The State of AI Agents in 2026 | [Sidecar Sync Episode 124]

Summary:

 In this special solo edition of Sidecar Sync, Mallory Mejias unpacks Anthropic’s 2026 State of AI Agents report and translates what it actually means for associations. Drawing from data across 500+ technical leaders, she explores how organizations are moving beyond basic chatbots to multi-stage AI agents that handle workflows end to end. From research and reporting to customer service and coding, the results are clear: speed and quality are driving ROI—not just cost savings. Mallory also addresses the biggest barriers—data infrastructure, integration, and employee resistance—and outlines practical next steps associations can take right now to unlock their institutional knowledge and better serve members. The technology works. The real question is: are your people and your data ready? 

Timestamps:

👥Provide comprehensive AI education for your team

https://learn.sidecar.ai/teams

📅 Register for digitalNow 2026:

https://digitalnow.sidecar.ai/digitalnow

🤖 Join the AI Mastermind:

https://sidecar.ai/association-ai-mas...

🎀 Use code AIPOD50 for $50 off your Association AI Professional (AAiP) certification

https://learn.sidecar.ai/

📕 Download ‘Ascend 3rd Edition: Unlocking the Power of AI for Associations’ for FREE

https://sidecar.ai/ai

🛠 AI Tools and Resources Mentioned in This Episode:

Anthropic’s 2026 State of AI Report → https://shorturl.at/NDZqK

Claude → https://claude.ai

ChatGPT → https://chat.openai.com

Perplexity AI → https://www.perplexity.ai

👍Please Like & Subscribe!

https://www.linkedin.com/company/sidecar-global

https://twitter.com/sidecarglobal

https://www.youtube.com/@SidecarSync

Follow Sidecar on LinkedIn

⚙️ Other Resources from Sidecar: 

• Sidecar Blog
• Sidecar Community
• digitalNow Conference
• Upcoming Webinars and Events
• Association AI Mastermind Group

More about Your Hosts:

Amith Nagarajan is the Chairman of Blue Cypress 🔗 https://BlueCypress.io, a family of purpose-driven companies and proud practitioners of Conscious Capitalism. The Blue Cypress companies focus on helping associations, non-profits, and other purpose-driven organizations achieve long-term success. Amith is also an active early-stage investor in B2B SaaS companies. He’s had the good fortune of nearly three decades of success as an entrepreneur and enjoys helping others in their journey.

📣 Follow Amith on LinkedIn:
https://linkedin.com/amithnagarajan

Mallory Mejias is passionate about creating opportunities for association professionals to learn, grow, and better serve their members using artificial intelligence. She enjoys blending creativity and innovation to produce fresh, meaningful content for the association space.

📣 Follow Mallory on Linkedin:
https://linkedin.com/mallorymejias

Read the Transcript

🤖 Please note this transcript was generated using (you guessed it) AI, so please excuse any errors 🤖

[00:00:00:14 - 00:00:09:17]
Amith
 Welcome to the Sidecar Sync Podcast, your home for all things innovation, artificial intelligence and associations.

[00:00:09:17 - 00:01:44:05]
Mallory
Today we've got a special edition episode of the Sidecar Sync for you. And I know what you're thinking if you listen to the pod, I often throw around the term special edition, but I would say today actually is a special edition because one, I'm flying completely solo for this episode. And two, I don't think we've ever done an episode where we deep dive into a report like this. So I know what all of you have been thinking lately, Mallory, Amith, what is the state of AI agents in 2026? I know that's been on your mind. So we want to cover exactly that in today's episode. Enthropic actually just released their 2026 state of AI agents report. They partnered with research firm Material and surveyed over 500 technical leaders, engineering leaders, IT execs, technical decision makers, across industries and company sizes. This is gonna be real data on what's happening with agents in production. Not predictions, not hype, what organizations are actually deploying and what they are seeing as of now. I'm gonna walk through the key findings of this report and translate what it means for you all as associations. The report itself is a PDF of over 40 plus pages. And I'm sure you, like me would probably not sit down and read all 40 pages. So what we're gonna do is highlight the quickest hard-hitting facts for you in a nice audio format.

