While we warmly invite everyone to join the meetup, please note that participation in the workshop is limited to one representative per company.
We kindly inform you that while everyone is welcome to attend the main meetup, workshop seating is limited. Participation in the workshop is restricted to the first 30 registrants on a first-come, first-served basis.
The Ignite Agentic Awareness Workshop is practical workshop that explores how AI Agents are redefining enterprise automation, requiring no coding skills and catering to both business users and developers. Participants will engage in hands-on labs to build a Customer Support Agent capable of classifying messages, detecting sentiment, and escalating cases. By the end of the session, attendees will gain a clear understanding of real-world AI applications and the expertise needed to scale automation within their organizations.
Alexey Chistyakov
Principal Sales Engineer
Opening the event with a welcome from the host, an overview of the agenda, and key goals of the session.
Toma Vevelstad
Managing Partner
The AI bubble makes it hard to tell a real result from a good demo. So instead of predictions, this talk gives you evidence in three implementations: everyday communication, recruitment, and legal work – described as they really went. What we built, what it changed, what it cost us in effort, what we learned, and which parts we’d warn you about before you start.
Dagmara Sysuła
Delivery Manager
Vi har mange store endringsprosesser i offentlig sektor, og med dagens fremadstormende teknologi har nyere forskning vist at 50-80% av alle endringsprosjekter og prosesser, så er brukerne misfornøyde som ett resultat at prosessene ikke møter ti forventingene enten på pris, tidsestimat eller forventningene om forbedringer og forenklinger. Nyere forskning fra Tolletaten og fra andre etater i offentlig sektor viser at RPA prosjekter og RPA teknologiske endringsprosesser skiller seg ut fra denne statistikken. Jeg viser da til kvantitativ og kvalitativ forskning på dette.
EN
A quantitative and qualitative study on the effects of robotization and automation in the Norwegian Customs Administration (Tolletaten) and in the public sector
We have many major change processes underway in the public sector, and given today’s rapidly advancing technology, recent research has shown that in 50-80% of all change projects and processes, users end up dissatisfied because the processes fail to meet expectations—whether in terms of cost, timeline, or the anticipated improvements and simplifications. However, recent research from the Norwegian Customs Service and other public sector agencies shows that RPA projects and RPA-driven technological change processes stand out from this trend. Here I am referring to both quantitative and qualitative research on this topic.
Cecilie Rochstad
Team leader RPA
Most organizations know UiPath for automation. Few realize it’s also transforming software testing with AI-powered test generation and intelligent automation.
Join us to see how UiPath Test Cloud helps teams accelerate quality, reduce manual effort, and increase confidence in every release through Agentic Testing, where AI agents can discover, create, execute, and optimize tests autonomously.
Kim Knudsen
Sales Manager
Christiania Qvartalet is a modern meeting centre in the heart of Kvadraturen, Oslo’s historic city centre district.
The conference sessions will take place in the venue’s recently renovated auditorium. It is bright and fully equipped for our presentations and workshops.
After the conference, we will move on to the Networking Part at NoName Sentrum Bar. Over snacks and drinks, participants will have time to talk, connect and exchange ideas in a relaxed atmosphere.
This meetup is for everyone passionate about automation, AI, RPA, Process Intelligence and Management.
Who might sit next to you? Here’s a couple of guesses:
Connect with like-minded people, gain practical insights from real automation projects and see how others are tackling challenges in RPA, AI and process improvement. No strings attached – just a relaxed space to exchange ideas and expand your perspective.
Get inspirations and ideas from our automation newsroom or just enjoy your morning coffee.
Amer Sports: an AI agent that answers “where is my invoice?” – and deliberately cannot see the amount
Amer Sports is home to Salomon, Arc’teryx, Wilson, Atomic, Peak Performance and Suunto. Its Kraków team has shipped over 200 automations. Their first global AI agent in production is MIA – My Invoice Assistant – a Teams chatbot that answers the single most common question accounts payable gets: has this invoice been paid?
The interesting part is what MIA is not allowed to do. It never touches SAP: it queries a read-only Power BI data model containing only fields that are safe to share. Amounts, approvers and cost centres were deliberately stripped out after review with legal, audit and cyber security – because individually harmless fields become a toxic combination when joined. Intent recognition is the only job given to AI; every data lookup runs on classical logic, which removes the room to hallucinate an invoice that doesn’t exist. Answers now arrive in seconds, not in a ticket queue.
