Features
Every Beluga service was born from a problem I had at the studio. Here is what each one actually does, and what it fixed for me.
The services
Nothing theoretical: these are the tools Haiko uses to run its own projects and its clients’ projects, every day.
The left-hand column gathers every project with its alerts. Each project has its board, and several agents work in parallel on different projects, or on the same one.
What it fixed for me: I juggled chat windows and never knew where each job stood.
A launched card runs on the server and carries on without me. The app installs on a phone: I launch, answer a question or read a report from anywhere.
What it fixed for me: Work stopped as soon as I closed the laptop. Today I code on the move too.
Each card keeps its request, its plan, its ticked task list, its report and its checks. A changelog is written automatically at every release.
What it fixed for me: I had to piece together by hand what had been done, and when. Now the history is complete with no effort.
Every project is born with its repository, and every launched card gets its own branch and its own copy. Work is committed and backed up, then released when I decide.
What it fixed for me: Two tasks touching the same code trod on each other, and a forgotten project was saved nowhere.
Columns that follow the real path of the work, scheduled starts, notes, and a project memory sorted by topic that agents read before acting.
What it fixed for me: Decisions got lost in conversations, and every new task started from scratch.
Passwords, keys, tokens, server and database access: one entry per secret, in a vault shared by every project. Agents look up the access they need on their own.
What it fixed for me: I copied the same credentials into every task, in every conversation.
Each client has a space of their own: they write requests, attach files and follow progress on a simple board. A request can become a Beluga card.
What it fixed for me: Conversations were scattered across emails, messages and calls. They are now in one place, right next to the work.
Each website gets its own backup recipe, prepared with an agent and then run automatically. Clients find their backups in their space.
What it fixed for me: Backups relied on different scripts from one server to the next. It has become a service I can offer.
An agent writes the monitoring for each website, then Beluga replays it on schedule. An outage sends an alert to my phone, and the agent who knows the website can take over the incident to analyse it and carry out a fix plan.
What it fixed for me: I used to hear that a website was down from the client. Now I know before they do.
The board
From request to archive, the card changes column at each stage. The characters act as landmarks: you see at a glance where the work stands.
The card is born here with its conversation. The need, configuration and understanding are refined without touching code; a start can also be scheduled.
The complete plan is written and reviewed here. The card stays at this stage until you approve it and start the work.
A full agent has its own branch and project copy. It reads, edits, tests and ticks off its task list; if it asks a question, work time stops until you answer.
The agent has returned its work and checks. You read a plain report with machine time, senior hours and estimated cost before choosing what happens next.
Approved cards form the next batch. Deployment can stay manual or be automated per project; production remains a separate, tracked stage.
Deployed work leaves the active flow without disappearing. Reports, branches, costs and history remain available, while the production version is shown at the top of the column.
Day to day
Less admin, more projects delivered — and going live stays my decision.
Several agents work at once, each on its own card. My job has become deciding and reviewing, not doing everything by hand.
The work runs on the server. A commute or a wait is enough to launch a task, answer an agent or approve a report.
Every change has its card, its report and its line in the changelog. Finding a decision or explaining a change to a client takes seconds.
Nothing goes live without my go-ahead. Deployment and production release remain separate, tracked steps.
They drop their requests in their space and see where things stand. Fewer reminders, fewer lost messages.
Automatic backups and constant monitoring: I hear about an outage before the client does, and I can restore what is needed.
