Reply drafters
Assistants that prepare a first draft of a reply in your tone, ready for you to check and send.
Home / Services / AIPractical, not magical
Assistants that draft, summarise and find answers in your own documents, then wait for a person to say yes. Useful AI for small teams, with your judgement firmly in charge.
01 · What we build
✍️
📚
🧾
🤖
💡
Practical AI for everyday work.
No robots taking over. Just focused helpers that take the first pass at the tedious stuff, so your team can spend its energy where it counts.
Ask your documents
Question-and-answer over your own files and policies, with a pointer to where the answer came from.
Summaries
Long threads, meeting notes and reports boiled down to the bits that need action.
Back-office agents
Small AI agents that prepare routine admin, like sorting requests or drafting reminders, then hand over for approval.
"Where can AI actually help us?" sessions
Not sure where to start? We'll look at your week with you, separate the hype from the genuinely useful, and suggest one small thing worth trying first.
02 · Ask the assistant
Helpful. Then it asks.
Pick a question and watch how a well-behaved assistant works: it does the legwork, shows its sources, and waits for your OK. This is a scripted demo with made-up data, not live AI.
Try asking
AIStudio assistantScripted demo · not live AI
👈 Pick a question to start the demo
Made-up business and documents. Nothing is sent, saved or connected.
03 · Human in the loop
The loop always runs through you.
The assistant prepares. You check, correct and approve. Only then does anything leave the building. It's simple, it's safe, and it means the AI earns your trust one draft at a time.
04 · Our AI principles
Always🔐
🔍
🙋
🌱
Four promises, no hype.
Your data, your rules
We design around what you're comfortable sharing, and test with made-up examples before real records.
Show the working
Answers point to where they came from, so you can check rather than trust blindly.
Human sign-off
Drafts, not decisions. A person approves anything that goes out or changes something important.
Start small
One focused use first. If it genuinely helps, we build on it; if not, we say so.
05 · How we work
01
02
03
Start narrow. Make it count.
Find the friction.
We look at where your team's time goes and where AI could genuinely take a first pass.
Build the useful part.
A focused assistant with clear limits, tested on synthetic examples and easy to review.
Prove the difference.
We check quality and reliability in practice before deciding whether to go further.
What can't AI do?
It can sound confident and still be wrong, it doesn't know your business unless you show it, and it shouldn't make decisions about people or money on its own. That's exactly why we keep a person in the loop.
What happens to our data?
We talk this through before building anything: what data the assistant may see, where it's processed and what's kept. We prefer setups where your information stays under your control, and we test with made-up examples first.
What does it cost to run?
AI services usually charge by usage, so running costs depend on how much the assistant is used. We'll explain the moving parts in plain English and design to keep them predictable.
Do we need to change our tools?
Usually not. Assistants can often work alongside the email, documents and systems you already use.
Is AI always the answer?
No! Sometimes a simple automation or a better form does the job better. We'll tell you honestly.
More from the studio
Other ways to make room.
Bring your questions, hopes and healthy scepticism. We'll help you find the one place where AI could genuinely make room for better.
Book a call ↗