Our role here was scoping and architecture: deciding what the agent must not be able to see, and keeping AI to intent recognition only.
First global AI agent in production ⬥ Teams-based, company-wide ⬥ Read-only data, never SAP ⬥ 7-year partnership
Amer Sports is home to Salomon, Arc’teryx, Wilson, Atomic, Peak Performance and Suunto. Its Kraków team has shipped over 200 automations. Their first global AI agent in production is MIA – My Invoice Assistant – a Teams chatbot that answers the single most common question accounts payable gets: has this invoice been paid?
The interesting part is what MIA is not allowed to do. It never touches SAP: it queries a read-only Power BI data model containing only fields that are safe to share. Amounts, approvers and cost centres were deliberately stripped out after review with legal, audit and cyber security – because individually harmless fields become a toxic combination when joined. Intent recognition is the only job given to AI; every data lookup runs on classical logic, which removes the room to hallucinate an invoice that doesn’t exist. Answers now arrive in seconds, not in a ticket queue.
Our role here was scoping and architecture: deciding what the agent must not be able to see, and keeping AI to intent recognition only.
First global AI agent in production ⬥ Teams-based, company-wide ⬥ Read-only data, never SAP ⬥ 7-year partnership
Inelo: 13 robots, 1M+ driver settlements – and employees asking “what can we automate next?”
Inelo, now part of the Eurowag Group, settles driver working time for transport companies across Europe – millions of activity records every month, with legal compliance riding on every calculation. Process Director Paulina Wołowiec didn’t start with technology. She started by measuring real working time, mapping every step, and costing the most repetitive tasks. Only then came the robots.
Today 13 UiPath robots run driver settlement end to end, 24/7, paying back within 5 to 12 months and generating around €44,000 of benefit every month. The programme is now expanding across the wider Eurowag Group. The part that doesn’t fit in a metric: teams named the bots Robert and Samurai, and now bring automation ideas forward themselves.
Paulina Wołowiec put it plainly: UiPath “didn’t just give us the platform – they connected us with Office Samurai, the best local partner who could guide us and support us from day one.”
13 robots running 24/7 ⬥ 1M+ driver records settled ⬥ ~€44,000 monthly benefit ⬥ ROI in 5–12 months
Inelo, now part of the Eurowag Group, settles driver working time for transport companies across Europe – millions of activity records every month, with legal compliance riding on every calculation. Process Director Paulina Wołowiec didn’t start with technology. She started by measuring real working time, mapping every step, and costing the most repetitive tasks. Only then came the robots.
Today 13 UiPath robots run driver settlement end to end, 24/7, paying back within 5 to 12 months and generating around €44,000 of benefit every month. The programme is now expanding across the wider Eurowag Group. The part that doesn’t fit in a metric: teams named the bots Robert and Samurai, and now bring automation ideas forward themselves.
Paulina Wołowiec put it plainly: UiPath “didn’t just give us the platform – they connected us with Office Samurai, the best local partner who could guide us and support us from day one.”
13 robots running 24/7 ⬥ 1M+ driver records settled ⬥ ~€44,000 monthly benefit ⬥ ROI in 5–12 months
Ignitis: 45 robots, 355 processes – and now agents that read 1,500 legal documents a month
Ignitis Group, the largest listed energy company in the Baltics, built its first robot in 2017. Today 45 robots run 355 processes across two Centres of Excellence, freeing up the equivalent of 200 people’s time. That maturity is exactly why agents were the logical next step – not a leap of faith.
The proof case was legal monitoring. One specialist used to spend two to three hours a day scanning two public websites for legislation across 26 areas of law. Now a robot collects the documents, an agent reads and classifies them, and a second robot routes each one to the people subscribed to that legal area – 1,500 documents a month at around 93% accuracy, giving the legal team back 60 hours monthly. Deliberately proven on public data first, before agents go anywhere near confidential material.
We guided Ignitis from the first agentic PoC to production and trained their CoE teams on Document Understanding, agent building and Agentic Automation, so the next five agents get built in-house.
45 robots ⬥ ≈200 FTE of capacity freed ⬥ ~93% agent accuracy ⬥ 60 h/month back to legal
Ignitis Group, the largest listed energy company in the Baltics, built its first robot in 2017. Today 45 robots run 355 processes across two Centres of Excellence, freeing up the equivalent of 200 people’s time. That maturity is exactly why agents were the logical next step – not a leap of faith.
