Skip to content

Solutions

AI consultants. From analysis to execution.

We are a strategy consultancy specialized in artificial intelligence. We analyze your business, decide where AI adds value and execute it with the best developers.

How we work

  1. 01

    Strategy first

    We are strategy consultants specialized in AI. The analysis decides what gets done, not the tool.

  2. 02

    Not tied to one technology

    We choose what fits each case best: our platform, other tools on the market or custom development.

  3. 03

    Highly experienced developers

    Senior engineers who build, integrate and launch everything that comes out of the analysis, with the robustness a real system demands. The best developers, not just good ideas.

Real case

A real project, step by step.

Lola is the WhatsApp receptionist of a veterinary clinic. This is how we went from its clients’ conversations to an agent built on our platform and put it to work.

01

Diagnosis

First, read what clients actually ask.

We analyzed a random sample of the clinic’s WhatsApp chats, from November 2025 to November 2026.

The ten most frequent requests, as a share of the total

  • Medication and doses20 %
  • Booking an appointment19 %
  • Symptoms and photos14 %
  • Product orders10 %
  • Prescriptions8 %
  • General information6 %
  • Invoices and payments4 %
  • Changing or cancelling an appointment4 %
  • Test results3 %
  • Prices3 %

Needs the vetAppointments and admin

  • 40 %of conversations arrive when the clinic is closed.
  • 77 %of what arrives when the clinic is closed waits until it opens: it only gets “Thanks for your message”.
  • 57 %of what people write is clinical: doses, symptoms, prescriptions, results and reports.
02

Design

Then, decide what the agent answers and what it doesn’t.

More than half of what comes in is clinical, and a vet answers that. Lola takes appointments and information, above all when the clinic is closed.

  • The first time we identify the client. After that, Lola recognizes them by their phone.
  • Appointments are left as “pending confirmation” in GestorVet, the clinic’s software.
  • While Lola is booking an appointment, the team does not reply over it.
03

Build

Build it on Neulogia Platform.

An agent with memory and 10 tools connected to GestorVet, the clinic’s management software, through its official API: it finds the client, looks up their pets, appointments and vaccinations, checks free slots and creates the appointment.

Lola’s real flow
Lola’s flow in Neulogia Platform: an AI agent with memory and ten tools to find clients, see pets, appointments, vaccinations and free slots, and create appointments.
04

Testing

Test it before it talks to any client.

Every design decision becomes a test: a simulated client writes as a real one would and Lola’s reply is checked. They are repeated with each version, in a testing environment separate from production.

Test suite · sample run
23 of 24 tests passed
  • They ask for an appointment while the clinic is openIntroduces herself and books it.Pass
  • They ask about a dose or send a photoSays nothing: the team replies.Pass
  • They write when the clinic is closedHandles it and says the clinic is closed.Pass
  • It is an out-of-hours emergencyGives the on-call phone number.Pass
  • It is the first time they writeIdentifies the client before booking.Pass
  • They ask for something she cannot solveLeaves it in writing for the team.Review
05

Rollout

Take it to production.

The version that passed testing is published to production as is, and Lola starts replying on the clinic’s WhatsApp.

  • The same version that passed testing, unchanged and with rollback.
  • The clinic’s WhatsApp Business number, connected to the platform.
  • The team, trained: what Lola does, what she leaves for them and when not to reply over her.
06

Analytics

Measure it.

Every conversation leaves a trace: how many Lola handles, how many appointments she books, how many arrive with the clinic closed and what goes to the team. They are the same questions as the diagnosis, now answered every week.

Lola’s analytics · one-week projection
  • 164Conversations
  • 47Appointments booked by Lola
  • 41 %With the clinic closed
  • 8 sTo reply

Conversations per day

How each conversation ends

  • Appointment booked29 %
  • Information given13 %
  • Passed to the team58 %
07

Savings

Work out what it saves.

With the analytics in hand, the saving stops being a promise and becomes a sum: the conversations Lola solves on her own, times how long each one used to take the team.

Estimated savings · per week
  • 2 ha week less of handling messages for each vet (there are 2)
  • 192 ha year for the clinic: 24 working days

Hours a week replying on WhatsApp

Before9.6 h
With Lola5.6 h
08

Continuous improvement

And improve it.

With that data the agent is tuned and we choose what else it can solve. The diagnosis itself already points to the candidates.

What Lola can solve next, according to the diagnosis

  • 10 %Product ordersof requests
  • 4 %Invoices and paymentsof requests
  • 4 %Changing or cancelling an appointmentof requests

And the wheel starts again: with what is measured, another diagnosis.

The diagnosis, the design and the flow are real, published with the clinic’s permission: a random sample of 26 chats with 210 requests. GestorVet is a trademark of its owner.

Our platform, as an accelerator.

We have our own platform to create, test and monitor text and voice agents. When it speeds up a project we use it: demos in minutes, automated testing and monitoring from day one. But it is not a condition: we do not work exclusively with it, and if another solution fits better, that is the one we propose.

Discover the platform

From analysis to execution, with the same team.

  1. 01

    Diagnosis

    Map of prioritized use cases and roadmap.

  2. 02

    Design

    Functional and technical design, fixed budget and success indicators.

  3. 03

    Testing

    Test report with results measured against the indicators.

  4. 04

    Rollout

    Solution in production and trained teams.

  5. 05

    Continuous improvement

    Managed service with periodic performance reports.

Types of agents we build.

Each use case calls for a different type of agent. These are the most common.

  • Customer service

    They handle calls, WhatsApp, web and email: they answer questions, manage issues and hand the case to a person when needed.

    • FAQs
    • Issues
    • Appointments
  • Sales and lead generation

    They engage interested visitors, qualify the request and pass good leads to the sales team.

    • Qualification
    • Bookings
    • Follow-up
  • Voice agents

    They talk on the phone naturally, on inbound and outbound calls, integrated with your phone system and tools.

    • Inbound
    • Outbound
    • Reminders
  • Internal agents

    They support your employees in HR, IT or finance, and instantly solve what used to go through several people.

    • HR
    • IT
    • Finance
  • Assistants

    Copilots for your teams: they search your company knowledge, draft, summarize and prepare the day’s work.

    • Knowledge
    • Drafting
    • Meetings
  • Process agents

    They run multi-step tasks across your systems, with a person overseeing what matters.

    • Back office
    • Reconciliation
    • Processing

How to hire us.

  • To startDiagnosis

    The way in: a fixed-scope analysis that ends in a plan you can execute with us or without us.

  • Per use caseOne project per use case

    Each case is an independent project, with a fixed or phased budget. You choose which ones and in what order.

  • Always onOngoing support

    Operation, monitoring and improvement of agents already running, so they keep delivering.

Start with what helps you most.

Ask for a tailored demo or tell us about your case and we will frame it together.