Dine4Fit · European SaaS, about 30 people, 9 functions
From individual AI chats to a company-wide AI operating model
How a European SaaS company of about 30 people gained 10% company-wide capacity in four months, enough to leave 3 of 30 planned positions unfilled.
-
+10%
company-wide capacity
-
97%
of people use AI at least 7 times a week (up from 48%)
-
3 of 30
planned positions left unfilled
In 4 months.
The challenge
A European SaaS company running a digital nutrition platform, with about 30 people across nine business functions and roughly $3M in annual revenue. Most people already used AI every day (24 of 29 in the first survey), but as a personal chat tool for writing and research. Only 11 of 29 used agents or skills, and AI wasn't connected to company data.
Much of the daily work in support, marketing, HR and development was still manual. Discount codes were generated in batches of 999, and email volumes were checked by hand at the end of every month to stay inside the tariff band. Know-how lived in people's heads, and some workflows had no clear owner, which made them hard to automate.
Management didn't want another productivity tool. They wanted an operating model where revenue grows faster than headcount.
What we did
MBI started with a three-week Discovery Assessment across all nine functions: interviews on site and remotely, 98 use cases scored for priority and feasibility by nine internal reviewers, and a roadmap of 6 initiatives and 34 use cases. A four-month transformation followed, changing people, processes and technology together:
- Weekly sessions by department and a weekly programme for internal AI champions, built around each person's actual work
- AI in every function: Claude Code and Jira review in development, reply drafts and Help Center translations in support, localisation and campaign reports in marketing, payroll inputs in HR, subscription analytics for management
- Shared skills instead of one-off prompts: a support reply skill, a Czech–Ukrainian glossary, a tone-of-voice skill for every app language and a company context index
- Scheduled tasks and agents: a morning email-volume check, 10 sales leads a day, weekly competitor and pricing reports, and an agent that summarises support tickets, suggests replies and routes bugs to Jira
- An internal MCP hub connecting Claude to nine company systems, including Slack, Jira, Confluence, Freshdesk, Gmail and Figma
- Governance from day one: personal sign-in, central control of connectors, Gmail limited to drafts, no deleting or sharing in Google Drive, and a human review of anything customers see
- Adoption tracked weekly from usage telemetry (with prompt text redacted), and impact measured in end-of-programme interviews
Ownership moved to the client's team during the programme. Their own people built 125 skills, and they have run the connectivity hub themselves since August.
The impact
- +10% company-wide capacity: 3 of 30 planned positions were left unfilled
- 86% of the people assessed now let AI do work for them, up from 43%
- Weekly prompts up 2.6x, and median sessions per person per week from 16 to 45
- Claude Code from about 100 to about 1,400 prompts a week; Jira ticket and epic review from a few hours to a few tens of minutes per sprint
- 125 skills built by the team, with weekly skill runs up 4.9x
- 100% of people active every month, against a 70% target
- Discount codes: 10,000 in one step instead of batches of 999, through a connector that replaces an API costing CZK 12,000 a month
- The email-volume automation paid for itself within about a month
Before and after, in their own numbers
Per task, from interviews with the team at the end of the programme.
- Localisation texts Before: 180 min With AI: 20 min
- Report built as an artifact Before: 180 min With AI: 20 min
- Spreadsheets and payroll inputs Before: 120 min With AI: 45 min
- Subscription analytics Before: 60 min With AI: 10 min
- Help Center article translation Before: 30 min With AI: 10 min
- Accounting file conversion Before: 20 min With AI: 0.5 min
- Draft reply to a support ticket Before: 15 min With AI: 5 min
- Daily stock check Before: 10 min With AI: 2 min
- Lead research for sales Before: 60% of working time With AI: 10% of working time
In their words
"At least five times faster. It changed my life."
"Developer before, architect now."
The service behind it: AI Transformation
Want a result like this?
30 minutes with a senior partner. No pitch deck.