Talk by Jess Grogan-Avignon and Jack Wang (Accenture) on why enterprise agentic AI projects struggle beyond the build phase. They recount building an agentic application in two weeks that then took 12 months to reach production because multiple enterprise teams (infrastructure, security, AI gateway, data governance, application) had to align — an organizational, not a technical, bottleneck. They outline five tensions that predict project success: turning human approval chains into executable code; funding AI like a VC portfolio rather than demanding committed returns from each project; running delivery through hypothesis-driven loops instead of milestone programs; building trust via staged autonomy (shadow mode, advisory mode, controlled autonomy) gated by outcome evidence; and treating living product memory from customer signals, rather than ERP data, as the real moat. # Channel Identity Observations — Video: "Most Enterprise Agentic Projects Are Doomed, Here's Why" ## Explicit Self-Description Evidence - **Speakers identify themselves by first name only**: > "Hi everyone. I'm Jess. This is Jack." - **Employer/organizational affiliation is stated in the title**: "Jess Grogan-Avignon & Jack Wang, Accenture" — the speakers do not name Accenture in the transcript itself, but present as consultants doing client delivery work. - **Claimed domain of expertise**: large-enterprise AI deployment. - > "We work in the world of enormous enterprises. So, telecoms, utilities, serving entire nations, government, healthcare… consumer products in your home right now." - **Claimed delivery experience**: > "learning from our experience deploying AI at large enterprises over the last couple of years." ## Claimed Scope - **Focus**: enterprise AI/agentic transformation — specifically why enterprise AI projects fail and how to make them succeed. - **Content structure**: a framework of "five enterprise tensions" (speed, value, delivery, trust, moat) plus prescriptive takeaways. - > "we're going to share with you today five enterprise tensions… if you understand these five, we think you can probably predict the success of your next AI project before you get started." - **Breadth of topics demonstrated**: enterprise governance/CI-CD debt, business-case/finance models, hypothesis-driven delivery for ML, progressive autonomy & trust ladders, and competitive moats via feedback loops — indicating a channel-level scope spanning **enterprise AI strategy, delivery practice, and organizational change**, not pure coding tutorials. ## Demonstrated Presentation Patterns - **Practitioner-consultant voice**: first-person plural client anecdotes, e.g. > "We had the pleasure of delivering agentic solution in a large corp… the whole application we built took about 2 weeks. It then took another 12 months to get that into production." - **Positioning against "slide-deck" consulting**: > "Others studied the games, crafted the playbooks, and then ran the workshop. But they went home. We stayed. We shipped through the reality. And that is our moat." — an explicit competitive identity claim. - **Data and citation use**: references Accenture-style research ("Our research show that 12% of companies reach what we called AI achiever"), GitHub commit statistics, and named examples (Klarna, Shein, Walmart, JP Morgan, Cursor). - **Interactive conference format**: audience polls ("Who here has had to start a project with a business case…?", "do we have any data scientist or machine learning engineers here?") and closing applause suggest a live conference talk, likely part of a developer/AI-engineering event. ## Intended Audience - **Primary**: practitioners working on or funding enterprise AI projects — AI/ML engineers, delivery leads, and enterprise decision-makers (CTO/CEO/CFO roles are addressed directly in the closing). - **Evidence**: direct addresses to roles — > "your CFO needs to think like a VC," "make the governance speed your CTO's top engineering problem," "that's for CEOs." The audience is assumed comfortable with AI tooling: > "probably all of us don't review every single code output." ## Claims vs. Demonstrated Distinction - **Creator claims** (stated, not independently verifiable in-material): years of enterprise deployment experience, research showing 12% "AI achievers" and 50% higher revenue growth, a client anecdote about a 12-month deployment delay. - **Demonstrated**: structured framework thinking, consultant-style storytelling, multi-stakeholder fluency (engineering, finance, governance, executive), consistent emphasis on enterprise-scale consequences ("bad deployment… can take down critical national infrastructure"). ## Private Motive Check - No private beliefs or motivations inferred beyond directly stated positioning (e.g., "We stayed. We shipped."). The talk does implicitly promote the speakers' delivery credibility, but no channel-wide conclusions about the channel's other content are drawn from this single video. # Delivery Observations: Jess Grogan-Avignon & Jack Wang — Enterprise Agentic Projects ## Overall Structure - **Two-speaker tag-team format.