AI Automation Dojo by Office Samurai

AI Automation Dojo by Office Samurai

By Office SamuraiTechnology
Download on the App Store

AI Automation Dojo by Office Samurai episodes

  • How to know when you need an AI agent orchestration platform

    When does a company need an AI agent orchestration platform? 3 signs, and how to read Gartner's report on BOAT platforms. #AIAgents #AIOrchestration #Automation

    A large international company started building AI agents on different tools and clouds. An orchestration platform – a system that manages all AI agents, RPA robots, systems and people in one process – seemed too expensive compared with simply paying for tokens (the usage fees for AI models). A year later, the company had a dozen AI agents, and nobody knew exactly how many there were, who owned them or what they cost.

    In this episode of the AI Automation Dojo, Andrzej Kinastowski, co-founder of Office Samurai, explains when paying only for tokens stops being the cheaper option and when a company needs an orchestration platform. He also walks through the 2026 Gartner Magic Quadrant for Business Orchestration and Automation Technologies (BOAT): how to read it and what to check with vendors before you choose a platform.

    What you'll learn in this episode:

    🔸 Why the cost of AI tokens doesn't show the real cost of managing a dozen AI agents

    🔸 What process and AI agent orchestration is, and what companies expect from an orchestration platform

    🔸 Why Andrzej disagrees with Gartner that RPA is optional and "computer use" (AI that clicks through screens) is mandatory

    🔸 How to read a Gartner Magic Quadrant: why the vendor cautions matter more than a Leader position

    🔸 Which cost risk Gartner found in all four Leaders: Appian, Pega, ServiceNow and UiPath

    🔸 3 signs your company needs an AI agent orchestration platform

    🔸 A simple exercise: a list of all your AI agents and automations, with an owner and a monthly cost for each


    21 min
  • How to use Process Intelligence data your team trusts

    How to use process intelligence data without losing your team's trust: 3 written rules to set before the first dashboard goes live. #ProcessIntelligence #Automation #GBS

    Years ago, in a BPO, one employee logged "stop" almost every afternoon in a simple work-logging tool. His manager called him in for a difficult conversation. The next week his afternoons were full of process clicks, and his output was exactly the same. He was one of the most efficient people on the team: he finished his work by lunch and simply logged the truth. After one 15-minute meeting, the company lost its most honest data.

    In this episode of the AI Automation Dojo, Andrzej Kinastowski explains why employees assume the worst when nobody has written down how data about their work will be used, and which three rules prevent it.

    What you'll learn in this episode:

    🔸 Why the same measurement builds trust or fear, depending only on how it is used

    🔸 Why training managers is not enough, and a written, signed rule is

    🔸 The Never List: 4 things process data should never be used for

    🔸 Why the first response to any bad reading should be a question: "What's blocking you?"

    🔸 Who should act on a finding: the process owner, not the people manager

    🔸 Why your first project should fix the oldest complaint on the floor, not chase the biggest ROI

    🔸 How to bring your works council a draft they can test instead of a vendor brochure

    Write the rules before the first dashboard. Otherwise your employees will write them for you

    26 min
  • How to deploy Process Intelligence without fear

    Why do corporate timesheets always add up to 8 hours? Explore #ProcessIntelligence #Automation #AIAgents with Andrzej Kinastowski.

    In this episode of AI Automation Dojo, we break down why traditional time tracking simply doesn't work. When companies try to track people in secret or force them to manually log hours in Excel, they don't get accurate data - they get fake busyness and reporting driven by fear.

    Instead of invasive surveillance, we share how to implement Process Intelligence tools built on clear, transparent rules. The goal is to spot real process bottlenecks fast without destroying team trust.

    You’ll learn:

    🔸 Why "stealth tracking" always produces corrupted data

    🔸 How to collect process data without recording screens or spying on people

    🔸 What a public application whitelist is and how it removes employee anxiety

    🔸 How to shift executive questions from "what did you do?" to "what is blocking you?"

    🔸 Why process measurement tools should cover management as well

    If you want to know how to set up process measurement on fair and safe ground, this episode is for you.

    24 min
  • 7 comfortable lies that kill AI transformations

    Why don't AI rollouts die - why do they just stall? #Automation #ProcessIntelligence #AIAgents #GBS #SSC

    Most enterprise AI deployments don't end in spectacular failure. They simply run on indefinitely as attractive demos, wheeled out once a quarter to impress the board. The technology itself works fine. The real problem sits in the layer nobody wants to talk about out loud: badly designed processes.