[00:01:45:08 - 00:02:53:04]
Mallory
 As I was prepping for this episode, I was thinking, you know, what good is a report like this? Obviously it gives us inspiration, it gives us guidance, it gives us knowledge on where this industry is and where it's heading, but also at the same time, and we're looking at some of these advanced technical organizations, for you as associations, you might be thinking, well, what does this have to do with me? If we had a budget of hundreds of millions of dollars, we could do very impressive things as well. And this is what it made me think of. Stay with me here for a second, Pinterest. So if you have listened to the podcast recently, you know that I just bought my first home with my husband, and it has been an incredibly exciting and stressful experience for the two of us, being first-time homeowners and just not knowing exactly what we're doing. So part of the process for me in buying a new home has been using Pinterest. If you're not familiar with the platform, you can curate photos of basically anything you can imagine. It could be fashion, it could be interior design, food, anything like that. So I've been using it to curate images of designs that I like in a home. But here's the important thing about Pinterest.

[00:02:54:06 - 00:04:26:19]
Mallory
 It's inspiration, it's a North Star, it's guidance. I'm never going to exactly copy any of the images that I'm seeing, right? Because I'm not an interior designer, I don't have access to their exact couch, their exact rug, their exact paint color. It's more of an idea of what I could do. So I take this inspiration, but then I look at my own scenario, I look at my own house and say, "Okay, how could I translate that for us in our home?" And I hope that you all can think of this state of AI agents report as the same thing, not anything that you need to copy, not beating yourself down as you're thinking, "Oh, if this organization can do this, why is my association slower to adopt this internally?" But more of a guidance, more of a North Star, more of a very nice curated image on Pinterest that you can use for inspiration. So with that, we will be breaking this episode into five parts. Let's kick it off with part one. The first part of the state of AI agents report from Anthropik is called the current landscape. That's where I want to start. Where exactly are we right now? Well, before we dive into all of that, I wanted to make sure that we're all clear on what we mean when we use the word agent. So many AI tools that you have used in the past or currently are reactive. You ask a question, you get an answer. You give it a task, it completes the task. You're driving, right? An AI agent is different. So an agent has three things that you need to be aware of. One, access to your tools and systems. Two,

[00:04:27:22 - 00:04:41:15]
Mallory
 instructions on how to behave. And then three, knowledge about your organization, your association. With those three things, it doesn't just complete tasks, it actually decides what task to do. Here's a concrete example.

[00:04:42:16 - 00:05:04:10]
Mallory
 A member emails asking about certification renewal. A basic AI tool could draft a response if you asked it to. An agent could check their membership status, look up their certification history, see that they're 30 days from expiration, draft a personalized response and flag it for followup without necessarily telling you each step or waiting for you to approve each step.

[00:05:05:10 - 00:05:32:22]
Mallory
 That's the shift this report is measuring. Organizations moving from AI that helps when asked to AI that handles workflows end to end. So now the juicy stuff. What types of agents are being deployed? The data shows organizations have moved well past simple automation. 57% of those surveyed now use agents for multi-stage workflows, so not single tasks. Multi-step processes where the agent makes decisions along the way.

[00:05:33:24 - 00:05:47:19]
Mallory
 16% have reached cross-functional processes, agents spanning multiple teams and departments. Only 10% are still doing single-step tasks, basic if this then that automation is now the minority.

[00:05:49:06 - 00:05:58:12]
Mallory
 So what is one of the most popular use cases of AI agents? Coding agents. And this matters even if you don't have a development team in your association.

[00:05:59:17 - 00:06:00:01]
Mallory
 Stay with.

[00:06:01:02 - 00:06:10:23]
Mallory
 90% of organizations are now using AI for coding. 86% are using it for production code, not just experiments or prototypes.

[00:06:11:24 - 00:06:16:22]
Mallory
 42% trust agents to lead development with humans reviewing.

[00:06:17:23 - 00:06:40:24]
Mallory
 Here's what's interesting. The time savings are nearly identical across the entire process. So for code generation, 59% time savings. For documentation, 59% time savings. For testing, 59% time savings. For planning, 58% time savings. It's not just AI writing your code faster, it's accelerating every phase.

[00:06:42:03 - 00:07:00:18]
Mallory
 So why does this matter for associations? Well, coding was the first knowledge work to get transformed by agents. If 90% of technical organizations are using agents for their core work, that wave is coming for every knowledge function, including research, member service, content development, just to name a few.