** Jess opens ("I'm Jess. This is Jack.") and delivers the framing and early sections; Jack takes the delivery-trench anecdotes and later sections. The handoffs are visible in the transcript but not formally announced after the opening. - **Numbered framework.** The talk is organized around a stated list: "we're going to share with you today five enterprise tensions." The speakers close by naming them: "speed, value, delivery, trust, moat." - **Preview and summary habit.** The talk opens with a promise ("if you understand these five, we think you can probably predict the success of your next AI project") and closes with a compressed prescription: "Bet like a VC, upgrade for machine speed, and engineer for trust with the feedback loop from day one." ## Tone and Register - **Conversational conference keynote style** with filler and self-corrections: "Cool. Let's get into it. Um 18 months ago…", "Okay, next up.", "All right, next one.", "Right, last one." - **Bold, declarative aphorisms** are a recurring rhetorical device: - "Human pace. Human pace. That is what shifted." - "It's a floor, not a fortress." - "Feedback is not the option. Feedback is the only moat." - "Engineer for trust, not just for completion." - **Dry humor and audience teasing**, especially in Jack's sections: "Um most of enterprises probably treated you guys like the modern IT crowd… kept you guys in the basement while the upstairs doing the real work through the Jira board at PI planning." ## Audience Interaction - **Repeated rhetorical questions to the room**, a stable pattern across the talk: - "Many of you here may ship here on Fridays and then rollbacks on Saturdays." - "Who here has had to start a project with a business case to unlock internal funding? Cool. Welcome to our world." - "do we have any data scientist or machine learning engineers here? Brilliant." - Jess also uses rhetorical questions for transitions: "So, how do you go faster?" ## Teaching Habits and Analogies - **Vivid extended analogies** to make enterprise friction tangible. The most developed example is the Google search analogy: > "think about Google search. Before you see the results come out, they're going to be three teams review the results first. Need to be a legal sign-off for the results. And then they say, wait for 2 weeks because we're quarter end, it's change freeze." - **Concrete numeric evidence** cited to support claims: "12% of companies reach what we called AI achiever," "80%", "1 billion commits… 275 million per week… 14 billion by the end of the year," "took about 2 weeks. It then took another 12 months to get that into production," "50% higher revenue growth." - **Named real-world examples** used as evidence: "Octopus, Klarna, Shein," "Cursor's user base of live coders," "Walmart… social media trend scanner," "JP Morgan." - **Coining/repeating branded terms** for concepts: "AI achiever," "progressive autonomy," "exposure ladder," "transactional memory" vs. "living memory," "trust account." ## Metaphorical Framing - **Trust as banking metaphor**, sustained across several sentences: "a deposit or a withdrawal into a trust account with your stakeholders, your leadership, with your end customers." - **Competition/race metaphor** in the moat section: "The day the day you ship is not the finish line. Far from it. It is when the race actually begins." - **Step-ladder scaffolding** for the autonomy progression: shadow mode → advisory mode → controlled autonomy → wider autonomy, each "gated by evidence in outcomes." ## Direct-Address Prescriptions - The closing section shifts to **imperative, role-targeted advice**: - "your CFO needs to think like a VC" - "make finance a transformation partner, not just a gatekeeper" - "make the governance speed your CTO's top engineering problem" - "that's for CEOs. Your moat is not in what you hold from yesterday." ## Self-Positioning - The speakers repeatedly distinguish themselves through presence: "Others studied the games, crafted the playbooks, and then ran the workshop. But they went home. We stayed. We shipped through the reality. And that is our mode." ## Delivery Quality Notes (Uncertainty Flagged) - **Jack's sections show more disfluency and grammatical irregularity** than Jess's ("And those are the things that is currently consumed the energy," "they will be the ones that learn to learn"), suggesting partially improvised delivery. This is an observable pattern in his segments but the transcript gives no way to confirm speaker attribution for every segment with certainty beyond handoff markers. - Transcription quality appears imperfect ("Our research