Practice Area 02 · AI Systems & Automation
Document extraction, cited knowledge search, and workflow automation built into your operation—with tests that measure accuracy, controls on cost, and a person in the loop where it matters. And if plain software does the job better, we’ll tell you.
Juan José López MartínezFounder · Senior Software & AI Engineer
Question
Answer
Enterprise contracts can be cancelled within 30 days of the start date for a pro-rata refund1, as long as the request comes in writing from the account owner2. Refunds are always processed within 24 hours.. Processing time isn’t specified in your documents.Not found
Enterprise Agreement§4.2 · Termination
Customer may terminate within thirty (30) days of the Effective Date and receive a pro-rata refund of prepaid fees.
Renewal terms are invoiced annually in advance.
Refund Policy§2 · Requests
Refunds apply to subscription fees only; professional services are non-refundable.
All refund requests must be submitted in writing by the account owner.
01 · Where AI Pays
Each one starts with your real data and ends with a system you can measure, audit, and own.
01Document Extraction
AI reads the document. Plain code checks the result. People only see the uncertain cases.
What stays in your repository
invoice_0142.pdf → recordReading
02Internal Knowledge Search
Answers grounded in your documents, with the exact source—or a plain “not found.”
What stays in your repository
Knowledge base · your permissionsSearching
Answer
Either party may terminate with sixty days’ written notice.MSA §9.1
03Workflow Automation & Agents
Agents that do the routine work—inside limits you define.
What stays in your repository
Support workflow · explicit stepsRunning
Audit log
04From Pilot to Production
Tests, cost control, and safeguards that turn a promising pilot into a dependable system.
What stays in your repository
02 · How We Work
A model is one component. Reliability comes from the software around it: validation, tests, and a person where the stakes are high.
How We Take AI to Production
A fixed-price test on your real data. If plain software or a simple query does the job better than AI, you hear it here.
We agree on the accuracy you need and test against your hardest real cases before building anything around it.
Deployed in your infrastructure with validation, fallbacks, monitoring, and human review where it matters.
Standards
Every response is checked against schemas and business rules before it touches your data or triggers an action.
Enterprise agreements with zero data retention, or models running in your own cloud. Your data never trains anyone else’s model.
Every system ships with tests and logs, so you know how it behaves—and when it changes.
Who you work with
The people who scope your system are the people who build it.
Founded and led by Juan José López Martínez, Senior Software & AI Engineer with 6+ years of production experience. Every architecture decision and pull request is reviewed by a senior engineer.
LinkedIn ↗03 · Start a Conversation
Tell us about the workflow or documents you want to automate. We’ll tell you whether AI is the right tool. We take on a limited number of engagements at a time, so you always get senior attention.
No. We use enterprise endpoints with zero data retention (AWS Bedrock, Azure OpenAI, Google Vertex AI, or direct enterprise APIs), or open models running inside your own cloud.
By limiting what the model is allowed to do. Knowledge answers must cite their source. Extraction and automation results are checked against schemas and business rules, and uncertain cases go to a person.
Every system ships with a test suite built from your real cases, plus tracking of accuracy and cost per transaction. You see the numbers before launch and after every change.
No. Your code talks to models through a clean abstraction, so you can switch or mix Claude, GPT, Gemini, or open models as prices and capabilities change.
Yes—real-time voice agents for phone or in-app conversations, connected to your CRM and business rules, with the same guardrails as any other workflow. As with everything else, we start with a feasibility test on your real scenarios.
A clear picture of the manual workflow you want to improve and a small sample of real examples—say, 10–20 representative documents or tickets. We handle the rest.
With a fixed-price feasibility test on your real data, so you know whether it works—and what it would cost to run—before committing to a full build. Scope and price are agreed in writing before work starts and depend on your case, not on a rate card.
Talk through your workflow, your data, privacy requirements, and whether AI is the right tool, with a senior DILAGO engineer.
Your message goes to the founder:
Juan José López Martínez
Founder · Senior Software & AI Engineer