[00:07:01:24 - 00:07:08:18]
Mallory
 I wanna take a second to talk about the build versus buy dilemma and how organizations are approaching this currently.

[00:07:09:19 - 00:07:15:16]
Mallory
 47% are taking a hybrid approach. So they're mixing some off the shelf with some custom.

[00:07:16:19 - 00:07:30:24]
Mallory
 21% are buying only and 20% are building only. Of course, no single approach works for everything. Most organizations use pre-built tools where they work, then build custom where it creates real advantage.

[00:07:32:18 - 00:08:40:20]
Mallory
 My take on the current landscape of AI agents, one, I think it's helpful to think of AI agents like new hires, and we've talked about this concept before on the pod, but remember, agents need three things. They need access to your tools and systems, instructions on what to do, and knowledge on your organization, your association. So if you think of hiring a brilliant new hire from a great undergrad college, coming to your organization, and they have just one or two of these things, that would be a problem. So if they only have instructions on what to do, but they don't have knowledge on your organization or access to your tools of system, they'll be helpless. If they come in and they have tons of knowledge on your association, what you do, who you serve, but they don't have any instructions on their specific role, and they don't have any access to your tools and systems, they're not gonna be able to do much. So think of AI agents like new hires. What exactly would a new hire need to know? What would they need access to? And how could we show this new hire or tell them exact step-by-step processes that we want them to assist with? That's a helpful way to me to think about AI agents.

[00:08:41:24 - 00:09:48:06]
Mallory
 I also think this hybrid finding is hopeful in terms of the build versus buy dilemma. You don't need to build from scratch, but I think you should be involved on your part in this process. You know your association better than anyone, full stop. Your staff knows your association better than anyone. So make sure that you have a seat at the table when it comes to partnering and building new software and new technology that serves your organization. I think perhaps in the past, it has made sense to delegate all technology decisions to your technology team, to your IT team. I think now in the age of AI, when you have non-technical people, vibe coding, software that they've never done in their life, that means that all of us can have a seat at the table. It doesn't need to be an IT function anymore. It doesn't need to be a, "Well, I'm not technical, so I guess I can't be a part of that AI conversation." That's all a thing of the past. So my advice to all of you listening to this pod is have a seat at the table and be a part of that conversation if you're going to partner with an organization.

[00:09:50:02 - 00:10:37:22]
Mallory
 The 57% doing multi-stage workflows shows that agents are pass the chat bot phase, which I felt like we're kind of, we already knew, but if you have an AI chat bot on your site that's answering FAQs for you, go you. I'm not downplaying that at all, but I would encourage you to think bigger, to look ahead six months from now, a year from now. What could we be doing to better serve our members? And also something that we talk about on the pod too, Amith especially, is thinking about what barriers exist now that would prevent us from better serving our members and kind of daydreaming with that, having a vision exercise. What would that look like if those barriers were gone? Because they very well might be in six months or 12 months, especially when it comes to AI.

[00:10:38:23 - 00:11:04:16]
Mallory
 And then my last takeaway here is, don't forget your members. I know that sounds obvious, but it's easy to focus on what AI can do, what AI agents can do, but what do your members need? That should be what's driving the conversation around AI agents, not it would be neat if we could have an audio agent or a research agent or a coding agent. No, what do our members need from us? How can we better serve them? And then let's reverse engineer that.

[00:11:05:19 - 00:11:22:22]
Mallory
 Part two of the report is called going deeper. So let's do exactly that. Let's dive deeper on what is delivering value in terms of AI agents. The top planned implementations for the next 12 months. At the number one spot, we've got research and reporting at 56%.

[00:11:24:05 - 00:11:27:20]
Mallory
 Number two spot, we've got supply chain optimization, that's 49%.

[00:11:29:01 - 00:11:33:09]
Mallory
 Then we've got product development, 48% and financial planning, 47%.

[00:11:34:16 - 00:11:50:08]
Mallory
 So let's think about that. Research and reporting is the top priority. The report explains why. It spans every function, it's lower risk than customer facing work, and it builds organizational comfort with agents before moving to more sensitive areas.

[00:11:51:16 - 00:12:23:06]
Mallory
 Now let's talk about the current highest impact use cases for AI agents beyond coding. First, we've got data analysis and report generation at 60%. This is a clear leader. Next up, we've got internal process automation at 48%. And then we've got managing internal knowledge bases at 41%. I think that sounds especially useful for associations. The pattern here is that the use cases that work best are knowledge work, research, analysis and synthesizing information.