show," "the societies these enterprises underpinning"), so fine-grained fluency judgments are **low confidence**. - Audience laughter/applause is captured only once ("[applause]"), so audience response patterns cannot be assessed reliably. # Position Record: "Most Enterprise Agentic Projects Are Doomed, Here's Why" — Jess Grogan-Avignon & Jack Wang (Accenture) ## Speakers - **Jess** — opens the talk; delivers sections on enterprise speed, business-case friction, trust/progressive autonomy, and the closing prescription. - **Jack** — delivers the enterprise scaffold/coding-agent bottleneck section, the data-science/delivery section, and the moat/"living memory" section. - Note: speaker attribution for individual sections is inferred from transcript handoff cues ("that Jess mentioned," "we had the pleasure..."); both speakers share overall positions. ## Central Thesis - Enterprise agentic projects fail primarily because **enterprise scaffolding — governance, process, sign-offs — designed for human speed** becomes the bottleneck, not AI capability itself. - The speakers propose **five enterprise tensions**: speed, value, delivery, trust, moat. --- ## Tension 1: Speed — Enterprise Scaffold vs. Machine Speed **Claims (Jack, with Jess framing):** - Large enterprises (telecoms, utilities, government, healthcare) built layers of control, process, and governance because "actions have consequences" at national scale — and this historically worked, "but always at human pace." - The barrier is *not* AI's code quality or engineers' context-handling: "It is the actual enterprise scaffolding itself. A human operating system that is designed for human and running at a human speed." **Evidence cited:** - Accenture research: only **12% of companies reach "AI achiever"** status; most are stuck piloting with poor returns. - Anecdote: an agentic solution integrated with a client's centralized AI gateway took **2 weeks to build, 12 months to reach production**, because infrastructure, security, AI gateway, data governance, and application teams all had to align; every configuration change required manual review. - Analogy: enterprise AI delivery is like Google Search if three teams reviewed results, legal signed off, and you waited two weeks for quarter-end change freeze. - GitHub stats cited: 1 billion commits in 2025; averaging 275 million per week this year, "on track for 14 billion by the end of the year." - AI coding agents turn PMs, designers, and domain experts into builders, exploding deployable code while approval/deployment infrastructure stays unchanged. **Prescription:** Every human process must become **adaptable, executable code** — "Not another meeting, not a sign-off chain, code." AI can help build this faster and cheaper; this is a mindset shift, not a new capability. --- ## Tension 2: Value — Business Cases Kill Projects **Claims (Jess):** - Business cases "aren't wrong per se" (oversight, ROI questions) but assume three things are knowable up front: scope/solution, expected value, cost/time. "With AI, this is often backwards — you learn the solution and the business case by doing the work." - When execution/prototyping cost nears zero, the question shifts from efficiency to **new capabilities** — attempting things previously "economically impossible." **Evidence cited:** - AI achievers show ~**50% higher revenue growth** than peers — "not from cost cutting, that's from doing entirely new things." - Emergent successes: Cursor's user base "didn't exist" at launch; Claude Code wasn't a pre-planned roadmap item; Walmart's social media trend scanner/generative designer now competes with Shein and Temu; JP Morgan productized an internal productivity tool into a new revenue stream. **Prescription:** "Your CFO needs to think like a VC." Portfolio of bets, not per-project guaranteed payback; the right question is "what is the cost of *not* doing this?" If finance can't think this way, "that is where your transformation should start." --- ## Tension 3: Delivery — Hypothesis-Driven, Not Feature-Driven **Claims (Jack):** - Data scientists/ML engineers have been treated like "the modern IT crowd" — ignored while "the upstairs" works through Jira boards and PI planning. "This is your opportunity. Agentic delivery is your world, not theirs." - Models are non-deterministic; agent behavior is emergent; agents cannot be scoped like feature builds or milestoneed like fixed programs — "yet that's what entire enterprises are trying to do." - Delivery energy is consumed not by building but by bridging the gap between how systems actually work and stakeholder expectations: "never-ending utopian design up front," guaranteed-performance conversations, endless status updates for decisions that never get made. **Prescription:** - Reshape programs around building **statistical confidence**: small loops of