from spreadsheets to SaaS tools.
2012 – 2013
Convergys· Entry into Latin America with its software solutions for customer-management BPO.
1998 – 2011
Accenture· Software architect in Spain, Chile and Brazil. Then senior manager for the Telefónica and Telmex accounts in Mexico, with teams of up to 50 people, and the telecom practice in South Africa.
Telecommunications Engineer, Universidad Politécnica de Madrid. D-1 program, IPADE Business School. Spanish and English.
Contact
What do you have that was never worth fixing?
It's the question every conversation starts with. If something came to mind while you were reading, that's probably where the work is.
