> ## Documentation Index
> Fetch the complete documentation index at: https://docs.wednesdayai.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Personas

> Give your AI a professional identity, connect it to the tools that role uses, tune how it runs, and extend it when something is missing.

# Personas

A **persona** is your AI assistant playing a dedicated role — a life coach, an IT expert, a marketing consultant, a teacher. Instead of a general-purpose chatbot, you get an assistant with a consistent character, its own instructions, and (over time) the tools and habits of that profession.

Setting up a persona happens in stages, and each stage is a guided conversation with your AI helper. This page explains what the three follow-up workflows do for you, in the order you'd normally use them:

1. [Persona toolkit](#persona-toolkit) — connect the persona to the real-world tools its role uses
2. [Optimizing a persona](#optimizing-a-persona) — a check-up of how the persona runs day to day
3. [Extending a persona](#extending-a-persona) — plan new abilities when settings alone can't do it

<Note>
  These workflows assume a persona already exists — someone has already run the persona
  builder, which writes the persona's character files (things like its soul, identity,
  guardrails, and tool notes). If you are starting from scratch, ask your administrator
  about creating a persona first.
</Note>

## Persona toolkit

Professionals use specific apps and services — a marketer lives in ad platforms and analytics dashboards, an IT expert in ticketing systems and server dashboards. The **persona toolkit** workflow finds those tools for your persona's role and works out how the AI can use each one.

What happens when you ask for it:

1. Your AI helper reads the persona's character files and works out the role — "this is an IT support persona".
2. It researches roughly the ten most-used tools for that job, worldwide or in your region if you say where you are.
3. For each tool, it checks whether WednesdayAI can connect to it — some tools offer direct connections, some have public interfaces, some only work through manual steps.
4. For every tool that can connect, it writes a small **skill** — a step-by-step recipe stored with the persona, so the AI knows how to use that tool when you ask.
5. It records everything in the persona's tool notes, so the persona knows what it can reach and how.

**What you get:** a curated list of tools for the role, ready-made skills for the ones that connect, and an honest note about any tool that has no automatic path (you'd keep using that one manually).

<Warning>
  The toolkit writes the recipes — it does not switch anything on. Real connections (logins,
  permissions, API keys) are set up afterwards with your administrator.
</Warning>

## Optimizing a persona

A persona that works well needs more than good character files — it needs good running habits. **Optimizing a persona** is a guided check-up, one topic at a time, with your AI helper asking questions and making recommendations.

The check-up reviews things like:

* **Check-in rhythm** — how often the persona looks at its own to-do list and pings you when something needs attention
* **Conversation freshness** — when an old conversation should start clean so the persona doesn't drag outdated details around
* **What it remembers** — which notes it keeps between conversations and which it should write down sooner
* **Its instructions** — whether its startup instructions are doing their job without bloating every reply

**What you get:** a plain-language report saved in the persona's folder, listing recommended changes with the reasons behind them.

<Tip>
  Nothing is switched on automatically. Every recommendation is a suggestion — you (or your
  administrator) decide what to apply. If the check-up finds something settings cannot fix,
  it will say so and point you to the next workflow below.
</Tip>

## Extending a persona

Sometimes a persona needs an ability WednesdayAI simply doesn't have built in — for example, quietly reflecting in the background between your conversations, noticing when you've gone quiet, or saving a summary every time a conversation wraps up. That's not a settings problem; it needs new building blocks.

**Extending a persona** turns those gaps into build plans:

1. Your AI helper reviews the persona and lists the missing abilities, prioritised for the role.
2. For each gap you want filled, it checks exactly where the system could support a new piece — verifying against the real software, not guessing.
3. It writes a precise build brief: what to build, where it plugs in, and what rules it must follow.

**What you get:** a hand-off document for a developer (or a separate build session with your AI helper). The brief is self-contained, so the person building it doesn't need to re-ask you everything.

<Note>
  This workflow plans — it doesn't build. Nothing about your running persona changes until
  the build is actually done and installed, usually by your administrator.
</Note>

## Troubleshooting

**The persona doesn't seem to use a tool the toolkit set up** — the recipe exists, but the connection behind it may not be switched on yet. Check with your administrator that the tool's login or access was set up.

**Recommendations from the check-up didn't change anything** — that's by design. The report recommends; someone still has to apply the changes. Ask your administrator to walk the report with you.

**A build brief was written weeks ago and nothing happened** — build briefs wait for a build session. Ping your administrator to schedule one, or ask your AI helper to re-check the brief against the current version first.

## Related

* [What is WednesdayAI?](/users/what-is-wednesdayai) — the basics, in plain English
* [Memory](/users/memory) — what the AI remembers between conversations
* [Slash commands](/users/commands) — controlling the AI in chat