build–evaluate–iterate. - Delivery teams need different skills: comfort with ambiguity, articulating *learnings* not deliverables, translating statistics into stakeholder confidence. "You need hire for it, you need train for it, and you probably need to value that as well." --- ## Tension 4: Trust — Progressive Autonomy **Claims (Jess):** - Society is learning to trust AI (no one cross-checks Google results; people in the room don't review every code output), but AI "isn't there yet"; AI engineers must bridge the gap. - "The completion of individual features is not necessarily the most valuable thing that you ship. The trust in the outputs... is the more valuable thing." - Agentic delivery as **deposits/withdrawals into a trust account** with stakeholders, leadership, and customers. - Agents differ from traditional automation: behavior is emergent and can't be fully foreseen or tested up front. **Prescription — the exposure ladder, each step gated by evidence in outcomes, not project milestones or pass/fail testing:** 1. **Shadow mode** — agent runs alongside humans, can't affect outcomes; human-vs-agent decisions used as iteration signal. 2. **Advisory mode** — agent runs live, recommends only; humans approve/reject. 3. **Controlled autonomy** — agent triggers actions in narrow, low-risk scenarios with clear limits and kill switches. 4. **Wider autonomy** — extended as confidence in target behaviors grows. > "Engineer for trust, not just for completion." --- ## Tension 5: Moat — Living Memory vs. Transactional Memory **Claims (Jack):** - In a recursive world "where AI codes AI, anything you ship can be cloned the minute it goes viral." - Existing enterprise knowledge (CRM, ERP, SOPs) is **"transactional memory"** — "a floor, not a fortress," since every competitor has a version of it. - The real moat is **"living memory"**: edge cases, corrections, emotional intent, and actual behavior at your specific scale and context, captured at the moment of customer contact. - "The day you ship is not the finish line. Far from it. It is when the race actually begins." - Every feature must either generate a feedback signal or deliver on what a signal has already taught you — otherwise "you're building something anyone can copy." > "Feedback is not the option. Feedback is the only moat." --- ## Closing Recommendations **Project playbook (Jess):** - Start now; measure in terms of **confidence**; shape projects around hypotheses rather than requirements/features; run small loops of experimentation, iteration, evaluation. - Make finance a transformation partner, not a gatekeeper; portfolio of AI bets; value beyond cost-out certainty. - Make governance speed the CTO's top engineering problem — "the ultimate technical debt." **For CEOs (Jack):** - "Your moat is not in what you hold from yesterday. It is in what you are learning and compounding every day." - Survivors won't necessarily be earliest adopters but "the ones that learn to learn," building memories through feedback loops and cultivating trust of people and customers — "you cannot buy that... you can only start building that by now." **Final prescription:** "Bet like a VC, upgrade for machine speed, and engineer for trust with the feedback loop from day one." --- ## Qualifications & Caveats Preserved - Business cases "aren't wrong per se" — they raise right questions; the objection is to their *assumptions of predictability*, not to oversight itself. - Stats (12% achievers, GitHub commits, 50% revenue growth) are attributed to "our research" and public figures as cited by the speakers; no independent verification within the talk. - The 2-weeks-to-build / 12-months-to-production anecdote is a single client experience, presented as representative. - Speaker division of sections is partly inferred from handoff cues; both present a unified set of positions. # Video Attribution & Provenance Record ## Video - **Title:** Most Enterprise Agentic Projects Are Doomed, Here's Why — Jess Grogan-Avignon & Jack Wang, Accenture - **Video ID:** `01a15ca9-1bad-7758-89d8-7fdc2693d3a2` ## Speakers & Roles ### Jess Grogan-Avignon - **Role:** Co-presenter / speaker (per title, affiliated with **Accenture**) - Self-introduced: *"I'm Jess. This is Jack."