Case · Knowledge
Internal documentation as an active AI assistant.
Systems that turn internal documentation into an active AI assistant. Eliminating repetitive HR questions.
Model case: we assume the company looks like this and show how much it would yield.
- Model case
- OpenAI
- Pinecone
- Slack
Starting point
An information black hole
Employees spend hours looking for information buried in Drive folders or an outdated Notion. HR keeps answering the same questions over and over. Knowledge gets lost and onboarding new hires drags on for weeks.
- Before
A frustrated team and lost data
MANUAL SEARCH
- After
Instant answers with a source
AI ASSISTANT
What we would build
System architecture
From documents to answers.
- 01
Knowledge sources
PDF policies, Notion, GDrive
- 02
Text preparation
Converting the text into a form the AI can search (OpenAI)
- 03
Retrieval
Search by meaning in the database
- 04
Answer
The AI composes the answer
- OpenAI · Text preparation and chat
- Supabase · Database for search by meaning
- Flowise · Connecting the steps
- Vercel · Hosting
Rollout steps
How the rollout would go.
- 01
WEEK 01
Collecting and cleaning data
Connecting to Google Drive and Notion and preparing the text for further processing.
- 02
WEEK 02
Preparing the search
Splitting the text into chunks and storing them in a database for searching.
- 03
WEEK 04
Launching the bot
Connecting to Slack/Teams and preparing team training.
Model assumptions
Where the result comes from.
We take a typical company from the sector and the situation from the brief, add labelled assumptions and calculate the result. If the inputs change, so does the result, so you can recalculate it for your own company.
Model company
A model company where employees search for information in Drive folders and an outdated Notion, and HR answers the same questions over and over.
Inputs
- HR questions per month (assumption)
- 200
- Share of questions the assistant resolves (assumption)
- 65 %
- HR time per answer (assumption)
- 10 min
- Labor rate (assumption)
- 24 EUR / h
Calculation step by step
- 01
Questions resolved by the assistant per month
200 × 65 %= 130
- 02
HR time saved
130 × 10 min= 1 300 min (21.7 h)
- 03
Value of the saved time per month
1 300 min ÷ 60 × 24 EUR= 520 EUR
This is a gross saving of time and money. We include the cost of rollout and running only where it is visible in the calculation.
Model results
What would change after rollout.
- Model case
−130 / month
200 questions × 65 % that the assistant resolves, with a source
HR questions
- Model case
−21.7 h / month
130 × 10 min, 520 EUR a month
HR time
Your numbers belong in the model, not ours. The calculator shows where to start, and we work out the real values on your data.
Start here
15 minutes is enough for you to know whether it makes sense.
You tell us what is bothering you. We tell you whether we can help, roughly what it costs and where I would start in your place. No presentation.
- jakub@transformuj.ai
- Phone
- +420 773 070 090
- Address
- Legerova 1820/39, Praha 2