How conflicting agent responses hurt trust

A customer asked one simple question. They received two completely different answers. That moment alone can damage trust faster than most companies realize. A customer contacts support. The first agent gives one answer. Later, another agent says something completely different. Now the customer is confused. And more importantly, trust is shaken. I have seen this happen more often than people think. Not because the agents were careless. But because the information behind them was not aligned. Different documents. Outdated notes. Or sometimes no documentation at all. So each agent answers based on what they know in the moment. The result is inconsistency. And inconsistency quietly damages customer confidence. From the customer’s perspective, it can feel like the company does not fully understand its own product. Many teams underestimate how damaging this can be. Customers can forgive delays. They can even forgive mistakes. But conflicting answers create doubt. And once doubt enters the conversation, every future interaction becomes harder. This is why internal knowledge systems matter more than many teams realize. Support teams need a 𝘀𝗶𝗻𝗴𝗹𝗲 𝘀𝗼𝘂𝗿𝗰𝗲 𝗼𝗳 𝘁𝗿𝘂𝘁𝗵. Clear documentation. Regular updates. A shared understanding of how things actually work. When that foundation is strong, something interesting happens. Agents respond with more confidence. Customers receive consistent answers. Resolution is faster because people are no longer guessing. Customer support is often seen as a communication role. But in reality, it is also a 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗿𝗼𝗹𝗲. The stronger the knowledge system behind the team, the stronger the customer experience. For anyone building or managing support operations, this is one of the quiet systems that makes everything else work better. When a team truly understands the product, customers can feel it immediately. And that confidence shows up in every conversation. For those working in support, CX, or operations: What is one knowledge gap you have seen cause repeated customer issues? #CustomerExperience #CustomerSupport #CX #CustomerSuccess #SupportLeadership #CustomerRetention #RemoteSupport #BusinessGrowth #Operations

AI Actions Without Context = Confusion, Not Collaboration (Backed by experimental data) 🔍 Problem When AI agents make decisions in team settings without explaining their actions, it disrupts trust and performance — especially when those actions contradict human expectations. → Users hesitate to rely on AI. → Misinterpretation increases. → Collaboration suffers. 📚 What Science Says This paper presents two empirical studies with 625 participants evaluating when AI should explain its actions in team scenarios, particularly when actions are unexpected. Core finding: AI explanations are most useful when its actions conflict with human expectations — especially in high-visibility or high-stakes moments. 💡 Key insight: AI doesn't need to explain everything — only when its behavior breaks human expectations or contradicts norms. 💼 How It Shows Up at Work In practical team settings, lack of timely AI explanations leads to: • ���� Humans second-guessing AI decisions • 😰 Increased anxiety in ambiguous scenarios • 🤐 Over-reliance or complete dismissal of AI input • ❌ Delayed task performance when AI decisions are unclear When AI offers proactive explanations during critical or unexpected moments, teams: • Improve shared understanding • Experience fewer coordination breakdowns • Report higher team trust ⚠️ Why It Matters Team success in Human-AI collaboration hinges on shared mental models. Without explanations, AI agents risk becoming "black boxes" that teammates either blindly follow or entirely reject. 🎯 Outcome: Trust + Team Performance suffer when humans can't predict or understand AI behavior. 📉 Ineffective AI teaming results in: Poor decision-making Reduced adoption of AI tools Misaligned roles and responsibilities ✅ Actionable Insights • Encourage humans to critically assess AI input rather than passively accept or reject it • Train employees to recognize when to ask for AI explanations 🛠 Practical Solutions for Managers 1. Triggered Explanations Only Don’t overwhelm users. Ensure AI agents are designed to explain only when their behavior deviates from expectations. 2. Design AI for Team Awareness Use explanation thresholds: Train AI agents to detect when their actions may surprise teammates and offer context proactively. 3. Role-Based Explanation Expectations Customize explanations depending on the human’s role: → Decision-makers need deeper context → Support roles may only need surface-level info 4. Foster Shared Mental Models • Integrate AI behavior briefings in onboarding • Regularly clarify how AI makes decisions • Simulate scenarios with unexpected AI actions during training #AIinTeams #HumanAITeaming #ExplainableAI #TrustInAI #AILeadership #SelfLeadership #AIandCollaboration #TeamPerformance #PsychologicalSafety #DigitalTeams #ResponsibleAI #AIProductivity

