AI consulting and custom software

We help you hand work to AI, one task at a time.

We automate in stages: first AI helps your team, then it prepares work for someone to review, and once it has passed testing, it does the job on its own. Your people focus on what they do best, and the same team gets more done.

Book your free assessment See our work

Services

What we do

Our core service

Consulting with engineers embedded in your team

Our engineers sit with your team, a model known as Forward Deployed Engineering. Together with you they decide which tasks move to AI and how much autonomy each gets, build the workflows and train your people with our method, the RL3 AI-First Playbook. Then they stay with you as AI takes on more of the work.

A map of today's process that shows where the hours go
An autonomy level for every task, from "a person does it" to "AI coordinates"
Integration with the tools you already use, with limits and a stop button from day one
Hands-on training designed to meet the AI literacy duty in Article 4 of the EU AI Act

Custom software

We build the software your business needs, with or without AI, starting from how your team works. You decide where it's hosted.

CRMs and internal tools
Web apps for your customers
Android and iPhone apps
Integrations and APIs that connect your systems

RL3 proprietary software

Our own platforms, built with our R&D investment and licensed to clients and partners.

AI feedback widget: bug reports and suggestions from your users, with screenshots, turned into tickets
AI onboarding engine: due diligence, KYC/KYB and screening, connected by API or run through its own portal
AI-first frameworks: the foundation every AI project of ours starts from
More products on the way

AI agents and automation

We automate processes and build AI agents. Some help your team; others do the whole job.

Customer service on WhatsApp
Appointments and bookings in your calendar
Data that moves between your tools without copy and paste
Your team gets a heads-up when something needs a person

Data and business analytics

Your data in order, and a dashboard to follow the business.

Data architecture
Data engineering
BI and dashboards

Our work

What we've built

Efímero ClinicHealthcareAesthetic medicineSpain

Sofía, an AI receptionist on WhatsApp

She answers patients in Spanish and English and books, moves and cancels their appointments directly in Flowww, the clinic's practice management software.

Starting point

At an aesthetic clinic, the front desk spends much of the day answering the same questions, moving appointments and chasing confirmations. Anything that arrives in the evening or at the weekend waits until the next working day.

What we built

The clinic's official WhatsApp Business number (Meta API), in Spanish and English
Books, moves and cancels appointments in the Flowww calendar, and confirms date, time and practitioner with the patient before saving
Tells a patient's usual practitioner apart from "whoever's free first", and handles joint appointments for two people

See details

New patient registration inside the chat
Confirmations, reminders, before and after treatment instructions, review requests and no-show follow-up
A staff console on the phone with notifications, a shared inbox and activity and cost dashboards, ready for several clinics
The clinic updates prices, aftercare and policies from the console, without touching code

Controls

Answers questions and quotes prices only from clinic-approved information
If it spots an urgent symptom, it doesn't diagnose: it passes the case to staff
Blocks attempts to manipulate the assistant (prompt injection)
Instant opt-out when a patient sends the opt-out keyword
If the calendar is unreachable, it tells the patient and passes the request to staff
GDPR: set retention periods, encrypted credentials, two-factor access and a tamper-proof audit log
Every change is checked against 40 test scenarios, including multi-message conversations

Technology

WhatsApp Business (official Meta API)
Flowww integration
Installable web app (PWA) with push notifications
Spanish and English

Capella and Velox SolutionsCorporate servicesComplianceUnited Arab Emirates

Compliance Brain, an AI-native CRM for corporate services firms

One CRM for the firm's three service lines: corporate services (CSP), regulatory authorisations and outsourced compliance. Clients are onboarded with KYC/KYB, each engagement is invoiced, and every step lands in an audit trail the regulator can inspect.

Starting point

A corporate services firm usually runs each service out of its own spreadsheets and inboxes. When the regulator asks who approved a client and on what evidence, someone has to piece it together by hand.