[00:12:25:00 - 00:12:29:23]
Mallory
 When asked what outcomes these organizations expected from AI agents.

[00:12:30:24 - 00:12:34:21]
Mallory
 The first, the top answer was faster task completion at 44%.

[00:12:35:24 - 00:12:49:14]
Mallory
 Then higher quality and accuracy, then improved customer satisfaction, then employee productivity. You're probably thinking, what about the obvious one? The fifth spot for expected outcomes was cost savings.

[00:12:50:20 - 00:12:59:17]
Mallory
 Organizations are prioritizing speed and quality over cutting cost. If you're pitching agents to your board as a cost cutting measure, maybe you're underselling the value.

[00:13:01:06 - 00:13:09:17]
Mallory
 Which functions are seeing the biggest impact with AI agents? We know from the first part of this episode, software development is first at 57%.

[00:13:10:24 - 00:13:25:01]
Mallory
 But the next in line for biggest impact is customer service at 55%. Hello, member service. Then we've got marketing and sales at 46%. And then another important one for associations, education and training at 36%.

[00:13:26:18 - 00:13:38:11]
Mallory
 Now, something that we hear all the time, and I especially remember hearing at previous digital nows is the conversation around return on investment for AI experiments.

[00:13:39:22 - 00:13:56:06]
Mallory
 The report is kind of hinting at the fact that maybe this ROI debate is over. 80% of those surveyed report agents have already delivered measurable financial impact. Only 20% are citing an unclear ROI as a barrier anymore.

[00:13:57:06 - 00:13:59:14]
Mallory
 We're seeing actual results being reported.

[00:14:01:06 - 00:14:37:14]
Mallory
 My take on part two of the episode, I think research and reporting being number one is huge. That is literally what associations do. Industry research, benchmarking, policy analysis, the top use case maps directly to you. And that's incredibly exciting. I also wanna remind you that AI excels at processing unstructured data. So think all of your many years of journals, think of your reports, think of legislation, which can be very hard to understand sometimes. AI excels at taking all that unstructured data and putting it into a usable format.

[00:14:38:14 - 00:15:24:24]
Mallory
 Something that we're also talking about on the podcast, which I hope to keep you all in the loop for, we're in the very early stages, is having an AI agent that does exactly this for the pod. So you all know we heavily use AI in the process of prepping for the podcast, but there's still some manual effort. I do many Google searches, then I might go to Perplexity AI and dive deeper on a topic, and then I'll go to Claude and say, these are my ideas for the topics. Can you synthesize the episode outline based on the style of previous outlines using this information? So very AI infused, but still very choppy. Me going to this platform, me going to another platform, me doing a manual Google search. So we are talking about having an AI agent, maybe that runs once a day, twice a day,

[00:15:26:01 - 00:16:33:16]
Mallory
 Google searches on its own, that's looking for the hottest items in AI, the news that's really catchy, but then also referencing those news items and comparing them to our previous episodes of the Sidecar Sync podcast, because oftentimes we see AI headlines that are interesting, but are not super relevant for you as associations, or maybe they're a bit too political leaning and we want to keep things practical and hard hitting on the pod. So having a research agent on our end that's checking the news, but also keeping in mind our most successful episodes, the formats that our audience loves the most, our previous transcripts to see if there's certain topics that Ameth and I excel at talking about more than others, and then spitting out a beautiful prepped outline at the end for us. Again, we have not done this just yet, but it's in the works in a preliminary phase. And I think, I hope that it provides some inspiration to all of you. If research and reporting being number one in terms of planned implementations, then I think that could be a really ripe area for experimentation for you all as associations.

[00:16:35:08 - 00:17:15:21]
Mallory
 Next takeaway here, data analysis as highest impact of AI agents, aside from coding. Think about how much staff time goes to pulling for reports, analyzing your member data, creating board presentations, agents can accelerate all of that. And then it got me thinking, well, what about worse? Maybe your staff doesn't even pull the reports. Maybe they don't have time. Maybe they don't know who to contact to get those. And it made me think that, and I'm guilty of this too, many times when we are planning strategically for the future of an organization or our team, or even just as an individual professional, a lot of times we have just been guessing.

