Dan Bălăceanu (Co-founder & CPO, Druid AI) argues that containment rate—the default metric on many AI agent dashboards—can reward the wrong agent. Using an example of two agents handling bank conversations (one at 96% containment, one at 80%), he shows how the agent escalating less may actually be performing worse, e.g., failing to route customers about to refinance to senior bankers. The talk presents a performance evaluation framework built with customers: seven pillars covering configuration, runtime behavior, and governance; 26 components each scored 0, 1, or 2 (missing, defined, or governed and tracked over time) and weighted by business impact, with governance and mandate weighted highest. Topics include error tolerance, risk, compliance, observability, drift, hallucination, tone and sentiment, agent mandate and scope, business case and budget, shared platforms, and outcome validation. Also covered: how a dashboard can stay green all year while a real score slides (83% to 60%), answering ROI questions, and industry-specific containment targets from Druid's 2026 AI Adoption Benchmark. Recorded at AI Engineer World's Fair 2026 in San Francisco. # Video Identity Observations: "Containment Is Lying to You: Measuring AI Agents Like Employees — Druid AI" ## Self-Description Evidence (Explicit Creator Claims) ### Who They Say They Are - Michael introduces himself as **"head of developer community at Druid AI"** and Dan as **"our co-founder and chief product officer."** - Druid is described as **"a conversation AI startup with hundreds of customers across EMIA and Americas"** that has **"been around for a few years."** - They claim a production-focused vantage point: **"because we sit with so many customers putting agents into production, we hear the same question over and over again."** ### Claimed Scope and Offerings - They claim to have built a proprietary methodology: **"we built our own reference model and measurement framework and we use it to help our customers evaluate their agents."** - The framework is presented as comprehensive: **"six pillars splitted in two groups with governance across all these pillars... all together 26 components that are observing and analyzing the agent behavior as a whole."** (Note: Dan later says "seven pillars," a minor internal inconsistency.) - They position the framework as free and shareable: **"we wanted to present it... share the framework free to use."** - They promote a data asset: **"the druid 2026 AI adoption benchmark... 15 months of the production AI agent data... free to download,"** with a QR code offered during the talk. - Physical presence at an event is stated: **"we're located at the booth LG11 right across from OpenAI Microsoft booth."** ## Intended Audience (Demonstrated) The presentation clearly targets a **business/enterprise practitioner audience**, not general consumers: - The opening scenario is framed for managers: **"You're a manager at a banking customer service center and hire two interns for the summer."** - Dan addresses people deploying agents and defending them to executives: **"once you start to engage with the business stakeholders let's say your CFO... the first question comes out is the ROI."** - He references consulting-style field experience: **"I see in my projects literally departments competing who should come first because they have a mandate from a board you should absorb AI."** - The benchmark data is broken out by enterprise verticals: **"financial services at 80, healthcare at 87, HR and IT at 93 and higher education at 99 containment rate."** ## Observable Presentation Patterns - **Analogy-driven explanation**: the intern metaphor structures the entire talk — **"Who doesn't love interns? So that's exactly how we'll walk you through today."** - **Contrarian hook**: the title and opening argue against a common metric — **"at first glance, intern B looks like the obvious hire... surprisingly, offer to intern B is the wrong offer."** - **Claims of proprietary data backing**: **"the intern was never hypothetical, that 80/20 split was a 2026 Druid AI adoption benchmark for financial services... backed by our anonymized aggregate AI agent production data."** - **Sales-adjacent education**: the talk teaches a framework while repeatedly anchoring it to Druid's customer work — **"this is a framework that we've used with multiple of our customers and it's been working out really well for us."** ## Claim vs. Demonstration Distinction - **Claimed**: hundreds of customers, proprietary framework, 15 months of production benchmark data. These are creator assertions, not independently verifiable within the video. - **Demonstrated**: the structure of the framework (weighted clusters: governance 39%, mandate 26%, platform 12%, outputs 9%), the scoring method (0/1/2 per component), and the worked A-vs-B comparison are laid out in detail during the talk. # Delivery Observation — Druid AI Talk ## Speakers & Structure Two presenters with clearly divided roles: **Michael** opens and closes (framing, story, call to action), **Dan** delivers the technical middle (framework walkthrough). Each handoff is explicitly verbal: *"Dan's going to walk you through that framework"* and *"Michael back to you."