What we built

Company formation and administration, licence applications and outsourced compliance in the same CRM. An enquiry is entered once and moves to its line without anyone retyping it
Onboarding runs in stages from first contact to approval, with the 10-day service level always in view
AI agents check every person and company against UN, UAE, OFAC, EU and UK sanctions lists, PEP sources, public registers and the press, and cite each source on the file

See details

Works out beneficial owners (UBOs) across the whole ownership chain and alerts the team when an ID document expires
Reminds the team ahead of annual returns, fees and renewals
Each engagement carries its invoicing and the expected regulator fees
Several companies on one platform, each with its data kept apart
Underneath runs the RL3 Onboarding Engine, our due diligence and KYC/KYB engine. We also license it to other companies, by API or through its own portal

Controls

The MLRO signs every onboarding after the person who prepared it, with a written rationale
No sanctions match is dismissed without an analyst's review, and a confirmed match stops the case that same day
Lowering a client's risk tier needs an MLRO rationale that can't be edited afterwards
The audit trail can't be altered and exports as a dossier for the regulator
Internal suspicious activity register (SAR)
Two-factor sign-in and role-based permissions

Technology

Multi-company CRM
AI-agent KYC/KYB
UBO calculation
Tamper-evident audit trail
RL3 Onboarding Engine

Does your business have a similar process?

Book your free assessment

Security and compliance

The frameworks we follow and the controls every system ships with.

We work in line with these frameworks. This is not a certification.

EU AI ActA risk level and human oversight for every workflow, plus AI literacy training for your team (Art. 4).
The Act sorts AI systems by risk. Before we build, each workflow gets a risk level from R0 to R4 that caps how much autonomy the AI has, and we write down who oversees it and how. We also train the people who will use it, because Article 4 requires your team to have sufficient AI literacy.
ISO/IEC 42001AI management system: inventory, risks and continuous improvement.
It's the international standard for managing AI inside a company, the way ISO 27001 is for information security. From it we take an inventory of every AI system with its purpose and risks, written responsibilities and a regular review of results. Each cycle of our method ends by measuring and adjusting, which is the continuous improvement the standard asks for.
NIST AI RMFGovern, map, measure and manage every AI risk.
It's the US standards institute's framework for managing AI risk. It works through four functions: govern (who decides and who is accountable), map (what can go wrong and who it affects), measure (tests before every change) and manage (what we do when something fails). We use it as a checklist for every workflow.
GDPROnly the data needed, defined retention periods and a record of who accessed what.
Each system only reaches the personal data it needs for its task, with defined retention periods and a record of who looked at what. You decide where the data is hosted: in the cloud, on your server or on your existing infrastructure.
Encryption and two-factor accessEncrypted traffic and credentials, and two-factor sign-in.
All traffic is encrypted. Access keys to your systems (APIs, databases, accounts) are stored encrypted and never in the code. Signing in to the dashboards takes a password and a second factor.
Role-based permissionsData kept separate per company. Agents only use the tools they're cleared for.
Each person only sees and does what their role allows. Every company's data is kept separate, so nobody sees another's. Agents follow the same rule: they only use the tools and data they're cleared for, and we filter out hidden instructions that try to manipulate them.
Audit logEvery action is recorded, whether a person or an agent took it.
We record who did what and when, whether a person or an agent: which data it read, what it answered and which action it took. It helps you understand a mistake, answer an audit and show the system does what it says.
Human oversightImportant actions need confirmation, and a stop button hands the work back to your team.
Sensitive or irreversible actions, such as sending something to a customer, changing a case file or deleting data, go through a person before they run. If something goes wrong, a stop button hands the work back to your team without taking the service down.

Contact

Which task eats up most of your team's week?

Describe it in two lines. We'll tell you how far AI can take it and where we'd start.

What happens next

We reply within one business day.
We call you for the free assessment.
If we decide to go ahead, we send you a written proposal.

Prefer WhatsApp?Message us on WhatsApp and we'll reply there.Contact via WhatsApp