[00:17:16:22 - 00:18:07:01]
Mallory
 Gut feelings, intuition, which are important. I'm not saying that. Intuition can serve you very well sometimes, but thinking about a future digital now conference for Sidecar and thinking, well, I know this keynote did well. So I think maybe a similar keynote would do well in that topic, as opposed to having data immediately accessible to me. It says X amount of people attended this keynote. In our survey that we sent afterwards, it was rated this way. And then perhaps predicting if this topic was most successful at digital now 2025, these adjacent topics might be successful at 2026. So really moving from the idea of guesswork when it comes to being strategic and moving into actual tangible data to inform your decisions, AI is making that more feasible and easy than it never has been.

[00:18:08:09 - 00:18:32:23]
Mallory
 Final takeaway here is that associations are expertise brokers. We've said this many times. The use cases that work best in terms of AI agents are your core function. You connect people to knowledge and wisdom that they cannot get anywhere else. AI can help you build a high speed train for your members to hop on and get to the information that they need stat.

[00:18:34:03 - 00:18:46:15]
Mallory
 Otherwise, they're on a nice horse and buggy. Maybe they have a nice view along the way, but it's taking them way too long to get to the information they need to better do their job. How is that fair to them?

[00:18:47:19 - 00:19:09:16]
Mallory
 I would say that it's not. So again, going back to what I said in part one, your members have to be central in this conversation. Not just what can agents do, not just, ooh, this other organization is creating a research agent. Maybe we should do that too, but what do your members need? What value do you deliver to them? And then how can AI help you deliver that value better?

[00:19:11:02 - 00:19:22:09]
Mallory
 The next part of the report is called the path forward. So basically how can we get from where we are now to where we want to be in the near term future? First, I wanna talk about the optimism gap.

[00:19:23:09 - 00:19:33:23]
Mallory
 So in terms of who is most optimistic or very optimistic about AI agents in 2026, top of that list, we've got Amith and Mallory. That's a joke.

[00:19:34:24 - 00:19:48:17]
Mallory
 We are not included in this anthropic report for now, but the top spot is enterprises, 78% of them. Then we've got mid-market, 64% of them. Then we have startups and SMBs at 38%.

[00:19:49:23 - 00:20:04:17]
Mallory
 This might be backwards from what you'd expect. Big organizations are more bullish than small ones. Why is that? They're seeing results at scale. They're not just running pilots. Confidence comes from measured outcomes, not experiments.

[00:20:06:01 - 00:20:09:17]
Mallory
 I wanna spend a little bit of time talking about the barriers that are blocking adoption.

[00:20:10:19 - 00:20:25:22]
Mallory
 One, I think is associations, this one might feel, this might resonate with you, it might feel relevant. Integration with existing systems was cited as the number one barrier, that's 46%. Then cost of implementation, 43%.

[00:20:27:02 - 00:20:32:05]
Mallory
 Next up, data access and data quality. That one might resonate too, that's 42%.

[00:20:33:06 - 00:20:44:03]
Mallory
 Then we've got security and compliance concerns. And then after that, we've got employee resistance. So I wanna look at number one and three in that list, that's integration with existing systems and data quality.

[00:20:45:04 - 00:20:59:07]
Mallory
 Both of these are data infrastructure problems. Agents can only work with what they can access. Let's go back to that, think of an AI agent like you're a new hire. A new hire can only be successful based on what they have access to.

[00:21:00:08 - 00:21:47:15]
Mallory
 If your data is scattered across disconnected systems, AI agents can't help you, or maybe they can, but maybe they can help you 10% of what they could do if they had better access to your systems. That stat should definitely concern you. I also wanna talk about this employee resistance as a barrier mentioned in the report. So for SMBs, 51% cite that as a barrier, and then for enterprises, 36%. So smaller organizations face more internal resistance, not less. The assumption that small organizations are nimble and big organizations are bureaucratic may not hold in this situation. For associations where you have had long periods of stability without asking your staff to adapt significantly, this is an incredible challenge.

[00:21:49:16 - 00:22:26:10]
Mallory
 I also wanna talk about where your freed up time is actually going after implementing AI agents. When agents handle more work, learning new skills is actually in the top spot of where your freed up time actually goes, then strategic and creative work, then relationship building. All of these sound great. These sound very human, things that we enjoy doing. I was excited to see that learning beat strategic work. Employees are upskilling more than doing higher value tasks. This is hopeful data for the AI will replace this conversation.