* ## Tone - **Conversational and rhetorical-question-heavy.** Both speakers rely on questions to pull the audience along: - *"Who gets a full-time job offer?"* - *"How do we actually tell if our AI agent is effective? Which numbers matters?"* - *"How do I know it's the success criteria?"* - **Slight informality in a business/enterprise register** — fillers like *"uh," "right?"* and hedged phrasing appear throughout, e.g. *"a simple analogy uh interns, right? Who doesn't love interns?"* - **Confession-style reveal** near the end lowers the register deliberately: *"a a confession to make about the opening the intern was never hypothetical."* ## Recurring Vocabulary & Phrases - **"Right?" / "here's what" / "so"** as sentence-launchers appear across both speakers: *"So here's a scenario," "Here's what had actually happened," "So let's dive into it."* - **"Containment"** is the pivot term, defined on first use: *"the percent of conversation handle without a human."* - **"Framework"** recurs constantly — roughly ten uses — as the core offering noun: *"we built our own reference model and measurement framework."* - **"Super important"** is Dan's signature intensifier, used at least three times: *"governance super important," "risk, very important, compliance," "agent mandate super important."* - **"We built"** frames everything as customer-driven: *"we build this framework sitting on six pillars"* / *"it's why we built a reference model and a scorecard."* ## Teaching Habits - **Extended scenario-first teaching.** The intern A/B story consumes the opening third before any framework content — the concrete case carries the argument. - **Narrative inversion as payoff.** The audience is set up to pick the wrong answer (*"surprisingly, offer to intern B is the wrong offer"*), then corrected — and the reveal is re-inverted again at the end (*"the intern was never hypothetical"*). - **Concrete numbers over abstraction.** Even illustrative content is quantified: *"83%," "39%," "97%," "containment... financial services at 80 healthcare at 87."* - **First-person practitioner grounding.** Dan repeatedly anchors claims in personal project experience: *"I see in my projects literally departments competing who should come first"* and *"I want to sleep at night knowing that my agent is only accessing these three endpoints."* - **Chunked framework delivery** with signposting: *"Let's look a little bit inside each cluster"*; clusters presented in sequence (governance, mandate, platform, runtime) with explicit weights. ## Communication Patterns - **Customer-consultative framing.** Both speakers position the framework as something used *with* customers: *"what customers typically tell us is that hey they finally have a some sort of framework."* - **Michael's closing follows a standard conference pattern:** value recap → free resource with QR code → booth plug (*"located at the booth LG11 right across from OpenAI Microsoft booth"*). - **Repetition for reinforcement:** the containment lesson is restated at least three times — in the story, in the final scores (*"a green containment that at the beginning looks the winner... can be a very different outcome"*), and in the benchmark data. ## Uncertainty Notes - **Dan's grammar shows non-native constructions** (*"splitted in two groups," "it's get zero"*) — noted only as an observable speech feature; no interpretation is offered. - The "signature phrase" pattern for *"super important"* is based on ~3 occurrences in a single talk — consistent within this video but insufficient to confirm it's a stable habit across talks. - Michael's segment contains more polished delivery than Dan's, but this is a single-video observation and may reflect preparation differences rather than a stable division. # Video Observation Record: "Containment Is Lying to You: Measuring AI Agents Like Employees — Druid AI" ## Speakers - **Michael** — Head of developer community at Druid AI (self-identified in transcript). - **Dan** — Co-founder and chief product officer at Druid AI (self-identified by Michael; confirmed when Michael hands off: "Dan's going to walk you through that framework"). Both speak on behalf of Druid AI, described as a conversational AI startup with hundreds of customers "across EMIA [EMEA] and Americas." --- ## Central Thesis **Containment** (the percentage of conversations handled without human involvement) is presented as a misleading primary metric for AI agent effectiveness. Both speakers argue that a high-containment agent can actually be less valuable to the business than a lower-containment one, because escalations/handoffs