Data and connections
Files, exports and anonymised data. AI and automation tools supported exploration without production connections.
From a few AI enthusiasts to teams building prototypes
A programme for the whole company, for talents and for directors, in which the client's employees built 21 AI prototypes on company data themselves. On Demo Day participants showed working prototypes to the company's management team, and selected cases went to the board.
Results
- 21
- AI prototypes built by employees
- 7
- directors' prototypes shown on Demo Day
- 40+
- ideas described by participants on canvases
How it works
Three programme levels
company, talent, directors
Ideas scored on a canvas
four dimensions
Data traffic light
what is allowed, what never
Employees build prototypes
on company data
Demo Day
prototypes shown to the management team
Directors programme ending with a Demo Day
Talent 15 people
Whole company foundational training
Sample idea
Repeatability
Impact
Risk
Speed of results
Process description
Allowed
Order export
After anonymisation
HR data
Never
including
- ERP documentation assistantprototype
- Multi-agent quotingprototype
- HR system plug-inprototype
7 directors’ prototypes on Demo Day
21 employee prototypes across the programme
Result21AI prototypes built by employees
01
Challenge
The board saw AI as the next wave of change, but the company had no systematic approach to it. Employees used various tools on their own, and the Copilot pilot covered only a small group. There was no common language and no method for assessing ideas or calculating return. The board wanted to see working solutions on company data before a larger investment.
Directors and employees worked on their own tasks. Prototypes used anonymised materials outside production systems.
02
Solution
The programme had three levels: foundational training for the whole company, a programme for 15 high-potential employees and a programme for directors ending with a Demo Day. In the hands-on programmes participants assessed ideas on a canvas and built prototypes on anonymised company data, while we ran sessions and coaching. The client's employees built the prototypes themselves, including an ERP documentation assistant, a multi-agent quoting prototype and a plug-in for the HR system.
03
How we built it
Participants worked with the tools available in the company and in their publicly available versions: Claude, ChatGPT and Copilot, and the more advanced ones also with n8n, Make and Python with model APIs. Before use in public tools, company data went through a traffic light rule: what is allowed, what only after anonymisation, and what never. Prototypes worked on files and data exports, with no integration with production systems. Instructions limited hallucinations: answers only from documents, internet only with the user's consent. Each idea was scored on four dimensions: repeatability, impact, risk and speed of results, and before deployment we recommended security testing.
ClaudeChatGPTMicrosoft Copilotn8nMakePython
Workflow
How the solution works
- 01
Task and baseline
The team chooses repetitive work and describes the intended improvement.
- 02
Prototype with working data
It tests the idea on files and exports without writing to production systems.
- 03
Demo Day and decision
Participants show working prototypes to the management team, with their assumptions and limitations. Selected cases go to the board, which chooses the next pilots.
The human role
Directors and teams assess usefulness. Leadership decides on further investment; a prototype is not a production integration.
Lessons
What made the difference.
- 01
A prototype instead of a presentation
Directors build a working solution on company data and show it to the management team on Demo Day. Selected cases go to the board before the budget decision is made.
- 02
Assessing an idea on a canvas
Every idea goes through four questions: repeatability, impact, risk and speed of results, with a return estimate prepared by its author.
- 03
Data traffic light
A simple rule says which data may be used in public tools, which only after anonymisation, and which never.
PrototypingIdea assessmentAI adoption
Company of the Future
How this builds the Company of the Future
A project delivers results on its own, but it also leaves organized context behind in the company. Every next AI solution builds on it. This project added to the AI Engine:
- Human+AI Teams
A shared language for AI across the company: foundational training for everyone, a programme for 15 high-potential employees and one for directors.
- Decisions
A method for scoring ideas on four dimensions and a traffic light rule for data.
- Skills
The first skills built by participants, e.g. a competency matrix for HR, comparing supplier quotes and tender search. They were built on participants' individual accounts.
Value
Where value is created
Time saved
Prototypes show which repetitive tasks merit a pilot.
New possibilities
More than 40 ideas described on canvases let the team assess a use case before investing in production integration. Participants rated the programme 8.7/10 on average in surveys, and the second talent session scored an NPS of +71.
What comes next
The results provide a basis for selecting pilots before connecting solutions to company systems. After the programme the company launched its own AI environment, a second phase started with domain training for 66 people, an internal quoting team was set up, and the board announced a competition for the best AI implementation.
Most of our projects are covered by NDAs, so we only share part of them, without client names or details, as inspiration. We are happy to tell you more about our experience in a conversation.
Let’s talk about the work you want to improve.
Tell us how a process works today. Together, we will explore where AI Engine can help and which module to start with.
Book a callOr write to: hello@wearefuture.ai