[00:22:27:22 - 00:22:30:05]
Mallory
 Now, what is coming in 2026?

[00:22:31:08 - 00:22:35:13]
Mallory
 81% of organizations are planning more complex agent implementations,

[00:22:36:15 - 00:23:09:01]
Mallory
 39% want to build multi-step department workflows, 29% want to deploy cross functional agents and only 3% are sitting out. Now, mind you, there's probably some bias here because if they're interviewing leaders from technical organizations, I'm honestly surprised that the numbers not 0% are sitting out. If they are more technically savvy organizations, I'm sure they are a bit ahead of the game, that's what I would think when it comes to AI agents, so keep that in mind. But the reality is almost no one is skipping this and I don't think anyone should.

[00:23:10:10 - 00:23:11:18]
Mallory
 My takeaway is for this part.

[00:23:12:18 - 00:24:04:04]
Mallory
 The top two barriers or two of the top three barriers are data problems. This is what we have been saying, I think since the inception of the SideCurse Inc podcast. Your data is scattered across your AMS, your LMS, your CRM, events, finance, so on and so forth. Building agents on top of these systems directly is like building on quicksand. They're changing constantly. You need, you need, you need a stable data layer and that is what an AI data platform provides. I'm actually just, we talk about this in the pod all the time but we just talked about this in last week's episode more in depth. So I'm not gonna spend too much time diving into that but if you want a reminder or refresher, what does Mallory mean when she says AI data platform? How is that different from a data warehouse or a data lake? Go check out last week's episode because we did a little bit of a dive on that.

[00:24:05:11 - 00:24:51:05]
Mallory
 The second barrier in that list was cost. A trend line if you're a regular listener that we've been following on this podcast is that cost is dropping rapidly when it comes to AI. So when you're planning an implementation, don't even think about honestly the current cost today. I would look ahead at six months and say, well, the cost is likely going to be a fraction of what it is today in six months. So that's what you should be planning for. And also on that note, if there's an AI agent implementation that you looked at maybe even a year ago and said that is just way too expensive in terms of using frontier models or whatever that may be, look at it again because costs are consistently dropping all the time when it comes to artificial intelligence.

[00:24:52:23 - 00:25:03:07]
Mallory
 The 51% employee resistance for smaller organizations is real, that one really got me thinking. And I feel like associations, all organizations really need to address this head on.

[00:25:04:16 - 00:25:53:04]
Mallory
 This won't come as a surprise to you, but you've got to invest in the AI literacy of your staff. Part of the reason people are resistant is because this is a technology they may not fully understand. Even us on the podcast, we don't fully understand it. It's a moving target. You might feel for one minute that you've got a solid grasp on something and then tomorrow a new model drops or a new company pops into the space, or there's a new domain specific AI that makes you rethink, well, back to the drawing board. I have to know how this works. I have to know the underlying mechanisms. The cost is dropping. We have to just keep constantly educating ourselves. And I do think your first line of defense as a leader, if you have staff that are resistant to AI is to invest in their education and invest in literacy.

[00:25:54:19 - 00:26:27:08]
Mallory
 Also helping create new pathways. That's kind of a part of upskilling, realizing that if you do add some AI agents to your member service department, adapting current member service roles and seeing where you can allow them to do more things that are more inherently human. Have deeper conversations with your members, maybe address more sensitive issues with your members with more care than they've been able to in the past because they were always crunched on time. And then of course, to be honest with your team about how rules will change.

[00:26:28:19 - 00:26:31:17]
Mallory
 If your people are fundamentally unwilling to adapt,

[00:26:33:10 - 00:26:46:13]
Mallory
 I would warn you from letting them anchor your organization in the past. I think that you all as leaders, it's our responsibility to do everything possible to bring your team along. But as Amith would say,

[00:26:47:19 - 00:26:50:04]
Mallory
 be prepared to make tough calls if that's not the case.

[00:26:51:21 - 00:27:03:12]
Mallory
 The next part of the report highlighted case studies, which is one of my personal favorite parts of reports. I like seeing what organizations are actually doing. So I selected three examples that I felt were most relevant for associations.