can be deliberately designed revenue-generating features. --- ## Position 1: Michael — The Intern Analogy and the Containment Trap ### The scenario as stated - A manager at a banking customer service center hires two interns with the same job description. - **Intern A:** contains 80% of conversations, escalates 20% to senior bankers. - **Intern B:** contains 96%, escalates only 4% — "dashboard is green." - Michael's claim: "At first glance, intern B looks like the obvious hire... But surprisingly, offer to intern B is the wrong offer." ### The stated reasoning - Intern A's example: a customer, authenticated at 7 a.m. on mobile, asks what it would take to pay off a mortgage early. The agent answers fully ("contained in every definition"), but the question signals the customer is **rethinking their mortgage — a refinance opportunity**. - The bank therefore writes a rule: *answer the customer, then route context to a senior banker*. That designed handoff is part of Agent A's 20%. - His characterization of containment: "It's easy to measure and it's on every dashboard" — implying its popularity, not its accuracy. - He frames this gap as the reason Druid built its own reference model and scorecard. --- ## Position 2: Dan — The Six-Pillar Measurement Framework *Note: Dan initially says the framework has "six pillars splitted in two groups," but later refers to "those seven pillars" (Michael's closing summary). This numerical discrepancy is preserved as expressed.* ### Structure of the framework - **Configured group:** what the agent is built on, before it runs. - **Runtime group:** tasks created, tasks executed, output. - **Governance** sits across all groups: error tolerance, risk, compliance, observability, accuracy/hallucination, tone and sentiment. - Total: **26 components** observing agent behavior "as a whole." ### Scoring method (as stated) - Each component scored: - **0** — component missing/doesn't exist - **1** — measurable/observable (manually is acceptable: "I can observe it and measure by hand equally good") - **2** — **managed** — "tracked, observed and analyzed over time," because "everything it's in a dynamic... What happens yesterday might be different today." - Components are **weighted by importance**; governance cluster weighted highest at **39%**, mandate **26%**, platform **12%**, outputs **9%**. - Explicit rationale: technical aspects "are weighted lower because we focus on the business effectiveness of the agent." ### Cluster highlights (Dan's claims) - **Governance:** Start here, not after the technical build; also useful in the envisioning phase — he reports seeing "departments competing who should come first" for AI mandates, and the framework helps decide. - **Error tolerance** is contextual, not absolute: possibly 50% is acceptable in some settings; higher targets for autonomous agents ("Should be perfect or cancel the job"); ~60% may be fine in human-to-AI collaboration where a human makes the final decision. - **Risk, compliance (SOC 2), observability/logging, drift, accuracy (3% hallucination, partly auto-detected, partly manual annotation), tone/sentiment** — all described as measured and tracked in the example. - **Agent mandate** (five elements): boundaries beyond the system prompt, hallucination control, business impact/ROI (e.g., "I expect this agent to achieve 70% contract renewal"), policies and isolated context, security ("I don't give him access to an MCP server with 4,000 tools and I don't know what he's doing"), and budget limits ("I don't want to have the surprise that one week later, my agent consumed 20K"). - **Platform:** shared memory, orchestration, and a **central handoff/action center** rather than per-agent handoffs. - **Skills/tools:** "the performer agent is not the one with the largest number of tools. It is with the effective uses of tools." - **Runtime:** logging, task creation with success criteria tied to individual outcomes ("if I look and I see that the agent renewed the contract, I want to measure if that renewal was successful or not"), explainability, and output validation. ### Applied result - **Agent A scores 83%; Agent B loses despite 96% containment** — A beat B in every cluster except skills; B "had more skills but was lack[ing] in explanability in tracking success and compliance with the governance." - Claim: the framework moves teams from debating whether an agent "feels ready" to deciding "based on mathematic rules." --- ## Position 3: Michael — Continuous Evaluation and ROI Reframing ### The interview never ends - > "When you hire a person, the interview ends, right? But when you hire an agent, it never ends." - Prescription: rerun weekly on live traffic against a human baseline — instrument, evaluate, improve, retest. - **Drift example:** the "vanity dashboard" holds flat at 92% all year (green line) while the weighted score slides from **83 to 60** (teal line) — "same agent, same