* - Delivers sections on: enterprise scale context, business cases / VC-style portfolio funding, trust and progressive autonomy (shadow → advisory → controlled autonomy → wider autonomy), closing synthesis of five tensions. ### Jack Wang - **Role:** Co-presenter / speaker (per title, affiliated with **Accenture**) - Co-introduced by Jess. - Delivers sections on: enterprise scaffolding / human-speed operating systems, the centralized AI gateway delivery anecdote, hypothesis-driven delivery and data science teams, the "living memory" moat argument. ### Audience - Referenced passively (polling questions, applause at end). No named audience speakers. ## Speaker Attribution Notes - The transcript contains no speaker labels; attribution of specific sections to Jess vs. Jack is **inferred** from discourse cues (e.g., *"the automation behind every delivery that Jess mentioned"*; *"we mentioned our moat earlier"*). Confidence is **moderate**, not definitive. ## Referenced Sources & Third-Party Claims - **"Our research"** (unspecified Accenture research): "12% of companies reach what we called AI achiever"; AI achievers show "about 50% higher revenue growth than their peers." - **GitHub stats (2025):** "1 billion commits" in 2025; "275 million per week" average this year; "on track for 14 billion by the end of the year" — figures attributed to GitHub, not independently verified in transcript. - **Named company examples (third-party claims):** - **Walmart** — social media trend scanner and generative designer competing with Shein and Temu. - **JP Morgan** — internal productivity tool productized into a new revenue stream. - **Cursor, Claude Code ("Cloud code" in transcript), Klarna, Shein, Temu, Octopus** — cited as examples of emergent/fast-moving AI product success. - **First-party experience claim:** Delivery of an agentic solution at "a large corp" integrating a centralized AI gateway — 2 weeks to build, 12 months to production. Company unnamed. ## Promotional Segments - No explicit product promotion or sponsorship identified. - The talk itself functions as a **credentialing/capability pitch for Accenture** ("We stayed. We shipped through the reality. And that is our moat.") — flagged as implicit self-promotion, not an advertisement. ## Availability & Limitations - **Comments:** No comment availability information supplied. - **Curated metadata:** None supplied beyond the title/transcript. - **Uncertainty:** Speaker-to-section attribution is inferred; no timestamps available in transcript. \>\> All right. Hi everyone. I'm Jess. This is Jack. And we're very excited to be here today. So, let us tell you about our world. So, we work in the world of enormous enterprises. So, telecoms, utilities, serving entire nations, government, healthcare that you just heard a little bit about, consumer products in your home right now. And when you operate at that scale, actions have consequences. So, bad deployment for example can take down critical national infrastructure. And so, over time these organizations have built structures for this reality. Control, process, repeatability, governance, layers and layers of it. And this has worked really well, right? Like for years these companies have seen massive successes and growth. But always at human pace. Human pace. That is what shifted. We're entering a world at a machine speed and transforming everything we know. Our work, our clients, and ultimately the societies these enterprises underpinning. Our research show that 12% of companies reach what we called AI achiever. It means that most of the company are still stuck piloting and spending millions and perhaps not getting too much in their return. 80% The tragedy is not just wasted spend. It's about falling behind in a world that accelerating beyond what they can compute. Many of you here may ship here on Fridays and then rollbacks on Saturdays. A decision you take an afternoon could easily take an enterprise 6 months or more. Octopus, Klarna, Shein, they think that's insane. And then they go on and redefine the games themselves. Others studied the games, crafted the playbooks, and then ran the workshop. But they went home. We stayed. We shipped through the reality. And that is our mode. When you stay, you learn things that the slide decks don't warn you about. So, for example, it's not just about data availability or API availability that impacts AI success. It's the entire enterprise scaffold itself. The very thing that has made these companies so successful, which is increasingly becoming the drag, the thing that is holding them back from capturing AI value at scale. So, we're going to share with you today five enterprise tensions, learning from our experience deploying AI at large enterprises over the last couple of years. And if you understand these five, we think you can probably predict the success of your next AI project before you get started. Cool. Let's get into it. Um 18 months ago, you were probably still needed to explain why AI mattered, why speed mattered. But that battle's gone. The C-levels are convinced. You know, the CEOs are terrified of being left behind now. But yet, the enterprise speed has not really shifted. It's not because AI cannot write good code. It is not because our engineers can't solve the context problem. I think it's something a lot more deeper. It is the actual enterprise scaffolding itself. A human operating system that designed for human and running at a human speed. The automation behind every