Agentic AI transformation is on the mind of every CEO I meet. But here’s what’s often missing from the conversation: when and how agents fail. And the truth is, they fail often. Microsoft’s AI Red Team notes that while many of these failures mirror those seen in LLMs, their frequency and impact are greatly amplified once agents operate with memory, tool access, and multi-agent collaboration . The whitepaper organizes failures into two groups: 𝗡𝗼𝘃𝗲𝗹 𝗳𝗮𝗶𝗹𝘂𝗿𝗲 𝗺𝗼𝗱𝗲𝘀 (𝘂𝗻𝗶𝗾𝘂𝗲 𝘁𝗼 𝗮𝗴𝗲𝗻𝘁𝘀): • Agent compromise or impersonation: attackers or even other agents pretending to be trusted components. • Multi-agent jailbreaks: one agent persuading another to ignore safeguards. • Flow manipulation and provisioning poisoning: poisoned configurations or orchestration logic redirect entire workflows. • Organizational knowledge loss: corrupted memories or over-delegation leading to long-term degradation of institutional knowledge. 𝗘𝘅𝗶𝘀𝘁𝗶𝗻𝗴 𝗳𝗮𝗶𝗹𝘂𝗿𝗲 𝗺𝗼𝗱𝗲𝘀 (𝗺𝗮𝗴𝗻𝗶𝗳𝗶𝗲𝗱 𝗯𝘆 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝘆): • Hallucinations: no longer just misleading text, but incorrect actions in enterprise systems. • Misinterpreted instructions: plausible but unintended workflows executed as if correct. • Bias amplification: skewed outputs scaling to affect whole populations of users. • Transparency and consent gaps: agents making consequential decisions without intelligible explanations. The paper stresses the effects of these failures: agent misalignment, action abuse, denial of service, incorrect decision-making, user harm, and erosion of trust . In practice, this means a poisoned memory entry can escalate into data exfiltration, or a misinterpreted instruction can lead to system-wide outages. Mitigation is possible and the taxonomy describes dozens of design controls. Distilling them, four stand out as foundational: 𝟭. 𝗜𝗱𝗲𝗻𝘁𝗶𝘁𝘆 𝗮𝗻𝗱 𝗽𝗲𝗿𝗺𝗶𝘀𝘀𝗶𝗼𝗻𝗶𝗻𝗴: Each agent should have a unique identifier with role-based access. This makes impersonation harder and enables granular auditability. 𝟮. 𝗠𝗲𝗺𝗼𝗿𝘆 𝗵𝗮𝗿𝗱𝗲𝗻𝗶𝗻𝗴: Memory must be treated as an attack surface. That means authenticated writes, restricted reads, and live monitoring to detect poisoning or leakage. 𝟯. 𝗖𝗼𝗻𝘁𝗿𝗼𝗹-𝗳𝗹𝗼𝘄 𝗴𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲: Autonomy requires constraints. Critical tool calls and data accesses should be deterministically gated and verified to prevent cascading failures. 𝟰. 𝗘𝗻𝘃𝗶𝗿𝗼𝗻𝗺𝗲𝗻𝘁 𝗶𝘀𝗼𝗹𝗮𝘁𝗶𝗼𝗻: Agents should operate in strong sandboxes so that a compromised or malfunctioning component cannot propagate failures beyond its scope. Agentic AI doesn’t just inherit old risks, it introduces new ones that are harder to anticipate and more damaging when they occur. Transformation without failure-mode awareness isn’t transformation at all. It’s exposure. Enterprises that succeed will be those that invest as much in designing for failure as in scaling capability.