green dashboard but different metrics across the time." ### The ROI fear and reframe - First stakeholder question (e.g., from a CFO) is ROI, and "the fear is that the only measurable thing about an agent... is whose job does it take away from" — a fear he says survives because "we mostly haven't measured the alternative." - Reframe: > "A good intern lightens your workload... Same test for an agent. What did it take off of your plate and what did that free the time to allow your business to do?" - The mortgage example reused: the agent hands off the balance check, so "the banker spends next morning with the customer who's rethinking their mortgage... The agent made the banker better at banking — and that's the column that belongs on your scorecard." --- ## Closing Disclosure: The Intern Was Real Data - Michael's stated confession: "the intern was never hypothetical" — the 80/20 split was a **2026 Druid AI adoption benchmark for financial services**, backed by anonymized aggregate production data. - **Containment benchmarks by industry (same metric, "four different right answers"):** - Financial services: 80% - Healthcare: 87% - HR and IT: 93% - Higher education: 99% - His interpretation: "higher ed isn't better than banking. Banking just design[ed] more handoffs in their agent AI design." Outcome weighting should be per-agent, not against one universal containment number. - The benchmark report (15 months of production data) is free to download; framework shared as "free to use"; booth LG11 noted. --- ## Tensions and Qualifications Preserved - **Pillar count inconsistency:** Dan says "six pillars"; Michael's summary says "seven pillars." Unresolved in transcript. - **Error tolerance is explicitly relativized** by Dan (50–60% acceptable in some configurations vs. high bar for autonomous agents) — not presented as a universal target. - **Containment is not rejected outright** — it is presented as insufficient alone; the escalation in the example is a *deliberate design choice*, and the framework still scores handoff quality under platform capabilities. - **The final scenario is benchmark-derived data**, per Michael's disclosure, though the two-intern narrative itself is a constructed illustration built on that data. # Video Attribution & Provenance Record **Video:** *Containment Is Lying to You: Measuring AI Agents Like Employees — Druid AI* **Video ID:** 01a15ca6-9fc6-74ad-9cca-d4197f20ab7f ## Speakers & Roles | Speaker | Stated Role | Basis | |---|---|---| | **Michael** | Head of developer community at Druid AI; opens and closes the session | Self-identified in transcript | | **Dan** | Co-founder and chief product officer at Druid AI; presents the measurement framework | Introduced by Michael; self-identified context | Both speakers present as **vendor representatives** — no guests, narrators, or third-party speakers appear. ## Affiliation & Context - **Druid AI** is described as a "conversational AI startup with hundreds of customers across EMEA and Americas." - The session is a vendor presentation of Druid's own proprietary framework (six-pillar/scorecard agent evaluation model). - No independent or third-party speaker validates the framework. ## Referenced Sources & Claims - **Druid 2026 AI Adoption Benchmark** — a report the speakers claim is based on "15 months of production AI agent data," stated to be "free to download." - Cited containment figures: financial services 80%, healthcare 87%, HR and IT 93%, higher education 99%. - The intern scenario's 80/20 split is attributed to this benchmark. - **Customer production data** — "anonymized aggregate AI agent production data" claimed as backing for the presentation. - **SOCS 2 compliance** — Dan states an evaluated agent "is compliant with SOCK 2 and back policies" (as transcribed; exact standard naming uncertain in transcript). - No external research, academic, or third-party sources are cited. ## Promotional Segments (flagged) - **Product/company promotion:** The entire session promotes Druid AI's proprietary measurement framework and consulting approach. - **Report download:** QR code / free download of the Druid 2026 AI Adoption Benchmark explicitly encouraged. - **Booth promotion:** Closing call-to-action — Druid is "located at the booth LG11 right across from OpenAI Microsoft booth," inviting attendees to visit. - Framework is offered as "free to use." ## Availability & Access Limits - No comment section availability is stated in the transcript. - No transcript-supplied metadata beyond title; speaker full names/last names, event name, and date are **not explicit** in the source (event context implied to be a conference). - Benchmark report availability beyond "free to download" is unverified in-source. ## Uncertainty Notes - Names given only as first names ("Michael," "Dan"). - "Six pillars"/"seven pillars" and cluster weightings (governance 39%, mandate 26%, platform 12%, outputs 9%) are as