delivery and that Jess mentioned about. Think about data access, security reviews, you know, deployment process. Most of the enterprises never needed to invest like a tech company. Corporate process balanced with minimal engineering investment with the complementary of stakeholder meetings. And that is how enterprise runs today. Fit for enterprise, fit for human. We had the pleasure of delivering agentic solution in a large corp and integrating their centralized AI gateway. You know, we've been given the essentially testing configuration templated to us. Every single configuration change required a manual review before you can actually hit the test to run. And we eventually have to automate it at and the whole application we built took about 2 weeks. It then took another 12 months to get that into production. And because their infrastructure team, their security team, their AI gateway team, their data governance team, their application teams, they all needed to align. Um the best way I can describe this is think about Google search. Before you see the results come out, they're going to be three teams review the results first. Need to be a legal sign-off for the results. And then they say, wait for 2 weeks because we're quarter end, it's change freeze. That is how AI in enterprise delivery today. So, how do you go faster? I guess you hand an AI coding agent to your developers and the next thing that you find is a massive bottleneck at the code review and the deployment stage, right? And this is going to get worse because these coding agents are turning everyone into a builder, PMs, designers, domain experts. And so, the the supply of deployable code is exploding. Some of you might have seen the GitHub stats. In 2025, they reported 1 billion commits. So far this year we're averaging 275 million per week, which means we're on track for 14 billion by the end of the year. And this is super exciting, but approval infrastructure, deployment infrastructure hasn't changed because these processes were ultimately designed for human speed. The real tech debt here goes beyond the legacy code that exists within applications. And it's the years of underinvestment in the engineering automation, CI/CD, et cetera that allows companies to move faster while maintaining control. So, what's the pathway? Every single human process needs to become adaptable, executable code. Not another meeting, not a sign-off chain, code. And the good news here is that AI can help you build this faster and cheaper, right? Like these aren't new capabilities, but it does represent a fundamental mindset shift for a lot of the organizations that we work with. Okay, next up. Who here has had to start a project with a business case to unlock internal funding? Cool. Welcome to our world. Um now, business cases aren't wrong per se. They, you know, they raise a lot of the right questions, they create oversight, they ensure that someone has thought about ROI. All of these are really good things. But they assume that three things are knowable up front. The scope and the solution, the expected value, and the cost and time to deliver. Now, with AI, this is often backwards, right? You learn the solution and the business case by doing the work. And more importantly, when the execution cost for prototyping, experimentation, building, et cetera drops down to near zero, this is no longer just about efficiency. This is now unlocking capabilities, entire new categories of things that weren't possible previously. It means you can now attempt things that were previously economically impossible for the organization. And that means things like new products, new services, customer experiences, they are just waiting to be reinvented. Um and we see this born out in the stats. So, in terms of the AI achievers that we mentioned earlier, we see them achieving about 50% higher revenue growth than their peers. And that's not from cost cutting, that's from doing entirely new things. If you think about as well recent AI product successes, often they've been emergent. So, you think about Cursor's user base of live coders, they didn't exist when they started building the product or when they released it. Cloud code wasn't something that was planned out on a product roadmap months and months in advance. And on the enterprise side, you have examples like Walmart for example, who built out a social media trend scanner and generative designer that's now allowing them to compete in entirely new ways with the likes of Shein and Temu. Um or JP Morgan who started building something as an internal productivity tool that they've now been able to productize and generate entirely new revenue stream. Now, currently the enterprise finance is wired for certainty, which means that generally your project starts life justifying itself in terms of committed benefits and predictable