Optional depth · Architecture

How your Whiia harness works

Four parts, each explained by what it does for your work. This page is for when you want to look closer — you do not need it to start.

Every part below is made of the same thing: files you can open, read and change.

What each part means for your work

The same harness, described from your side of the desk: what your AI starts from, what it can point back to, how it should behave, and what it uses to do the job.

Part 1

Your business context

What your AI starts from: the goals, people, agreements and preferences you have saved. It matters because you stop re-explaining the same background in every conversation.

See what the files contain →
Goals and priorities you set Your direction

What you are working towards this season sits in front of the work, so a task you ask for is prepared against your priorities instead of from a blank prompt.

People, agreements and preferences Saved knowledge

The contacts behind a collaboration, what was agreed and how you like to work are stored as files you can open. You can check what is saved and correct anything that is out of date.

Corrections you have made Your corrections

When you correct an interpretation, the correction is kept with the context it belongs to, so later work has the chance to start from what you confirmed rather than the first guess. You can open the file and see for yourself.

Part 2

The sources you connect

The messages, notes, records and files you choose to give it access to. They matter because a summary is only useful when you can see where it came from.

You choose what it can use Your access

Begin with one source or one file. Add more when the next task needs it. Nothing is connected just because it exists.

The messages and records behind the summary Traceable output

A brief or an update should point back to the material it was built from, so you can see the evidence rather than take a conclusion on trust.

Found, suggested and confirmed Honest labels

Information gathered during setup, suggestions it proposes and facts you confirm stay distinguishable. A source is not treated as verified simply because it was fetched.

Part 3

How you like to work

The instructions you write for your AI: tone, format, boundaries and what needs your review. They matter because the work comes back in the shape you actually want.

Your voice and format House rules

How you like messages, documents and summaries to read, kept as instructions your AI applies instead of preferences you restate each time.

What waits for your review Oversight

You decide which actions are prepared as drafts and which can proceed. For anything that leaves your business, the rule is: review the draft before sending.

Boundaries you confirm Your limits

The limits you set — what needs asking first, what is out of scope — are written down as part of your instructions, and you can change them whenever your business changes.

Part 4

Tools and calculations

What your AI uses to carry the work out: calculation tools, file handling and, where your plan includes them, research and specialist music methods.

Calculation tools you can check Numbers

The workflow uses calculation tools and shows the inputs and results for checking, instead of asking you to trust a number that appeared in a sentence.

Files and documents Documents

Reading what you provide, drafting what you ask for and saving the result next to the context it came from, so the source stays attached to the document.

Work you request, not work that happens on its own On request

Research and monitoring tools run when you ask for them, or on a routine you configure and start yourself in a running environment. What runs and how often is visible to you.

A private harness you control

The business information and working instructions you save stay in a private setup you control. The AI application, model providers and work services you choose may process information used for your tasks — what is shared depends on those services, their settings and the access you authorize. Read privacy and ownership.

See it in practice

Enough theory — try one task.

Create an account and put the harness to work on something you need this week.

Find your first useful task