Seradyx Icon
Seradyx
Software Lab
← Services
AI automation

Hand the repetitive work to software.

Custom AI agents and desktop tools that read your documents, fill your spreadsheets, write your reports and handle your inbox, with your own API keys or local models, and approval before anything important happens.

MailDriveExcelCRMAGENTworking…Send 3 emails?OK

What you get

Agents that do the work

Read folders, extract data, write Word, Excel and PowerPoint files, the way Hiel Cowork does.

Your keys, your models

OpenAI, Anthropic, Gemini or local models through Ollama. You pay providers directly, no markup.

Human in the loop

Approval steps before sending, deleting or paying, plus a log of every action taken.

Connected to your tools

Email, Google Drive, databases, CRMs and internal APIs, through MCP connectors or direct integrations.

Desktop or server

A desktop app for your team, a bot on Telegram, or a workflow that runs on your VPS.

Measured

Hours saved and error rates tracked, so you know the automation pays for itself.

How it works

  1. 01

    Map

    We watch how the work is done today and pick the steps worth automating.

  2. 02

    Prototype

    A working version on your real documents within days, not months.

  3. 03

    Harden

    Approval gates, error handling, logging and cost limits.

  4. 04

    Roll out

    Installed for your team, with training and a feedback loop.

Built in-house

Hiel Cowork (an AI coworker for Windows) and HielJobs (job search on autopilot) are agents we design, build and run ourselves.

See the apps
Tools we use
Python.NET / WPFMCPOpenAIAnthropicOllaman8nPocketBasePython.NET / WPFMCPOpenAIAnthropicOllaman8nPocketBase
Start a project

Tell us what you need

Two minutes. We reply within two working days with questions or a short call, then a fixed quote.

What should be automated?
AI provider
How sensitive is the data? *
Where should it run?
Budget
Timeline

We only use your details to answer this request. No newsletter, no tracking.

Questions about AI automation

We only use providers and settings that do not train on your data, or local models when the data must never leave your machines.

Other services