    In this episode, Andrzej Kinastowski takes an unvarnished look at why technology transformations fall apart. We examine what happens when you industrialize chaos - automating processes that are fundamentally broken and should have been deleted outright rather than sped up. You'll also learn why your organization may not actually run on the expensive ERP system at all, but on hidden Excel files that modern analytics tools are blind to.

    This one is made for directors, managers and leaders in finance, HR and administration who want to stop investing in comfortable half-measures and start implementing solutions with a real, visible return in the P&L.

    What you'll learn in this episode:

    🔸The 7 comfortable lies of transformation – from blind faith in analytics platforms to pointlessly building new AI teams.

    🔸Industrializing chaos – why using technology to speed up a broken process only compounds the problem.

    🔸The process mining blind spot – where the work actually leaks away, out of sight of your ERP and CRM.

    🔸The personal productivity trap – why saving a few minutes on chatbot-written emails will never show up in the financial result.

    🔸The executive sponsor – why without one, your odds aren't lower, they're exactly zero.

    🔸What actually works – how simple process mapping and fixing the handoffs between departments recovers real money.

    Instead of looking for shortcuts, choose the correct option over the comfortable one. Fix your processes before you hand them to artificial intelligence.

    21 min
  • AI in Shared Services: where it works, function by function

    What you offshored Is already your AI Roadmap. Your Shared Services Center already ran this exact filter 30 years ago and it's the same filter AI is running today.

    Somebody in your company already figured out where to automate with AI. They did it years ago, when they decided what to move to your Shared Services Center / GBS / BPO. The criteria never changed: high volume, repetitive, rule-based. That's the same list AI is good at right now.

    In this episode, we walk the floor of a shared services organization function by function — Finance, HR, IT, Procurement, Documents, Legal, Customer Service and flag exactly where to point AI first, where to keep a human close, and where it's still too early.

    What's actually in this episode:

    🔸 An invoice-status agent that answers "did we get paid?" in plain language, without ever guessing

    🔸 Vendor mailbox: from writing 200 near-identical emails a week to approving them

    🔸 IT Service Desk: at some clients, access requests and revokes are up to 40% of all tickets


    🔸 CV screening that took candidate review from two weeks down to 48 hours

    🔸 A legal-monitoring system reading up to 600 regulatory acts a month, sorting them with 93–95% accuracy — in some cases beating the specialist who used to do it by hand

    🔸 Document processing that cut prep time by 70–90% on hostile, multi-language scans

    🔸 A logistics client whose automation went from a planned 30% to north of 60%

    🔸 Why a customer-service AI pilot opened at 68% accuracy — and why that's the point, not the problem

    Download the AI Automation Playbook here: 👉 https://office-samurai.com/2026-ai-automation-playbook/?utm_source=youtube&utm_medium=organic&utm_campaign=ai_playbook_2026_description_20260819

    21 min
  • Is AI a massive disappointment? The 2026 state of the enterprise report

    Is #AI a "massive disappointment"? Dissecting the WRITER 2026 report on #EnterpriseAI. #Automation #AIAgents #Strategy #BusinessAutomation.

    In this episode of AI Automation Dojo, Andrzej Kinastowski takes a "katana" to a glossy 36-page report on AI adoption. While the document looks professional with its expensive gradients, it reads more like a horror novel where nearly half of executives (48%) admit that AI adoption at their company has been a "massive disappointment".

    Download the AI Automation Playbook here: 👉 https://office-samurai.com/2026-ai-automation-playbook/?utm_source=youtube&utm_medium=organic&utm_campaign=ai_playbook_2026_description_20260729

    We analyze data from the companies supposedly "winning" the AI race, only to find that 75% of their strategies are "more for show", essentially a LinkedIn post wearing a trench coat. We dive into the corporate paradox where 69% of companies are cutting staff based on the promise of AI productivity, while 39% admit they have no formal plan for how that AI will actually generate revenue.

    What you will learn in this episode:

    🔸 Why satisfaction with AI tools dropped across every single dimension measured this year.

    🔸 The rise of "AI sabotage"- why 44% of Gen Z employees are intentionally feeding garbage to models.

    🔸 Why 75% of employees would trust an algorithm more than their own manager for work tasks.

    🔸 The risk of promoting AI to the C-suite when one in three leaders isn't sure they could "pull the plug" if things went wrong.

    🔸 How to stop "automating vibes" and start tying AI to measurable business outcomes.

    Instead of chasing sci-fi hype, we discuss the unglamorous but essential recipe for success: fixing processes before automating them, governing without strangling innovation, and keeping a Human-in-the-Loop on the kill switch.

    Don’t be next year’s cautionary statistic.