cost phasing. And that framing can kill projects before you even begin because it's asking a question based on can we justify this specific thing based on predictability, rather than asking about what now becomes possible. And so, the right question to be asking is, what is the impact of not doing this? What is the cost of not doing this? So, your CFO needs to think like a VC. At least when it comes to agentic transformation. Um a VC doesn't bet on one project and demand like 3 years fixed guaranteed payback because they know the certainty from a business case is a fantasy. Rather, they back a portfolio, they're knowing most of the bets may not pay off, but they will basically looking for those ones that compound. They knows that where the true value really lies is in that beyond certainty and that power law exponential growth. Enterprise investment works the same way for AI. The question is not can we justified this project, but are we placing enough bets across the portfolio that we going to hit the right ones that's going to change everything for us. So if your finance function cannot think like that, that is where your transformation should start and because everything else is downstream. Cool. Next one. Um do we have any data scientist or machine learning engineers here? Brilliant. I think hopefully you guys going to like this part. You guys been doing something different everyone else, I hope. Um hypothesize and and experiment, statistical confidence. Um most of enterprises probably treated you guys like the modern IT crowd. Brilliant, quietly right and just very kindly ignored. Um kept you guys in the basement while the upstairs doing the real work through the Jira board at PI planning. However, I think this is your opportunity. Agentic delivery is your world, not theirs. Models are non-deterministic. Agent behavior is emergent. And you should not scope it like just a feature build like software traditionally. And you cannot milestone it like a fixed program. And yet that's what entire enterprises are trying to do. When you are in that delivery trench, what we see is that the more enormous effort we have to spend to really not about building the things, but to really try to bridge that gap between and how any system actually work versus what our stakeholders actually expect. You know, the um the never-ending utopian design up front and the constant conversation you have to talk about guaranteed performance and those kind of endless status updates to for those decisions that never gets made. And those are the things that is currently consumed the energy when we come to delivery this the agentic systems. The IT crowd is never the problem, right? The organization just need to learn your language. And it's the only language we think is matter for agentic delivery. Um the team needs to upskill themselves to learn hypothesis driven delivery. You need to reshape your program around one goal, which is building that statistical confidence. Small loops of build, evaluate, iterate, fast evidence. And then and and actually your delivery team needs to look quite different, too. People who are comfortable with ambiguity who can articulate what they have learned, not just what they have delivered. And most importantly, they can translate those statistical number into stakeholder confidence. And those are different kind of skill set. You need hire for it, you need train for it, and you probably need to value that as well. All right, next one. Now, as a society, we are collectively learning to trust AI. I mean, no one cross-checks their Google results anymore, right? And we in this room are probably quite comfortable with using AI tools. Um probably all of us don't review every single code output that we generate. Um now, AI is on that same trajectory, right? But it's not there yet, and it's our job as AI engineers to bridge that gap. And for large enterprises, that trust gap is not small. Um and what we have learned is that the completion of individual features is not necessarily the most valuable thing that you ship. Um the trust in the outputs and the AI that you build over time is the more valuable thing. And and when I say trust, I mean in the broader sense, you know, in the content, in the accuracy, responsible use, privacy, all of those things that collectively allow end users to trust an AI system. You can think about agentic delivery in some ways as a deposit or a withdrawal into a trust account with your stakeholders, your leadership, with your end customers. And what survives over time isn't necessarily a specific feature. It's that trust that you've built as things evolve and things change. And so the question we ask is, how do you build trust at speed, deliberately, and with evidence? So we spend a lot of time talking to companies about progressive autonomy. Many companies still