May we count your visit?
We use Google Analytics to see which pages are read. It sets cookies, so we ask first. The site works the same either way. Privacy policy
Assistants that work from your own content, and step aside when a person is what it takes.
What it is
In the way: the same questions are answered by hand, the answers are spread over guides, inboxes and people's heads, and a general chatbot guesses.
What we do: we connect an assistant to your own guides, prices and systems. It answers from those and says where it found the answer. It does not guess, and it hands the whole conversation to a person when that is what it takes.
What it covers
It answers from your own guides and prices, in the language the customer wrote in, and links to where it found the answer.
Here: the assistant on ptpro.no.
People search in their own words. The search finds what was meant, across everything written or said.
Here: Helsekoden, searching what was said in the episodes.
A time that is free in the team's calendars, picked and booked in a few taps, with the meeting link sent at once.
Here: booking a demo with PT PRO.
Behind the chat
The assistant on ptpro.no gets three kinds of question: how to do something in the app, a wish to see it, and something only a person can fix.
Below: what it does with each of them, step by step.
Asks how to do something
Chat on ptpro.no
How do I connect Stripe so my clients can pay by card?
The reply
Open Payments in the app and choose Connect Stripe. Stripe asks for your business details and sends you back when it is done.
Every step is in the guide: Connect Stripe
The assistant on ptpro.no
Understood
Searched by meaning, not by word
Read the guide
The help centre
Every guide and question, in three languages
In the same chat, in the language they wrote in, with a link to the guide.
Wants a demo
Chat on ptpro.no
Can someone show me how this works for a studio with several trainers?
Who should we ask for?
Name
Send
Pick a time for the demo
09:30 11:00 14:00
Book
The reply
You are booked. The meeting link is on its way to your e-mail.
The assistant on ptpro.no
Understood
A short form, in the chat
Checked the team's calendars
Leads
Everyone who has asked, for the team to follow up
The team's calendars
When each colleague is free, and who had the last demo
A booking card in the chat. Nobody writes back and forth to find a time.
Needs a person
Chat on ptpro.no
I think I was charged for two subscriptions this month. Can you fix it?
A person from the team is joining
In the same chat
Hi, I am on the PT PRO team. I have read the conversation and I am looking at it now.
The assistant on ptpro.no
Understood
Handed to the team
Answered from the admin
The team's inbox
Chats, leads and requests in one place
The customer does not have to explain it twice.
MCP
In the way: a system that does a lot has a screen for everything, and most of the team only knows a few of them.
What we do: everything the admin can do is also a tool for the AI assistant your team already uses. They ask in their own words. Anything that changes something is shown first, done only after a yes, and logged in their name.
Below: three things PT PRO's team asks for, and what the server does with each.
Translate a guide
PT PRO's team, in their own AI assistant
Translate the new guide to English and Swedish.
Back in the assistant
Both drafts are ready to read. Nothing is published until you approve them.
The staff server on ptpro.no
Signed in as themselves
Two translations drafted
Shown before anything is public
Roles and permissions
What each person may do in the admin
The help centre
Every guide and question, in three languages
The log
Every change, and who made it
Logged in their name, like everything else they change.
Who is waiting?
PT PRO's team, in their own AI assistant
Who is waiting for an answer?
Back in the assistant
A list of the people who are waiting, with what each of them asked for and a link to answer from.
The staff server on ptpro.no
Signed in as themselves
Looked across the inboxes
Roles and permissions
What each person may do in the admin
The team's inbox
Chats, leads and requests in one place
Only a read. Nothing was changed.
What do people search for?
PT PRO's team, in their own AI assistant
What did people search for this week?
Back in the assistant
The searches that found nothing, the most asked first. That is the next guide to write.
The staff server on ptpro.no
Signed in as themselves
Read the help centre's searches
Roles and permissions
What each person may do in the admin
The help centre's searches
What visitors typed, and whether it found anything
The team can ask it to draft that guide next.
Things staff ask their assistant
Follow-ups, meetings and the people who showed interest since yesterday.
A list of what is missing, instead of sixty screens to click through.
New episodes come in and are queued to be written out.
Speech to text
In the way: hours of talk that nobody can look anything up in.
What we do: the sound is sent to a transcription service, which writes it out and tells the voices apart. A person says who is who. Then the text is cleaned up, cut into passages and turned into vectors, so a search goes by what was meant and not by which words were used.
Below: one episode of Helsekoden on its way from the podcast feed to a search, and what a search does.
A new episode
The podcast feed, to the team
New episode: Sleep, evenings and small habits.
Who is speaking?
Voice A
Voice B
Ready
To the team
The episode is ready. Listeners can now search everything that was said in it.
Helsekoden's platform
Sent to be written out
The ads are told from the talk
A person says who is who
Cleaned up by fixed rules
Cut into passages, and embedded
AssemblyAI
Turns speech into text, and tells the voices apart
A language model
Reads the opening minutes and marks what is an ad
The clean-up rules
What the transcription keeps getting wrong, and what it should say
OpenAI embeddings
Turns a passage into a vector: numbers that stand for what it means
Naming the voices was the one thing a person did.
A listener searches
Search on helsekoden.no
What did they say about magnesium in the evening?
The answer
The guest, in “Sleep, evenings and small habits”: “I hear about magnesium all the time too. For me the routine matters more.”
Play from 18:40
Helsekoden's platform
The question becomes a vector too
The nearest passages are found
An answer, from what was said
OpenAI embeddings
Turns a passage into a vector: numbers that stand for what it means
The vector index
Every passage of every episode, in the platform's own database
A language model
Writes a short answer from the passages it is given
Every result opens the episode at the minute it was said.
A site that sells the app, and support that answers before a person has to.
With a school since it started in 2005. Today that is the website, and the system that keeps track of everyone who gets in touch.
A podcast you could only listen to became a place its listeners use.
Every muscle, bone and exercise AFPT teaches, in one library and three languages.
A site that says what you do, answers what people ask, and takes them all the way to a sign-up or a sale.
The system the business runs on: courses, a podcast, a library, a CRM. Built for how you work, and for the people who use it every day.
A name on the door: the logo, the colours, the type, and how it all looks on a page, a shirt and a car.
Tell us what you want to build, and we will work it out together.