    25 min
  • Want to build a 21st-century Digital Workforce? (The 5 levels of maturity)

    Can you build a 21st-century digital workforce on a hunch? Mastering #ProcessIntelligence is the only way to scale #Automation and #AIAgents without flying blind.

    In this episode of the AI Automation Dojo, Andrzej Kinastowski dissects the desperate corporate quest to figure out what employees are actually doing all day. Most organizations still operate in the "Wild West" managing by intuition and the foundational premise that "Bob usually copies this into Excel on Tuesdays."


    We break down the five levels of process maturity - from manual "spreadsheet purgatory" that burns 5% of your workforce's capacity to the holistic zen of Productivity Mining. This conversation is for executives demanding efficiency gains and managers tired of outdated, "coffee-stained" process maps that don't reflect reality.

    You’ll learn how to move beyond the "Happy Path" trap of traditional recording and see the "dirty" work happening in Outlook and Excel that your ERP systems are blind to. We explore how to find the truth of your operations without falling into the "Level 666" trap of corporate spyware.


    Data doesn't fix a broken culture; it just illuminates it. Stop optimizing a manager's hunch and start fixing reality.

    Gut feeling: double-checked.

    24 min
  • Agentic AI in finance: automating P2P inquiries (case study)

    Can artificial intelligence handle the constant stream of finance queries? #AIAgents #Automation #ProcessIntelligence


    In this episode of the AI Automation Dojo, we host Mateusz Ślemp from Amer Sports. We discuss the deployment of their first global AI agent, a tool designed to help employees reclaim time spent on manual payment status checks in corporate systems.

    Mateusz explains why the agent is not connected directly to the SAP system and how PowerBI acts as a secure data layer to prevent system overload. We also cover the importance of data security and protecting confidential information in large organizations. You will learn how to manage AI projects in environments regulated by strict SOX audits.


    What you will learn in this episode:

    🔸Replacing thousands of support tickets with a simple chat on Teams,

    🔸Using PowerBI architecture instead of overloading the core SAP system,

    🔸Why invoice amounts were intentionally removed from the agent database,

    🔸The evolution from traditional RPA robots to intelligent agents,

    🔸How a phased testing strategy helped detect business errors before launch,

    🔸AI manages the heavy lifting while humans focus on exceptions.


    1 hr 3 min
  • AI Agents case study: 93% accuracy in legal document monitoring

    How to safely implement #AIAgents and #UiPath in a large enterprise? Discover a real-world case study from the #Energy sector, where #Automation and #AI are supporting the legal department.


    👉 Download the AI Automation Playbook here.


    In this episode of AI Automation Dojo, Andrzej Kinastowski breaks down a practical example from a Lithuanian energy company that successfully combined innovative AI solutions with strict cybersecurity requirements. You’ll see how to overcome “cloud paranoia” by implementing AI agents in a critical infrastructure environment, where a single mistake could create a real risk to power delivery.

    18 min
  • What if you could see how work in your company really happens?

    Are you managing your team based on facts or just "monthly averages"? It’s time to look at the real data behind #ProcessIntelligence, #WorkforceOptimization, and #OperationsManagement.

    In this episode of The AI Automation Dojo, Andrzej Kinastowski discusses why official company flowcharts and PowerPoints rarely match the reality of daily work. While management often relies on averages to distribute tasks, the raw data usually reveals a different story: some teams are overwhelmed with 12-hour workdays while others have hidden capacity.

    We explore how large organizations can use task mining to identify the real causes of low productivity. Instead of blaming "slow employees," we look at structural issues like system lag, too many meetings, and the "tool fragmentation" where people have to jump between dozens of different applications just to complete one process.

    This conversation is a guide for operations leaders and GBS/SSC managers who want to use data to balance workloads, improve employee morale, and make better decisions about automation and team size.


    What you will learn from this episode:

    🔸 Why "monthly averages" hide the real bottlenecks in your operations

    🔸 How to identify the "quiet heroes" – high-performing employees who are often overlooked

    🔸 The impact of "tool fragmentation" (e.g., using 6 different SAP instances simultaneously)

    🔸 Case Study: How Vesuvius used data to prove 12-hour shifts during month-end and redesigned their process

    🔸 How to reduce team size through natural attrition and better workload balancing

    🔸 Why system performance, not employee speed, is usually the reason for delays

    19 min

About AI Automation Dojo by Office Samurai

From the publisher's feed

Tired of processes so old they remember when 'the cloud' was just a weather phenomenon? The Office Samurai Podcast slices into the heart of modern business, showing how AI and automation aren't just…