see agents as basically the same as traditional automation where you complete some tasks, you run them, you deploy them, and they run, and that's that. But agents aren't just built and then turned on, right? Like you can't foresee every single response or behavior up front and test for it. Um their behavior is emergent. And this is particularly relevant in the context of autonomous processes, which is something we do with quite a lot of our clients. Um and so the eval suite, we've heard a lot about it over the last few days, that's super important, um and you have to evaluate the right things, but we also talk about how you actually deploy into production and increase autonomy over time. So you follow this exposure ladder. So you start with a shadow mode where an agent might run alongside human processes, but it can't actually affect outcomes. Um you compare the human decisions that are made to what the agent is saying, and you use that as a signal to iterate um and build up confidence in the specific behaviors that you're you're trying to achieve. Um you keep iterating, you then move up to more of an advisory mode where the agent runs live, um but it only recommends. So the humans are still playing an active role in the workflow. They can approve or reject the outcome. And again, this provides you with another signal that you can incorporate and you iterate again. Then you shift up to controlled autonomy where the agent is able to run and trigger actions, but in a narrow, low-risk uh scenarios. Um and it has clear limits, kill switches. And over time, you can extend that up to um to to wider autonomy based on achieving the right level of confidence in the target behaviors that you're trying to drive. And the key here is that each step is gated by evidence in outcomes. It's not based on completion of activities in a project plan or pass-fail testing. It's entirely about the confidence, the trust in those outcomes. So, engineer for trust, not just for completion. Right, last one. Um we mentioned our moat earlier. Now, what is yours? In a recursive world where AI codes AI, anything you ship can be cloned a minute it goes viral. So ask this, what is unique only to you? I think your existing enterprise knowledge, the CRM, the ERP, the SOPs, they got you to the table. We call it the your transaction your transactional memory. However, every competitor has one version of that. It's a floor, not a fortress. The real moat is in the moment when your customer touches your product. Edge cases, corrections, emotional intent, and actual behavior at your specific scale in your specific context. Those signals belong to you. And we call it your living memory. The day the day you ship is not the finish line. Far from it. It is when the race actually begins. How quickly you can compound and iterate. How fast you can turn a signal into value. It is a race against yourself to engineering your own competitive edge recursively and constantly. The pathway requires a fundamental shift in your engineering vision. Every feature you ship should either generate that feedback signal or deliver on what the signal has already taught you. Um because if it does neither, you're building something anyone can copy. Feedback is not the option. Feedback is the only moat. So, five enterprise tensions, speed, value, delivery, trust, moat. How do you then apply what we've learned here to make sure that your next agentic project is a success? So first, we say start now. Deliver differently, measure in terms of confidence, uh shape the project around hypotheses rather than requirements or specific features, and run delivery in small loops of experimentation, iteration, and evaluation. Second, make finance a transformation partner, not just a gatekeeper. Um create a portfolio of different AI bets across the organization rather than justifying each project in isolation. And see value beyond the certainty of cost out. And third, um make the governance speed your CTO's top engineering problem. The ultimate technical debt you want to rectify. And the not last one, um that's for CEOs. Your moat is not in what you hold from yesterday. It is what it is in what you are learning and compounding every day. The technology will accelerating. Those who will survive won't be the ones that has to be the earliest adopter, but they will be the ones that learn to learn. They'll be ones that living and building the the memories through the feedback loops. They will be the ones to cultivate the trust for their people and their customer. And you cannot buy that. And you can you cannot copy that. You can only start building that by now. And by never treating the journey as finished. So, prescription is simple. Bet like a VC, upgrade for machine speed, and engineer for trust with the feedback loop from day one. Thank you. Thank you. \>\> \[applause\]