stated in the transcript; internal consistency is not assessed here. - No claims in this record are verified; all figures are attributed to the speakers. Morning. I'm Michael, head of developer community at Druid AI, and this is Dan, our co-founder and chief product officer. Quick context on who we are. Uh, Druid is a conversation AI startup with hundreds of customers across EMIA and Americas and we've been around for a few years. And because we sit with so many customers putting agents into production, we hear the same question over and over again. How do we actually tell if our AI agent is effective? Which numbers matters? Which artifact should we inspect? There was no clean answer out there. So we built our own reference model and measurement framework and we use it to help our customers evaluate their agents. And one of the ways we explain it to our customers is a simple analogy uh interns, right? Who doesn't love interns? So that's exactly how we'll walk you through today on this session. So here's a scenario. Uh let's take a business scenario, right? You're a manager at a banking customer service center and hire two interns for the summer. Same team, same job description for both interns. By August, here's what their dashboard says, right? Intern A contains 80% of conversation, escalates 20% to their senior bankers. Intern B contains 96% and escalates four almost nothing and dashboard is green. Now you have one return offer to give. Who gets a full-time job offer? At first glance, intern B looks like the obvious hire here, right? Intern B hired handle almost everything alone. Didn't keep pulling their senior bankers into a conversation. But surprisingly, offer to intern B is the wrong offer. Here's what had actually happened with intern A as an example. The customer is authenticated at 7 a.m. on a mobile app, asked what it would take to uh ask what would it take to pay off a mortgage early. The agent can answer the math, pull the balance, run the calculation and give a complete response contained in every definition. But that question is a signal. The customer is rethinking their mortgage. That's a refinance. That's a real revenue on the table here. So the bank writes a rule, answer the customer, then route the context to senior banker. That design handoff is part of agent A's 20%. The metric doing this is a containment uh the percent of conversation handle without a human. It's easy to measure and it's on every dashboard. This is the exact gap our customer kept hitting and it's why we built a reference model and a scorecard and Dan's going to walk you through that framework. \>\> Thank you. Thank you, Michael. Um this is the framework that we build because our customers and prospects asked us how do you know that the agent our agent it's delivering to the job? How we know that the agent is effective? So we build this framework sitting on six pillars splitted in two groups with governance across all these pillars. The configured group, it's what the agent is built off, what exists before the agent runs. Under the runtime, it's what the agent is doing at the runtime. What tasks creates, what tasks execute, and what is the output across all these groups. It's governance. From the business perspective, what is the expected boundaries? And under governance, we observe and we monitor error tolerance, risk, very important, compliance, observability, accuracy and hallucination and tone and sentiment. all together 26 components that are observing and analyzing the agent behavior as a whole. So with that framework, you stop debating whether the agent feels ready and you move the discussion based on the facts and you level the field and you can take a decision based on mathematic rules. So let's dive into it. Each individual component it's scored on three levels. It's get zero if the respective component is missing doesn't exist. It it gets one point if we can measure it. We can absorb it, observe it and we can measure it not to that not necessary automatically. So I can observe it and measure by hand equally good like it is the risk defined what is the risk I can read it I know it and the best gets two points in score if it's managed managed means it's tracked it's observed and analyzed over time because everything it's in a dynamic What happens yesterday might be different today. And to accumulate all these different metrics, we wave them based on the importance. And you observe that the governance cluster is the most important because I am analyzing the effectiveness of the agent from the business per perspective. they produce the expected business outcome. So the governance group gets 39%. Further on the mandate and we will see what it means gets 26%. Platform 12 outputs 9 and so forth. And you will notice that the technical aspects are measured are contributed but are weighted lower because we focus on the business effectiveness of the agent. Based on