Skip to content

Your process, running on its own.

Automation, applied AI and custom software.

A beam of light enters a prism and leaves split into six rays: process diagnosis, automation, applied AI, integrations, custom software and ongoing operation.

What we do

Six capabilities, combined to fit the problem.

Shows, under each item, how it is done and with what.
  • Mapping · bottlenecks · metrics

    Process diagnosis

    First, we measure where the work stalls.

    Flow mapping with the people who do the work, time and volume measured per step, an improvement backlog ordered by impact and effort; the map becomes the spec of what gets built.

  • Workflows · queues · bots

    Automation

    The repetitive work runs on its own.

    Event-driven workflows with queues and workers, safe retries (idempotency, dead-letter queues), schedules, bots for systems with no API; every run logged.

  • LLM · agents · MCP

    Applied AI

    Assistants and agents working on your data.

    Language models through APIs, agents with tools, MCP (Model Context Protocol) servers so the AI reaches the systems you already use, with scoped access; evaluation sets before anything goes live.

  • API · ERP · webhooks

    Integrations

    The systems you already have start talking to each other.

    REST and GraphQL APIs, webhooks, queues, ERP connectors, idempotent retries and dead-letter queues; a transactional outbox when the source is a database we own.

  • Web · internal tools

    Custom software

    When nothing off the shelf fits your flow.

    TypeScript end to end, PostgreSQL, a typed API, two-factor sign-in and role-based permissions, three languages, versioned migrations and tests; the same base as our own products.

  • Monitoring · support

    Ongoing operation

    We run what we built.

    Monitoring with alerts, service health checks, error queues watched, small and frequent updates, a support window agreed in writing.

How we work

What holds in every job.

  1. Measured before and after.

    Time per step, volume, errors and rework measured in the diagnosis, recorded by the system afterwards, and compared in writing at the end of each delivery.

  2. Backup every day. Restore tested every week.

    Daily restic backup, encrypted on the client, to a bucket off the server; an automated weekly restore test comparing row counts per table with the backup's.

  3. Everything that changes is recorded.

    A built-in audit trail: every change made by a person or an automation produces a record with author, action and affected data, queryable inside the system.

  4. Security and LGPD by design.

    Two-step sign-in, role-based permissions, retention with automatic purge, secrets server-side only; it is what runs in our own product.

  5. No lock-in.

    PostgreSQL and open formats, full data export, repository and documentation handed over, no component only we know how to run.

In productionWe run our own software: BusinessManager, scheduling and customer service for service businesses, and the clients' site templates.

Contact

What is your challenge?

Pick one and message us. The message writes itself.

Rework and spreadsheetsSystems that don't talk to each otherI want to use AIOverloaded customer serviceDecisions without data
I'd like to understand what you could do for my business.