Summary:
In this annual predictions spectacular, Amith Nagarajan and Mallory Mejias kick off 2026 by looking back at how well their 2025 AI forecasts held up (spoiler: pretty darn well) and then boldly lay out what’s coming next for artificial intelligence, associations, and knowledge work. From real-world AI recruiting agents and voice-first interfaces to the rise of AI literacy as a job requirement, the shift from SEO to AEO, and the very real possibility of AI-native platforms disrupting traditional associations, this episode is packed with practical insight and big-picture thinking. Whether you’re an association leader or just trying to stay ahead of the curve, this conversation is both a wake-up call and a roadmap for making 2026 the year you move from experimenting with AI to truly leading with it.
Timestamps:
00:00 – Welcome to 2026 & Why This Year Feels Different
👥Provide comprehensive AI education for your team
https://learn.sidecar.ai/teams
📅 Find out more digitalNow 2026:
https://digitalnow.sidecar.ai/digitalnow2026
🤖 Join the AI Mastermind:
https://sidecar.ai/association-ai-mas...
🎀 Use code AIPOD50 for $50 off your Association AI Professional (AAiP) certification
📕 Download ‘Ascend 3rd Edition: Unlocking the Power of AI for Associations’ for FREE
🛠 AI Tools and Resources Mentioned in This Episode:
Gemini 3 Pro ➔ https://deepmind.google/technologies/gemini/
Claude Haiku 4.5 ➔ https://www.anthropic.com
ElevenLabs ➔ https://elevenlabs.io
ChatGPT ➔ https://chat.openai.com
Sora 2 ➔ https://openai.com/sora
Veo 3 ➔ https://deepmind.google/technologies/veo/
NotebookLM ➔ https://notebooklm.google
Gamma ➔ https://gamma.app
https://www.linkedin.com/company/sidecar-global
https://twitter.com/sidecarglobal
https://www.youtube.com/@SidecarSync
⚙️ Other Resources from Sidecar:
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
🤖 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 - 00:00:25:17]
Amith
My name is Amith Nagarajan.
[00:00:25:17 - 00:00:27:15]
Mallory
And my name is Mallory Mejias.
[00:00:27:15 - 00:00:31:20]
Amith
And we are your hosts and welcome to 2026.
[00:00:33:09 - 00:00:36:20]
Mallory
Unbelievable, Amith. Everybody, happy New Year. How does it feel?
[00:00:36:20 - 00:00:53:05]
Amith
It is, it's kind of crazy. 2025 went by really, really quickly. And, you know, all the, all the exciting things in AI, all the fun that we're having helping our friends in the association market, it was year to zoom by. How about you?
[00:00:53:05 - 00:01:16:18]
Mallory
You know, 2025 for me was really a great year. It's been awesome, like you said, helping our association friends, getting to see everyone in person at DigitalNow. 2025 is always a highlight. That's something I'm looking forward to in 2026. And it has just been a good year for me in Atlanta, I've got to say. So if 2026 is going to be anything like 2025, I am excited and ready for it.
[00:01:16:18 - 00:01:53:17]
Amith
Well, we're officially in the second half of the decade. And we talk about educating a million association folks on AI. By the end of the decade, as our mission, we've got four years, a little bit over four years left to do it. Um, I think we're going to hit that number, Mallory. We're making great progress towards it. And, uh, it's pretty exciting. It's not a linear thing. It's obviously a growth curve, but, uh, things are moving along and the interest in AI is enormous. I'm really excited about that. Uh, just a couple of years ago, it was pretty quiet, you know, people were starting to dabble a little bit, but now it's in full blown, let's go do this thing mode, which is pretty exciting.
[00:01:55:07 - 00:02:13:14]
Mallory
And I mean, this is our third annual prediction episode. I'm sure everybody's been patiently waiting for this. So we had one for 2024, 2025, and now we're going into it with 2026. And I am pleased to present that we hit eight out of 11 of our predictions for 2025. How does that feel to me? Are you surprised?
[00:02:13:14 - 00:03:08:00]
Amith
I'm very happy to hit eight out of 11. I mean, I think some of them were probably abundantly obvious to people, but, uh, I'd say we, we did a pretty good job with that. I'm hoping that we will be as accurate this year and maybe a little bit inaccurate, because if you're too accurate on the predictions, maybe you're not thinking big enough, that would be the only downside of that. Um, I would say that my starting prediction for this episode is that we will have record attendance at digital now 2026, and it is not too soon to put this on your calendar folks. It is coming up in the Washington DC area, October 25th through 28th in DC. And again, the year is now 2026. So put that in your calendars, sign up, block off everyone in your team's calendars as well, and get there. It is going to be awesome. So that's my first prediction is we're just going to blow the doors off for digital now 2026.