this metrics, we will see that a green containment that at the beginning looks the winner, it's based on a fact and can be a very different outcome when we measure all these components. Let's look a little bit inside each cluster. governance super important and many times it's forgotten to be approached from the very beginning. We focus on the technical aspects and then we tend to observe the business effects afterwards with this framework. We start with that. It helps a lot in the envisioning phase as well. Choosing what functional area to approach with AI and I see in my projects literally departments competing who should come first because they have a mandate from a board you should absorb AI. So all of them are looking to do it and the framework helps them to decide. So anytime I'm defining an agent, I start setting his boundaries. What is accepted tolerance? And you might say zero. I want to be perfect. I want to be 90%, we know that AI is not always perfect, but sometimes 50% is good as well. In autonomous agents, I want to have a higher target because there is no human in the loop. Should be perfect or cancel the job. But in a humanto AI collaboration, I know that the final decision is to a human. So AI is creating content and then human decides. So 60% error tolerance might be good risk. It is the risk defined. I am in synergies with all the other departments. So I know that my agent should run under this risk uh uh blueprint. I need to look somewhere and have these risks defined. And uh if you look at the numbers the agents we measure in here are good 97% because there is risk compliance defined for this agent. Compliance it's good. We evaluated the agent is compliant with sock 2 and back policies. So got a good score. Observability it's also good. I look the agent is logging everything from input reasoning decisions and then call integrations everything is logged good score on there drift good score we are still in the boundaries in the base established for with the first go live date it is measured and tracked accuracy 3% hallucination part of them automatically determined other set manual annotation but altogether accuracy it's measured and tracked and tone and sentiment because these agents are conversational agents I observe and extract the tone and sentiment from the conversation so all together you see how many seven very important metrics measured agent mandate super important I know we have the system prompt which says the boundaries but it's just this is just a part what is the agent definition what this agent should do the other things unlimited scope should not be done by the agent so hallucination control what's the business impact Why we are doing this agent? We are not just experimenting technology. I expect this agent to achieve 70% contract renewal. This is an ROI. We will measure it at the end to judge the business success. What are the policies boundaries based on what facts my agent should reason? Isolated context. These are the documents. These are connected systems. security, what systems this agent is allowed to engage with. Super important. I want to sleep at night knowing that my agent is only accessing these three endpoints. I don't give give him access to an MCP server with 4,000 tools and I don't know what he's doing. Isolated context. And secondly, who is allowed to engage with my agent. And last but not least, I want to stay in the budget. I don't want to have the surprise that one week later, my agent consumed 20K with whatever resources he engaged with. So all these five pillars creates the agent mandate. platform capabilities. We want the agents to work together. We don't want the agents to reinvent the wheel. If one agent is performing good because he is managing right the memory, humans handoff and orchestration on his skills, I want all the agents to behave the same. So I am measuring that as well. very important handoff. I don't want each agent to handle his own handoff. I want to have a central action center where people are engaging with all agents in my organization. So I am evaluating if each particular agent is talking with this superior uh service. I'm getting more into technical topics which are contributors to the overall score but all of them are under governance and the business purpose. So yes there are skills defined but how granular, how discoverable they are, how orchestrated, how thin or how large it's the uh the prompt combining multiple skills are the skills coming from a central repository are built on the spot or are human coded. What tools are available? You will see at the final judgment that the performer agent is not the one with the largest number of tools. It is with the effective uses of tools. So very important this aspect as well. And with this I concluded the definition. But what happens at the runtime? I need to log. I need to create tasks based on the mandate and then execute