ENESStart a project
Artificial intelligence

Put AI to work in your business.

We build AI into your apps and systems so they serve your customers, find information, process documents and take on the daily work. Behind every project is a team of specialists, from the idea to production.

Tell us your use caseMessage on WhatsApp
Support assistantexample

Where is my order DZ-1001?

Looking it up…find_order({ id: "DZ-1001" })
{ status: "on_the_way", rider: "Mateo", eta_min: 12 }

Your order is on the way with Mateo and arrives in about 12 minutes. Want me to let you know when it is close?

Illustrative example: a support assistant for a delivery app.

Two ways AI pays off

AI as part of what you sell, and AI as part of how you work. Most projects start with one and grow into the other.

For your customers

AI in your product

Features your users see and value: an assistant that answers at once, search that understands what they ask for, recommendations and forms that fill themselves in.

  • A better experience
  • More conversion and retention
  • A product that stands out
For your team

AI in your operations

Internal work that eats hours today: reading invoices, preparing quotes, answering the same questions every day, searching through manuals.

  • Less manual work
  • Fewer errors
  • Faster answers

What we build with AI

AI that finishes a task inside your product, not a demo on the side. Each piece below can live in a web app, a mobile app or WhatsApp.

LLM

Assistants and chatbots

  • Support, sales and operations, in your app, site or WhatsApp Business
  • Connected to your CRM, ERP or database
  • Hand over to a person when they should
Agents · MCP

Agents that act

  • They read, decide and call your systems through APIs or MCP
  • A person approves the critical steps
  • Every action logged and traceable
RAG

Search over your documents

  • Answers with citations from manuals, contracts and databases
  • Permissions per user or team
  • Knowledge that updates as your documents do
OCR · Vision

Documents into data

  • Invoices, IDs and contracts turned into fields
  • Checked against your rules
  • Doubtful cases go to a person for review
iOS · Android

AI inside mobile apps

  • Voice, vision and recommendations
  • Forms filled from a photo or a voice note
  • On the device or in the cloud, as the case needs
Evals

Evaluation and costs

  • Automatic tests of answer quality
  • A panel of quality, cost and latency
  • Model cost per user under control

Use cases that tend to work first

We start where the process repeats, the data exists and the result can be measured.

Customer support

Problem
The same questions all day, and slow replies at night.
Solution
An assistant that answers from your policies and hands over to a person.
What we measure
Time to first reply and share solved without a person.

Finance and back office

Problem
Invoices typed in by hand, with errors.
Solution
Fields extracted automatically, with doubtful ones sent to review.
What we measure
Documents per hour and error rate against typing.

Internal knowledge

Problem
Answers buried in manuals nobody finds.
Solution
A conversational search with citations and permissions by team.
What we measure
Search time and accuracy on a test set.

Sales and quotes

Problem
Every quote rebuilt from scratch.
Solution
An agent drafts it from your ERP data for the seller to approve.
What we measure
Time per quote and quotes sent per seller.

Field operations

Problem
Long forms filled on a phone, outdoors.
Solution
An app that reads a photo or a voice note and fills the form.
What we measure
Time per record and fields filled correctly.

Digital products

Problem
A product that needs AI to stand out, at a cost that adds up.
Solution
AI features built in, with model cost per user controlled.
What we measure
Retention, conversion and AI cost per active user.

How we take AI to production

Four steps, and what you get from each one.

  1. 01

    A measurable use case

    The process with the most impact, the data you have and the metric. What the AI does and what a person keeps doing.

    You get: Case, data and metric in writing
  2. 02

    A prototype on your data

    We check quality, cost and adoption before building the whole product.

    You get: A prototype and its first evaluation
  3. 03

    Secure integration

    Connected to your apps and systems, with access control, defences and monitoring.

    You get: The AI live in your product
  4. 04

    Measure and improve

    Evals, adjustments to prompts, models and data, and a report against the first metric.

    You get: Continuous evals and adjustments

Models and technology

We choose the model by quality, cost, latency and privacy, and the product can switch models without being rewritten.

Models
  • Claude
  • GPT
  • Llama
  • Mistral
Connection
  • MCP
  • Tool calling
  • REST APIs
  • WhatsApp Business
Data
  • RAG
  • Embeddings
  • Hybrid search
  • PostgreSQL
Quality
  • Evals
  • Tracing
  • Cost per user
Security
  • OWASP LLM Top 10
  • Least privilege
  • Audit log

Safe AI, by default

An assistant that can act needs the same care as any system with access to your data.

  • Prompt injectionDefences and tests against instructions hidden in user text and documents.
  • Permissions per userThe AI only sees what the person asking is allowed to see.
  • Human approvalCritical actions wait for a person before they run.
  • Everything loggedEach answer and each action keeps its trace for review.
  • Your data stays yoursBusiness APIs that do not train models on your data.

What you get

  1. The use case defined, with its metric
  2. A prototype on your real data
  3. The solution integrated, with source code that is yours
  4. A set of evals and a quality and cost panel
  5. Documentation and a handover to your team

Specialists behind every line

Every project is designed, built and reviewed by experienced people. AI is one more tool in their hands: it speeds up drafts and tests, but the decisions, the architecture and the quality are ours.

  • Specialist judgementArchitecture, security and product decisions are made by people who know the trade.
  • AI as supportIt helps us write drafts and tests faster; it never decides for us.
  • Reviewed by peopleEvery change is read, tested and approved before it reaches your repository.
Illustrative example: a developer working with a coding assistant.

Questions about AI

Which models do you use?
Claude, GPT or open models such as Llama and Mistral, chosen by quality, cost, latency and privacy. The product is built so the model can change.
Chatbot or agent?
A chatbot answers. An agent also acts: it calls your systems and completes a task, with a person approving the critical steps.
Can it work on WhatsApp?
Yes, through the WhatsApp Business Platform, with the same knowledge and rules as in your app or site.
Is my data protected?
We use business APIs that do not train on your data, permissions per user, and a log of every answer and action.
How do you know it answers well?
With evals: a set of real questions with expected answers that runs on every change, plus the metric we agreed at the start.
Do I need tidy data to start?
No. The first step looks at what you have; often documents and spreadsheets are enough for a prototype.
Can you add AI to an app I already have?
Yes. We connect it to your current backend and screens, web or mobile, without starting over.
How much does it cost?
It depends on the case. After the first call you get a written estimate with the development and the expected monthly model cost, separately.

Which process should AI solve first?

Start a projectMessage on WhatsApp