The proof case was legal monitoring. One specialist used to spend two to three hours a day scanning two public websites for legislation across 26 areas of law. Now a robot collects the documents, an agent reads and classifies them, and a second robot routes each one to the people subscribed to that legal area – 1,500 documents a month at around 93% accuracy, giving the legal team back 60 hours monthly. Deliberately proven on public data first, before agents go anywhere near confidential material.
We guided Ignitis from the first agentic PoC to production and trained their CoE teams on Document Understanding, agent building and Agentic Automation, so the next five agents get built in-house.
45 robots ⬥ ≈200 FTE of capacity freed ⬥ ~93% agent accuracy ⬥ 60 h/month back to legal
1,000 invoices per person: the monthly average that hides how work really happens
“We process 10,000 invoices a month, we have 10 people on the team, so that’s a clean 1,000 invoices per person.” Andrzej Kinastowski calls the monthly average the most dangerous lie in modern management, because it ignores the real complexity of the work. At Vesuvius, Elena’s team sat at their desks for 12 or more hours a day during month-end closing. The same averages also miss an army of high-performing introverts who keep their heads down, crush their work and sometimes stay past midnight, without ever talking up what they do.
Often the problem isn’t the people, it’s the tools. At one insurance company the system lagged so badly that calls fell into awkward, endless silences. To cope, call centre agents typed customer details into electronic sticky notes during the call and copied them into the official system afterwards. Elsewhere, one person worked in six SAP instances at the same time, and employees had to switch between more than 120 different applications. On top of that come hour-long status meetings with ten people, which add up to ten hours of stolen productivity, and email chains that should have been a two-minute phone call.
Process Intelligence shows what actually happens, not what the flowchart says should happen. After Vesuvius rolled out KYP.ai, it cut its workforce by 8% purely through natural attrition and broke even on the software in six months. The same data makes incentive programmes transparent, instead of leaving them to office politics or whoever talks the loudest. The goal isn’t to squeeze people harder: “We hire humans because of their intellect, their creativity, and their ability to solve complex problems. It is time we stop forcing them to act like bots.”
10,000 invoices, 10 people ◆ 120+ applications ◆ 8% leaner through natural attrition ◆ Break-even in 6 months ◆
“We process 10,000 invoices a month, we have 10 people on the team, so that’s a clean 1,000 invoices per person.” Andrzej Kinastowski calls the monthly average the most dangerous lie in modern management, because it ignores the real complexity of the work. At Vesuvius, Elena’s team sat at their desks for 12 or more hours a day during month-end closing. The same averages also miss an army of high-performing introverts who keep their heads down, crush their work and sometimes stay past midnight, without ever talking up what they do.
Often the problem isn’t the people, it’s the tools. At one insurance company the system lagged so badly that calls fell into awkward, endless silences. To cope, call centre agents typed customer details into electronic sticky notes during the call and copied them into the official system afterwards. Elsewhere, one person worked in six SAP instances at the same time, and employees had to switch between more than 120 different applications. On top of that come hour-long status meetings with ten people, which add up to ten hours of stolen productivity, and email chains that should have been a two-minute phone call.
Process Intelligence shows what actually happens, not what the flowchart says should happen. After Vesuvius rolled out KYP.ai, it cut its workforce by 8% purely through natural attrition and broke even on the software in six months. The same data makes incentive programmes transparent, instead of leaving them to office politics or whoever talks the loudest. The goal isn’t to squeeze people harder: “We hire humans because of their intellect, their creativity, and their ability to solve complex problems. It is time we stop forcing them to act like bots.”
10,000 invoices, 10 people ◆ 120+ applications ◆ 8% leaner through natural attrition ◆ Break-even in 6 months ◆
12 ERP systems, one employee: what task mining uncovered at Vesuvius
At Vesuvius, a global leader in molten steel products, one employee worked in 11 to 12 ERP instances every day. They all had the same names but different configurations, versions and layouts, and each one needed its own password. That’s what happens in a company that keeps buying businesses around the world, each with its own ERP. Elena Olichowska-Szydłowska, Automation Excellence Manager at Vesuvius, found this after she brought in task mining with KYP.ai instead of running Lean workshops with sticky notes.
It wasn’t the only thing the data showed. Record-to-Report teams were heading for burnout around month-end and year-end closes. In the same team, some people worked overtime while others had spare capacity, and monthly averages hid the difference. The data also showed that work could be shared out by how complex tasks really are, not just by how many there are.