the tasks. Are the tasks written? Is the input logged? Is the required context present? How do I know it's the success criteria created together with the task? Because if I look and I see that the agent renewed the contract, I want to measure if that renewal was successful or not. If I don't have the success criteria linked with that individual contract, it will be impossible to judge. So the agent should explain everything he's doing. This is what the task topic is. And at the end, I want to validate the output. there is one single score that evaluates if the agent was successful or not. So in summary, combining all these metrics, agent A scores 83%. If you remember at the beginning, his containment was 80%. Escalated 20 times. Agent B escalated just four times. However, observing and measuring the governance, the mandate, the final outcome, the final score, it's in favor of agent A. And if you look at the clusters, in each individual cluster, A was better than B except the skill. B had more skills but was lack in explanability in tracking success and compliance with the governance. So all together this framework gives the true winner Michael back to you. \>\> Thank you. Thank you Dan. So those seven pillars and the scorecard uh that's the framework and this is a framework that we've used uh with multiple of our customers and it's been working out really well for us and what customers typically tell us is that hey they finally have a some sort of framework that they can use to measure the effectness of the agents that deploying and be able to have a discussion intelligent discussion with their business stakeholders. So here's where going back to the inter uh intern metaphor breaks right in your favor. Uh when you hire a person, the interview ends, right? But when you hire an agent, it never ends. You rerun it every week on a live traffic against a human baseline instrument, evaluate, improve, retest. Again, going back to what Dan said earlier, the drift, right? The green line you see on the right here is the um the vanity dashboard as flat 92% all year. The teal line is the wheel score sliding from 83 to 60 because you're measuring it on a weekly basis. So same agent, same green dashboard but different metrics across the time. So now one of the things especially if you once you start to engage with the business stakeholders let's say your CFO or something you know the first first question comes out is the is the ROI right and the and the fear sets in the fear is that only measurable thing about an agent that you're working on you're deploying is like whose job is it does it take away from right and that fear survives because the we mostly haven't measured the alternative alternative. So here's a reframe that we try to typically work with our customers. A good intern lightens your workload. The routine works moves off to your desk and you get sharper at the work that's left. Right? Same test for an agent. What did it take off of your plate and what did that free the time to allow your business to do? Right? The agent hands a balance check at 7 and 9. So the banker spends next morning with the customer who's rethinking their mortgage. The agent made the banker better at banking and that's the column belongs on your scorecard and that's some of the things that we try to work with our customers as well. Now a a confession to make about the opening the intern was never hypothetical that 8020 split was a 2026 root AI adoption benchmark for financial services. So everything you just heard is backed by our anonymized aggregate AI agent production data. Uh the same metric containment four different industries four different right answers financial services at 80 healthcare at 87 HR and IT at 93 and higher education and 99 containment rate. Now when you look at the data higher ed isn't better than banking. banking just design more handoff in their agenting AI design that's what you wait the outcomes per agent against one universal number the numbers different across different industry this is the druid 2026 AI adoption benchmark uh we gathered 15 months of the production AI agent data and uh it contains many useful data uh across different industry as well it's free to download and if you Look at the next slide here. this slide uh this is where you can um QR code you can scan it you can uh download the report and it's the framework that we developed is really about defining the mandate contract the boundaries score the whole agent run it weekly uh against the human baseline and that's a framework that we typically uh work with our customers to kind of so we wanted to present it you all kind of share the framework free to use and uh one Last thing is uh we're located at the booth LG11 right across from OpenAI Microsoft booth. So feel free to come by and I would love to have some conversation with you all. Thank you very much. \>\> Thank you. \>\> \[music\]