[00:03:08:00 - 00:03:20:10]
Mallory
I agree with you. I, but to your point, I mean, that might be an obvious one. I feel like we've just seen steady growth with digital now. So maybe that might be a little bit too easy of a guess, but I'm going to, I'm going to agree with you on that prediction as well.
[00:03:20:10 - 00:03:23:22]
Amith
I did not need Gemini 3 pro to figure that out.
[00:03:25:06 - 00:03:38:08]
Mallory
And it makes me think also to that point of me of the idea of, of going too safe in general with piloting AI projects. We often say, look, if all your experiments and all your pilots work, well, you may not be pushing the boundaries enough.
[00:03:39:16 - 00:07:33:23]
Amith
Totally. You know, it's interesting because as 2025 comes to an end, I have been working with our team very actively on hiring a new crop of what we call technology fellows. And this is a position within Blue Cypress Labs where we hire fairly early career computer scientists and we pull them into a cohort. And what we do is really intensively train these folks over a three year window of time in a very practical way. They work on products, they work with our clients and we're hiring a group of five of these folks. And if you're not familiar with the current situation in the labor market, unfortunately it's a pretty weak labor market for, especially for candidates just leaving college. And so when we posted this position, we received an overwhelming response. I think almost a thousand applicants in a very short period of time. So we said, well, how do we solve this problem? We want to talk to as many people as possible. And on paper, many of these people look amazing, right? They all have these, these great degrees. They all worked really hard. They have good GPAs, blah, blah, blah, blah, blah. And you can't possibly talk to hundreds of people even though you'd like to. So we said, you know, what if we set up an AI agent for that, right? And there's an AI agent for that. So we built a really cool recruiting agent. And what this thing does is it basically reaches out to the candidates that pass some basic criteria and it invites them to speak to us at any time of their choosing, they just click a button, they go to our website and they can go and have an audio interview with our audio AI agent, which is using a combination of 11 labs and Claude Haiku 4.5. So it's a very high end experience. It's real time audio. They're, it's like they're speaking to a computer science professor. Um, and they're asked a bunch of questions to kind of grade their overall computer science knowledge. Um, but it's, it's a very interactive experience. So that's been pretty cool. And then we get the audio out of that, not transcript, but the actual audio. And that audio is piped over to Google's brand new Gemini 3 Pro AI model, which is extraordinary at reasoning over multiple modalities, including audio. So the audio goes in along with the rubric that is essentially evaluation criteria. It says, Hey, like evaluate this person in terms of their enthusiasm, evaluate them in terms of how quickly do they learn new concepts? Because part of what we're doing is this computer science professor in big air quotes, um, which is the AI is kind of guiding them because our goal isn't to have perfect recall of textbook knowledge, but rather how good of a problem solver is somebody. So the AI is designed to talk to them about a problem and then help them a little bit and then see if they can pick the ball up where the AI left it and then run with it from there. And many of these candidates have done an extraordinary job with that. Well, Gemini is really good at figuring that out and saying, Oh, well, you know, Mallory, uh, missed this one piece, but as soon as the AI, you know, gave her a hint, then Mallory just crushed it and ran with that ball and scored. Um, and so Gemini 3 Pro does that. And then we get this great report and then we're able to then narrow the field down to the people who seem to have the best aptitude for our environment. And then of course we actually talk to these folks. We don't actually hire them through the AI, but maybe one day, but I don't think that's going to happen, but, uh, but I just wanted to share that with our audience because I think this is super relevant to the association world. You guys are constantly out there, um, trying to match people with other people, maybe not hiring at scale for paid positions, but you're bringing on volunteers, you're matching, uh, individuals with other individuals from a networking perspective and audio modalities can be an extraordinary tool for that. So I'm just particularly pumped about this because I'm right in the middle of this process, could be more excited about bringing on a fresh crop of brand new CS grads to join our team in this fellowship position. But using AI, I was also giving these folks a flavor of what they're getting themselves into. He's like, Oh, who are these nuts that are doing these, like this weird AI interview? And, uh, it's like, well, welcome to our world. If you, if you sign up for this, this is what you're going to help us build.
[00:07:33:23 - 00:07:43:22]
Mallory
I mean, that is so fascinating. This is my first time hearing about that. Can you give us an idea of the numbers? So you said you had about a thousand people apply and then kind of how many have gone to each step?