Meet the Quasi-Creature. We name the thing, set it beside the data, pour a cup of tea, and the strangeness becomes visible. The paper argues that the “uncanny valley” of GenAI is not about looks at all. It’s about agency - how agent-like a system seems versus how reliably it behaves. When autonomy outruns reliability, trust collapses. That is the uncanny valley of agency. The authors ground this idea in an illustrative study (‘Move 78’), involving 37 participants working with a tuned chatbot on a creative task. The sample is modest, the signals striking. On NASA-TLX - a standard workload scale from NASA - frustration averaged ~15 out of 20, with mental demand similarly high. Two patterns stood out. First, inefficiency drove frustration. When the system felt inefficient, frustration spiked (r ≈ −0.85). Second, more interaction wasn’t collaboration - it was repair. Higher message counts correlated with higher frustration (r ≈ +0.74). What looked like “engagement” was struggle. There was an expert twist. Those most familiar with AI were the most annoyed. Richer mental models meant sharper disappointment when the system broke its own implied rules. Regression confirmed it: inefficiency, negative reactions, and prior familiarity explained nearly 80% of frustration, with inefficiency the strongest factor. Beneath the stats lies what the authors call rupture and repair. We approach the system as a tool. It stumbles, we try to fix it as a tool. But when it resists in ways that feel agent-like, our perception shifts: no longer a hammer, but a house-guest. The Quasi-Creature sits across from us, uncanny and unpredictable. That flip is the valley in motion. The design implications are blunt. Stop chasing flawless mimicry. Instead, build seams and signals: show when context is lost, expose uncertainty, offer reasons instead of coy confidence. Make the system’s limits visible so people can predict failure and recover. This isn’t just UX polish - it’s an ethical stance about how much opacity and helplessness we normalise in the infosphere we now share. So invite the creature in, pour the tea, and place it beside the tables. Seated with the evidence, the oddity becomes diagnosis. That is the power of naming it: to see clearly why these systems can feel brilliant, baffling, and strangely exhausting - all at once. Here comes the sun. It’s alright.

The Email I Regret Sending: How One Harsh Response Damaged a Supplier Relationship It was late. I was tired. A supplier had missed a critical delivery deadline, and I was frustrated. Without thinking, I fired off a blunt email: “This delay is unacceptable. We expected better.” I hit send and went home. The next morning, I got a reply. It was short, defensive, and cold. The tone of our relationship had shifted. What had been a decade-long partnership now felt strained. The supplier stopped sharing proactive updates. Communication slowed. Trust eroded. That email taught me a painful lesson: procurement is diplomacy in action. We deal with people, not just contracts. Emotional intelligence isn’t a soft skill—it’s a strategic one. My reactive message cost us more than a delivery—it cost us rapport, reliability, and respect. I’ve since changed how I communicate. I pause before responding, especially in moments of stress. I ask myself: Is this message constructive? Does it preserve the relationship? I’ve also trained my team on communication etiquette and emotional intelligence. We now treat every supplier interaction as a chance to build—not burn—a bridge. Procurement isn’t just about negotiating terms. It’s about nurturing trust. One email can strengthen a partnership or shatter it. Choose wisely. Have you ever sent a message you wish you could take back? How do you manage supplier relationships under pressure? Share your thoughts—I’d love to learn from your experiences. #ProcurementDiplomacy #EmotionalIntelligence #SupplierRelationships #StrategicCommunication #TrustInSourcing

If you're ignoring this in your team, you're paying for it. You just don't see the bill yet. The silent killer in teams isn't conflict. It's how you respond to it. I've seen partnerships dissolve, deals fall apart, and talented people walk away. Not because they disagreed. But because no one knew how to disagree well. Here's what I've learned building teams in this industry: 1.Conflict that builds: ~ Someone questions your pricing strategy → You explore it together ~ Two agents disagree on market positioning → They debate with data, not ego ~ Your team challenges a decision → You listen before defending 2. Conflict that destroys: ~ Shutting down questions because "that's how we've always done it" ~ Blaming team members when a listing doesn't perform ~ Making someone feel small for seeing things differently The gap between these two? One word: Respect. What's worked for me: When tension rises, I pause and ask: "What problem are we actually solving?" I separate the person from the issue. Always. If I'm too close to it, I bring in someone neutral to mediate. Always remember your team is watching how you handle disagreement. Every single time. If you dodge it, they'll dodge it. If you weaponise it, they'll fear it. If you navigate it with integrity, they'll trust you more, not less. The strongest teams I know don't avoid conflict. They've just learned to make it productive. What's your biggest challenge when conflict shows up in your team?