[00:27:04:14 - 00:27:18:03]
Mallory
 First case study here was Thomson Reuters, which we've talked about on the pod before. What do they have? Well, they have 150 years of legal expertise. They've got 3000 domain experts, massive archives of case law and legal analysis.

[00:27:19:03 - 00:27:53:06]
Mallory
 The problem, a lawyer researching a case had to manually search thousands of documents across multiple databases, hours of work, before even starting the actual analysis. So what does the agent do? A lawyer can ask a question in plain English, what precedent exists for X type of case in Y jurisdiction? The agent searches across all their archives, case law and expert analysis, then synthesizes a comprehensive answer with citations, not just search, but synthesis. The result, what took hours now takes minutes.

[00:27:54:12 - 00:28:13:08]
Mallory
 In terms of the association parallel, you're probably already have your gears turning, but this is exactly what you have. Decades of journals, proceedings, research, member expertise. Imagine a member asking, what is our industry published about workforce trends and getting a synthesized answer drawing from 20 years of your content?

[00:28:14:17 - 00:28:22:19]
Mallory
 Next step, I wanna talk about L'Oreal. The problem that they were facing, a regional marketing manager wants to know how a product performed last quarter.

[00:28:24:13 - 00:29:12:15]
Mallory
 Submit a report request, wait for the data team to build a dashboard, decisions are stalled. What the agent does, employee asks a question in natural language, what were the sales of X product in France last quarter compared to the year before? The agent figures out which databases to query, pulls the data and returns an accurate answer. No SQL, no dashboard requests, just ask. The result, 99.9% accuracy, 44,000 employees now have self-serve data. The association parallel, this is your staff member who needs member engagement data but has to wait for IT. Imagine anyone on your team asking, how many members attended events last year but haven't renewed and getting an answer in seconds.

[00:29:14:16 - 00:30:16:14]
Mallory
 The last case study I wanna cover here is the Norwegian Sovereign Wealth Fund. So the problem that they were facing as an investment analyst had to read research reports, market data, regulatory filings and multilingual news sources every day. Overwhelming volume, needed help synthesizing but couldn't sacrifice accuracy. What the agent does, analysts use the agents as a research partner. It reads across all their information sources, helps identify relevant insights and assist with analysis but humans stay in the loop for that final decision. It's classic case of augmentation, not automation. The result, 20% weekly time savings across all departments, 600 users in two months. The association parallel, this is the research heavy association where human judgment still matters. Think of your policy team monitoring regulatory developments or your research team tracking industry trends. The agent does the reading and the initial synthesis, your experts make the calls.

[00:30:17:16 - 00:30:32:12]
Mallory
 The common thread to me with all of these case studies is that all three had deep institutional knowledge that was underutilized. The technology unlocked it. And that is the association opportunity making your expertise accessible at scale.

[00:30:33:18 - 00:30:38:10]
Mallory
 These companies and these organizations show you what is possible with AI agents.

[00:30:39:16 - 00:31:40:17]
Mallory
 They should inspire you, they should guide you. And even though at the beginning of the pod, I said, you don't need to feel like you necessarily need to copy any of the things that you're hearing. I will tell you that the technology is available to you. The technology to create a research agent like the one I just mentioned is accessible to you. The technology to create an agent, a member service agent that can respond to your members, email them back, reference their renewal status, that is possible. An agent that can generate reports for you on the fly, that is accessible to you as association. Don't feel discouraged by these big companies, but also remind yourself that you are capable of doing what these organizations are doing. It is a matter of resources. And yes, I'm talking about dollars, but also time that you're willing to dedicate to it and staff involvement. So use this as your North Star and where you wanna head in the next, hopefully soon, one, two, three months.

[00:31:42:00 - 00:32:02:03]
Mallory
 I know at this point in the episode, you're thinking, "Mallory, you have thrown a lot at us and it wasn't me, it was the state of AI report from Entropic, but I wanted to spend the ending part of this episode consolidating what we've talked about and discussing ways that you can practically think about everything that we've covered. So here are some concrete next steps in my opinion.

[00:32:03:04 - 00:32:33:17]
Mallory
 If you are just getting started, I think a good first thing to do probably would be audit your data, where does your member data live? Do you have access to it, unfettered access? How many systems do you have? Can you access your data easily? If the answer is scattered or I'd have to ask IT, that is gonna be a first major problem in terms of AI implementation. Think about that new hire. Brilliant, you brought them on your team. Ooh, they know a lot about your organization.