To get there, Elena had to earn people’s trust first. Employees were afraid of being watched. So she started with a 13-person pilot, met each person one-on-one to explain what data was collected and who could see it, and presented the tool as a way to spot where people need support. Her one regret is not giving employees personal dashboards from day one. The investment paid for itself in 6 months. Vesuvius also cut its workforce by 8% without layoffs, by not replacing people who left.
11–12 ERP instances for one employee ◆ 13-person pilot ◆ Paid back in 6 months ◆ 8% smaller team, no layoffs ◆
At Vesuvius, a global leader in molten steel products, one employee worked in 11 to 12 ERP instances every day. They all had the same names but different configurations, versions and layouts, and each one needed its own password. That’s what happens in a company that keeps buying businesses around the world, each with its own ERP. Elena Olichowska-Szydłowska, Automation Excellence Manager at Vesuvius, found this after she brought in task mining with KYP.ai instead of running Lean workshops with sticky notes.
It wasn’t the only thing the data showed. Record-to-Report teams were heading for burnout around month-end and year-end closes. In the same team, some people worked overtime while others had spare capacity, and monthly averages hid the difference. The data also showed that work could be shared out by how complex tasks really are, not just by how many there are.
To get there, Elena had to earn people’s trust first. Employees were afraid of being watched. So she started with a 13-person pilot, met each person one-on-one to explain what data was collected and who could see it, and presented the tool as a way to spot where people need support. Her one regret is not giving employees personal dashboards from day one. The investment paid for itself in 6 months. Vesuvius also cut its workforce by 8% without layoffs, by not replacing people who left.
11–12 ERP instances for one employee ◆ 13-person pilot ◆ Paid back in 6 months ◆ 8% smaller team, no layoffs ◆
The Visio map vs. a busy Tuesday afternoon: what your reports don’t show
Corporate reports show results, the “what”, but not how the work actually gets done. That leaves a Visibility Gap between the clean process map drawn in Visio and the messy reality of a busy Tuesday afternoon. Most organisations fill that gap with assumptions, which the article calls “Corporate Fiction”, and end up making operational decisions on intuition rather than evidence.
KYP.ai closes the gap with task mining. Process mining reads structured logs from ERP and CRM systems. Task mining is “the camera inside the delivery truck”: it captures the work done on the desktop, in Excel, email and legacy apps. That’s where the differences show up. In a transactional setting, someone on the lower end hits around 17,000 actions a day, while a top performer reaches about 35,000.
The platform works in three steps: see, understand, change. It’s automation-agnostic, so it diagnoses the problem before recommending a fix. The fix might be RPA, a training programme, a change to the process, or simply better software. Privacy controls like app whitelisting and PII masking on the desktop are built in. Allied Global saw a 25% increase in active productive FTE. Hollard Insurance got a 20% productivity leap and saved 307 hours of manual work a month.
35,000 vs 17,000 daily actions ◆ +25% active productive FTE at Allied Global ◆ 307 hours saved a month at Hollard ◆
Corporate reports show results, the “what”, but not how the work actually gets done. That leaves a Visibility Gap between the clean process map drawn in Visio and the messy reality of a busy Tuesday afternoon. Most organisations fill that gap with assumptions, which the article calls “Corporate Fiction”, and end up making operational decisions on intuition rather than evidence.
KYP.ai closes the gap with task mining. Process mining reads structured logs from ERP and CRM systems. Task mining is “the camera inside the delivery truck”: it captures the work done on the desktop, in Excel, email and legacy apps. That’s where the differences show up. In a transactional setting, someone on the lower end hits around 17,000 actions a day, while a top performer reaches about 35,000.
The platform works in three steps: see, understand, change. It’s automation-agnostic, so it diagnoses the problem before recommending a fix. The fix might be RPA, a training programme, a change to the process, or simply better software. Privacy controls like app whitelisting and PII masking on the desktop are built in. Allied Global saw a 25% increase in active productive FTE. Hollard Insurance got a 20% productivity leap and saved 307 hours of manual work a month.
35,000 vs 17,000 daily actions ◆ +25% active productive FTE at Allied Global ◆ 307 hours saved a month at Hollard ◆
SPS: 50% less admin work, 25% more productive capacity – across 20+ countries
SPS runs multi-client BPO operations for banking, insurance and health from Zurich, with over 8,500 employees across more than 20 countries. Nothing was broken – governance and delivery were already mature. What was missing was real-time visibility into how work actually ran across regions, and a clean path from that data into billing.