[00:07:43:22 - 00:09:15:23]
Amith
I think, um, we probably invited maybe 250 people in that range to do the audio interview in the audio interviews. We thought they would last 15, 20 minutes, but it turns out that Claude Haiku 4.5 really enjoys being a computer science professor and makes these interviews last like in some cases, 45 minutes to an hour. Um, and they're quite involved. It's pretty interesting to listen to these, these audio transcripts or listen to the audio itself, actually. Um, so from the 250 invites we sent out, I think we had maybe a hundred or so people complete the audio interviews. Um, and then from there, you know, I went and listened to excerpts of every single one of those personally. I didn't listen to all the men to end, but, um, what part of what Gemini 3 does is it says, Hey, here's some highlights that you should listen to both positive and negative to get a sense of the candidates. So for example, when someone figures something out, do they, do they really like have that light bulb moment? Did it click and did they get excited about it? You know, were they notably, you know, uh, moving forward in their thinking after like this little missing piece was presented to them. Um, and that's part of what you get through that audio modality that you wouldn't get that from a transcript, no matter how good the LLM is. Um, you know, we talked about it in the pod, the idea of modalities, having information loss where you go from video to audio, audio to text. And, uh, this is a clear example of that where you get so much richness out of that audio. But 250 and down to a hundred. And then, and then, uh, then I think we're going to have interviews with maybe 30 people to get to five candidates.
[00:09:17:01 - 00:09:38:10]
Mallory
That's awesome. I mean, on the one hand, someone might hear that and be like, Oh, having AI involved in the recruiting process might be a missed opportunity for you to connect with people. But on the flip side of that, having 250 people be invited to do that audio interview, that's just something that would not have happened before. So you're probably getting much better exposure to the applicants than you would before.
[00:09:38:10 - 00:10:28:10]
Amith
I want to think too, and I've gotten this feedback from some of the applicants, even people that we didn't move forward with saying they were really, um, their eyes were opened by the experience because even though they're, you know, you think, Oh, these people are computer science grads. They just know all this stuff. Well, no, not at all. In fact, computer science grads, in fact, even people with PhDs in AI oftentimes have no idea what AI can do. As strange as that may sound, because their focus area might be some particularly narrow, deep discipline that is not, you know, at all about the breadth of application functionality that we talk about on this pod. And a lot of other practitioners are engaged with. So, um, it's quite interesting as a recruiting tactic, but also it's a way of helping these people in their journeys because that interview, I think could be helpful for them as they go forward, interview with other companies as well.
[00:10:28:10 - 00:10:29:04]
Mallory
Hmm.
[00:10:30:21 - 00:10:39:24]
Mallory
Well, with that, I want to dive into a recap of our 2025 predictions so we can discuss which of the eight out of 11 that we covered last year, we got correct.
[00:10:41:07 - 00:11:07:20]
Mallory
Our first 2025 prediction was that AI audio would match human latency and conversational response time. And we nailed this one. Industry benchmark is now 800 milliseconds as acceptable with some leaders in the space advertising sub 200 millisecond latency as standard. Real time voice AI with natural conversational flow is now production ready. Multiple platforms are achieving 300 to 500 millisecond latency routinely.
[00:11:09:01 - 00:11:23:19]
Mallory
Our second prediction was that a fortune 500 company would fall victim to an advanced AI scam. And unfortunately we got this one right. Goldman Sachs executives were impersonated via deep fakes to run investment scams targeting retail investors.
[00:11:24:19 - 00:11:33:18]
Mallory
Deep fake fraud losses hit $897 million cumulatively with $410 million in just H1 2025.
[00:11:34:24 - 00:11:41:24]
Mallory
North Korean operatives also infiltrated 100 plus fortune 500 companies using deep fake job applicants.
[00:11:43:08 - 00:13:14:13]
Mallory
Our third prediction from 2025 was that AI video models would produce extended high quality content. In particular we predicted that you would be able to generate a 10 minute long video from a single prompt. We didn't quite get this one obviously we made huge strides as it comes to AI video but Sora 2 for example maxes out at 15 to 25 second clips and then VO 3 which we've also covered is hovering around eight second clips at this moment. Native audio with video is now standard but 10 plus minutes from a single prompt has not yet arrived so we're going to carry this one over to our 2026 predictions. Our fourth prediction was that AI models would exhibit human like reasoning capabilities. Yes I would say we hit this one on we've seen some major advances. GPT 5 scores about a 94.6% on Amy math benchmarks. A one in three reasoning models demonstrate step by step logical problem solving the cloud for family handles sustained tasks lasting seven plus hours and models are now quote unquote thinking through problems rather than just pattern matching. PhD level performance on many technical benchmarks is what we've seen in 2025. Our fifth prediction was that we would see significant advances in predictive capabilities. I'd say we've seen substantial progress here across the board AI weather models now outperform traditional forecasting medical diagnosis AI is achieving radiologists level accuracy.