Masks at Work: How They Hurt More Than They Help When words change with every audience, trust quietly walks out of the room. It’s something we all notice but rarely talk about — how some people say different things to different people, simply to match what the other person wants to hear. While it’s natural to adjust our tone or style depending on the situation, when the message itself keeps changing, it raises a deeper question: Can I trust this person? In professional spaces, many wear a “mask” — trying to please everyone, fit in, or avoid conflict. But when words shift based on the listener, it starts to show. It shows in contradictions, in inconsistencies, and eventually in a loss of credibility. And slowly, it chips away at trust. Trust is built on authenticity. When people feel that what you say reflects what you truly believe — consistently — they are more likely to connect, collaborate, and commit. But when they sense that your words change depending on the audience, doubts begin to grow, and relationships weaken. Organizations thrive when people are consistent — not playing roles, but standing by their values with respect and clarity. It doesn’t mean being rigid or unkind; it means communicating with honesty and integrity, whether in a boardroom or a casual conversation. Consistency in words isn’t just good communication — it’s the foundation of trust.

The most expensive sentence in Punjab’s supply chain is only four words long: “Sir, leave it on me.” This single line has cost suppliers millions, largely due to non-professional agents. Yet many suppliers continue to trust it blindly. How non-professional agents destroy value: Blocking Communication: Agents block direct contact between suppliers and buyers. Factories operate in the dark, risking all their finances while agents control every step. Forcing Unrealistic Timelines and Prices: Agents push impossible deadlines and unsustainable rates. Suppliers must learn to say no to protect both quality and investment. Unauthorized Changes: Fabric choices, quality levels, or product specs are altered without buyer approval, leading to failures in testing and final inspections. Kickbacks and Hidden Commissions: Agents misuse buyers’ names for personal gain, creating ethical and financial risks for everyone involved. Avoiding Accountability: When things collapse, agents disappear, leaving exporters and suppliers to deal with the damage. Real example: One agent’s mismanagement caused a supplier to lose five million dollars in shipments, now rejected and sitting in a warehouse. No direct access to buyers, no clear communication, no final approvals. And the predictable excuse at the end: “I am in the hospital because of an emergency.” The bigger issue: Many buyers chase the lowest price instead of focusing on the product. This pressure encourages agents to cut corners, putting suppliers at even greater risk. Key Lessons: For suppliers: Stand firm. Reject unrealistic prices and impossible timelines. For buyers: Work only with credible, professional agents who promote transparency and allow direct communication. The wrong agent can destroy millions and permanently damage trust.

Recently on one of our projects, I faced a situation that reminded me what project management truly means in practice. The contractor submitted his valuation. The Consultant QS reviewed it, did his physical confirmation and felt the figures did not reflect the actual work done. His recommendation to the Project Manager came in very lower than what the contractor expected. When the contractor saw it, he was upset. Threats to slow down work started coming up. Tension rose immediately. What struck me was this: There had been no prior reconciliation between the contractor and the Consultant QS before the recommendation was issued. From my experience, when a valuation is submitted, there is usually some level of engagement — discussion, clarification, measurement checks before the Consultant QS finalizes his recommendation. That alignment often prevents exactly this kind of situation. But this time, there was a communication gap. And I found myself right in the middle of a dispute that was waiting to happen. The challenge was that the recommendation had already gone to the client. Approval had been given. The certificate process had even commenced. So this was no longer a simple correction — it had become a matter of dispute resolutionand relationship management. After I had some discussions with the Consultant QS agreed to take a more accommodating position in the next valuation to balance the gaps identified. But the real lesson here was not about the numbers. It was about communication. When the Contractor QS and Consultant QS do not engage early, the Project Manager inherits a preventable conflict. And if such patterns continue, project timelines, relationships, and trust begin to suffer. One of the realities of being a Project Manager is this: You must always be ready to step into the role of dispute resolution. Not because something went terribly wrong, but because simple professional communication didn’t happen when it should have. Many site disputes are not technical problems. They are communication problems wearing technical clothes.