KYP.ai supplied the intelligence layer: process signals captured from existing workflows, an Integration Database consolidating data from multiple systems, and validated data flowing straight into SAP via IDoc. Administrative alignment work dropped by more than half, productive capacity rose 25%, and two separate systems collapsed into one.
As implementation partner we ran the user needs analysis and process mapping, configured KYP.ai against how SPS actually works, and onboarded the teams as certified KYP.ai Platform Evangelists.
50%+ less admin work ⬥ 25% more productive capacity ⬥ 13 pp more time on client work ⬥ 2 systems → 1
SPS runs multi-client BPO operations for banking, insurance and health from Zurich, with over 8,500 employees across more than 20 countries. Nothing was broken – governance and delivery were already mature. What was missing was real-time visibility into how work actually ran across regions, and a clean path from that data into billing.
KYP.ai supplied the intelligence layer: process signals captured from existing workflows, an Integration Database consolidating data from multiple systems, and validated data flowing straight into SAP via IDoc. Administrative alignment work dropped by more than half, productive capacity rose 25%, and two separate systems collapsed into one.
As implementation partner we ran the user needs analysis and process mapping, configured KYP.ai against how SPS actually works, and onboarded the teams as certified KYP.ai Platform Evangelists.
50%+ less admin work ⬥ 25% more productive capacity ⬥ 13 pp more time on client work ⬥ 2 systems → 1
35,000 vs 17,000 actions a day: the top performers your reports don’t see
In transactional operations, a low performer completes up to 17,000 actions a day. A top performer completes 35,000. Standard reports don’t show that gap, and they don’t show who carries the load. Adam Bujak, CEO of KYP.ai, has a name for many of them: an “army of high-performing introverts” who support several teams, never talk up what they contribute, and end up working after midnight until they burn out.
The same blind spot drives bad automation decisions. In his Capgemini days, Bujak watched finance teams put “a few millions” into RPA and get back about 80k in savings, even with 280 business analysts hunting for use cases. When a process runs in countless variants and the licence costs too much, a bot only automates the chaos.
KYP.ai was built to close that gap: see what’s happening, understand the details, then change things to get the outcome. Continuous task mining instead of one-off workshops, and privacy-first data capture that collects only what the goal needs. The result is evidence about people, processes and automation that stands up in front of a CFO. One KYP.ai customer recently reached an ROI of 4,000%.
35,000 vs 17,000 daily actions ◆ “A few millions” in RPA, ~80k saved ◆ 4,000% ROI at one KYP.ai customer ◆
In transactional operations, a low performer completes up to 17,000 actions a day. A top performer completes 35,000. Standard reports don’t show that gap, and they don’t show who carries the load. Adam Bujak, CEO of KYP.ai, has a name for many of them: an “army of high-performing introverts” who support several teams, never talk up what they contribute, and end up working after midnight until they burn out.
The same blind spot drives bad automation decisions. In his Capgemini days, Bujak watched finance teams put “a few millions” into RPA and get back about 80k in savings, even with 280 business analysts hunting for use cases. When a process runs in countless variants and the licence costs too much, a bot only automates the chaos.
KYP.ai was built to close that gap: see what’s happening, understand the details, then change things to get the outcome. Continuous task mining instead of one-off workshops, and privacy-first data capture that collects only what the goal needs. The result is evidence about people, processes and automation that stands up in front of a CFO. One KYP.ai customer recently reached an ROI of 4,000%.
35,000 vs 17,000 daily actions ◆ “A few millions” in RPA, ~80k saved ◆ 4,000% ROI at one KYP.ai customer ◆
AI Automation Dojo by Office Samurai
Tired of processes so old they remember when “the cloud” was just a weather phenomenon? The Office Samurai Podcast gets to the heart of modern business, showing that AI and automation are more than just buzzwords your boss likes to throw around. We explore practical ways GenAI, RPA, and intelligent automation can truly transform business processes.
Tired of processes so old they remember when “the cloud” was just a weather phenomenon? The Office Samurai Podcast gets to the heart of modern business, showing that AI and automation are more than just buzzwords your boss likes to throw around. We explore practical ways GenAI, RPA, and intelligent automation can truly transform business processes.
