AI has moved from assisting people to increasingly doing the work itself. It can plan, analyse, identify problems, recommend decisions, and execute complex workflows. But as AI becomes more autonomous, one question becomes increasingly important: who remains accountable when something goes wrong?

KEY CONCEPTS COVERED:

• How AI has evolved from an assistant into increasingly autonomous agentic systems

• Where businesses should draw the line between AI automation and human decision-making

• Why governance, data security, accountability, and human expertise become more important as AI gets more capable

In this episode of Chain Reaction, Michael Ostroumov is joined once again by Waj Akram, CTO of FLOX, to revisit their conversation about AI, automation, and human judgement. They explore how rapidly AI has evolved from requiring constant supervision towards systems capable of planning, executing, and reviewing increasingly complex work. Waj also explains how FLOX is putting this into practice through its multi-agent software development system, customer-facing assistant Holly, and Wingman, an AI engine embedded directly into the logistics platform to help users identify opportunities, prioritise exceptions, and make better operational decisions.

Michael and Waj then tackle the bigger question: how much control should AI actually have? They discuss why effective AI requires clear rules, guardrails, testing, human oversight, and careful control of sensitive data, as well as why accountability must remain with people even when AI can operate autonomously. The conversation also looks ahead to how increasingly capable AI could reshape logistics roles, software development, small businesses, and the skills people need as AI becomes embedded into everyday operations.

Watch this episode on YouTube: https://youtu.be/tyjzJ8ja-DY 

TIMESTAMPS:

(0:00) – Introduction and How much has AI actually changed?

(7:39) – Building an automated software development lifecycle

(11:13) – AI benchmarks vs what actually works in practice

(16:49) – What FLOX is actually building with AI

(23:52) – Reducing administration rather than replacing people

(33:37) – Data security and AI governance in logistics

(38:15) – Guardrails, hallucinations, and AI reliability

(40:47) – Where AI goes next

(43:36) – Could AI help smaller businesses compete?

(46:35) – AI, profitability, and the end of traditional economies of scale?

(49:21) – Magic wand question: don't let AI take the wheel

CONNECT WITH THE HOST & GUEST:

Michael Ostroumov –https://www.linkedin.com/in/michaelostroumov/

Waj Akram – https://www.linkedin.com/in/akramwajahat/ 

CONNECT WITH FLOX:

Website:https://www.flox.is/

LinkedIn:https://www.linkedin.com/company/flox-logistics-services/

ABOUT THIS EPISODE:

In this episode, Chain Reaction explores how quickly AI has evolved from an assistant into increasingly capable agentic systems and what that means for logistics, software development, and business operations. Michael Ostroumov and Waj Akram discuss multi-agent AI, human oversight, autonomous decision-making, data governance, security, specialised AI models, and FLOX's experience building AI directly into its technology, while addressing a critical question: just because AI can make a decision, should it? They explore why accountability must remain with people, why businesses need control over the data AI can access, and how combining human expertise with increasingly capable technology could reshape logistics roles and business models.

#ChainReactionPodcast #SupplyChain #ArtificialIntelligence #AI #AgenticAI #Logistics