[00:32:34:20 - 00:33:19:08]
Mallory
 You've given them some instructions on how they should respond to your members if they're in your member service department, but uh-oh, they don't have access to any data or they have to fight tooth and nail every time they want to see renewal status or look at your payment platform. That's going to be a problem. So the first step I think is audit your data. Where is it? Is there a way for you to bring it together into a consolidated unified place, like an AI data platform? Also on that note, we talked about this in the previous episode, but we have Member Junction, which is a part of our Blue Cypress family of companies. This is a free open source AI data platform that is available to anybody and the association space are outside of it.

[00:33:20:12 - 00:33:25:03]
Mallory
 Next up, I would say pick one use case. Research and reporting,

[00:33:26:06 - 00:33:59:17]
Mallory
 I think is a pretty good place to start. Of course, this is gonna depend on your organization, but I'm thinking that it is high impact and also lower risk. It's not necessarily member facing and so this could be a really good way to get you and your team comfortable with an AI agent, but ultimately think about where your team spends hours that an AI agent could do in minutes and especially an area where if they had that time freed up, it would allow them to provide higher value service to your members in other ways.

[00:34:01:07 - 00:34:32:05]
Mallory
 Lastly, you gotta start change management now. Actually, you needed to start yesterday, but start now if you haven't. Tell your team roles will change, they'll adapt and evolve. Invest in their AI literacy and create some new pathways, AI trainers, member success specialists. Don't wait until you're implementing to have these types of conversations. Now, if you're already experimenting, name your barrier. Is it still data access, integration? Is it staff resistance, get specific?

[00:34:33:18 - 00:34:47:11]
Mallory
 Try to start moving from tasks to workflows. So identify a multi-step process an agent could own end to end. I was actually talking to a meet about this yesterday in reference to us having a research agent for the podcast.

[00:34:48:18 - 00:35:43:09]
Mallory
 And he was telling, he gave me good advice, which was to go and talk to Claude or chat GBT or your preferred tool, preferably audio, because I think sitting down and typing all of this out would be very daunting, but truly have a mind dump, a brain dump is what I'm trying to say. With Claude or chat GBT and say, here's the process, I log into my computer, I go to this platform, this is the exact tool I use, then I do a Google search, then I take that Google search and I go to Perplexity AI and I drop it in there and I ask it clarifying questions. Then I think about previous episodes of the podcast, I think which of the information presented might be most relevant, then I string it together into an outline and I order the topics based on maybe most complex, the least complex or the reverse, so on and so forth. Have a conversation with an AI about a process that you want to use an AI agent for,

[00:35:44:10 - 00:36:00:11]
Mallory
 then have the AI draft a lovely document for you, a flow chart, basically full documentation of a process that an agent could own. I feel like that's a really good first step if you're not sure where to start in terms of implementing an AI agent.

[00:36:01:19 - 00:36:16:05]
Mallory
 And then my final note to you all, which I said before, but I'll double down on is don't let resistance anchor you, do what you can to bring your people along, do everything you can. But if someone's fundamentally unwilling to adapt, don't let them hold your organization back either.

[00:36:18:07 - 00:37:05:02]
Mallory
 It was a wild episode we had today, lots of information, lots of stats, lots of percentages in there, but I hope that you were able to walk away from the state of AI agent's report from Anthropik with something tangible, with something informative. If you are interested in reading that 40 plus page PDF, be my guest, the full report will be linked in the show notes. I think to wrap this up, I want to say the technology works. Like we know AI works. I don't really think there's a question about that. This report is showing us that ROI seems to be more and more proven than it has been in the past. The question is whether your team, your data, your people are ready. If you are the trusted source of expertise in your space, you owe it to your members

[00:37:05:02 - 00:37:10:15]
 (Music Playing)

[00:37:21:08 - 00:37:38:07]
Amith
 Thanks for tuning into the Sidecar Sync podcast. If you want to dive deeper into anything mentioned in this episode, please check out the links in our show notes. And if you're looking for more in-depth AI education for you, your entire team, or your members, head to sidecar.ai.

[00:37:38:07 - 00:37:41:13]
 (Music Playing)