[00:13:15:14 - 00:14:27:02]
Mallory
JP Morgan estimates AI is delivering up to 1.5 billion dollars in value with fraud detection and prediction as key drivers and we've seen drug discovery timelines compressed from five to six years to one year with AI assistance. Our sixth prediction was that AI education would become a major priority and it has I would say this is a clear winner. 66% of leaders are saying they won't hire someone without AI skills AI literacy demand grew seven X over the past six years. Open AI is targeting 10 million Americans for AI certification by 2030 Microsoft Google AWS all launched AI credential programs and we've seen a 142 X increase on linkedin members adding AI skills to their profiles. Our seventh prediction is that AI agents would gain autonomy and specialization we've seen some progress here but I would stay we were a little bit early. The tools have launched we've seen you know the open AI agent framework we've seen pod computer use we've seen sales forces agent force but mainstream adoption I don't think we're quite there yet so we're carrying this one forward to 2026.
[00:14:28:05 - 00:15:06:22]
Mallory
Our eighth prediction was the rise of domain specific AI tools I would say we got this one right, this was a clear trend, for example Harvey AI which is legal reached an $8 billion valuation and it's used by 50 of the top and law 100 firms. we've seen over 1250 FDA authorized AI medical devices up from 950 and 2024 specialized AI tools are now outperforming generalist models and many domains. And we've seen finance AI deployed as scale across major banks are nice prediction was a shift from SEO to a EO or AI engine optimization.
[00:15:08:01 - 00:16:26:11]
Mallory
I would give this one like a partial success it's mixed so a EO is now a recognized concept but traditional SEO is still dominating. Organizations are aware of a EO but definitely haven't fully shifted behavior we're going to carry this one over to 2026. Our 10th prediction was humanoid robots would become more common yeah i'm going to say we got this one right Tesla optimists deployed in Tesla factories. agility digit robots were deployed at Amazon warehouses, we saw the figure robots backed by major VCs with open AI integration and Nvidia Newton physics engine collaborations were announced. Our 11th and final prediction for 2025 was that AI would fuel some significant breakthroughs in scientific research and I would say overwhelmingly we got this one right. AI developed drugs are showing 80 to 90% phase one success rates versus 40 to 65% traditionally Google deep mine continues material science discoveries. drug discovery timelines have dramatically compressed and AI models are now generating novel research hypotheses so. 8 out of 11 achieved and three partial will say that we didn't get them, but I would say partially we did that brings us to an 82% accuracy not too bad.
[00:16:27:23 - 00:16:56:17]
Mallory
Moving on to 2026 predictions, our first one is probably not going to come as a surprise to you all it's kind of a carry over from last year. AI agents will be moving from pilots to production, so we believe AI agents will transition from experimental pilots to production ready systems deployed at scale. Gartner predicts 40% of enterprise applications will feature tasks specific AI agents by the end of 2026 up from less than 5% and 2025.
[00:16:57:18 - 00:17:28:21]
Mallory
These agents will autonomously handle things like scheduling research document processing customer service and more. And for associations, this might mean deploying member service agents data analyst agents marketing agents, maybe even recruiting agents as we talked about. So me we have been bullish, I would say, to say the least on AI agents. Since we've started the podcast really, but we haven't quite seen them take associations by storm just yet or maybe you disagree with that. But why do you think that is?
[00:17:28:21 - 00:17:37:24]
Amith
Well, first of all, I agree with the prediction. I think we're going to see a tremendous amount of adoption in 2026. I believe that the reason adoption has been slower.
[00:17:38:24 - 00:18:13:13]
Amith
In 2025, let's say is two things. One is technology being ready for prime time. There's some of that, but a lot of it's just people getting ready where, you know, if you think about most people haven't really having dug into tools as simple as chat GPT and use them. And I think that's a big thing for their own tasks day to day to say like, oh, we're going to unleash a semi autonomous or even autonomous agent is a big mental leap. So I think the familiarity that people have grown with AI tools just as users of AI tools themselves is a prerequisite for organizational readiness psychologically.
[00:18:16:24 - 00:19:27:03]
Amith
What does AI do if you've never used chat GPT or cloud or Gemini as an end user, how can you conceptualize what an agent does? Let's say, for example, the member services realm. If you want to have an agent, let's say automatically answer email based on all the knowledge that you have in your organization or set up an SMS number that people can text to and do the same thing there. If you don't understand how AI is able to reason over multiple modalities of content, how it's able to summarize content, how it's able to take into account different people's requests and, you know, the context of their past requests. If you don't really have any kind of mental framework to lean on where you've done this, it would be really hard to say, I'm going to unleash a system like that. It's just it's just something that's hard for people to conceptualize. So I think that's the biggest impediment historically in the last two years. Also, the technology has gotten so much better that the accuracy level is good enough to really say, hey, like, let's just let this thing rip. So I think that's the reason why 2026 is going to be different is there's this combination of much greater understanding, awareness and appreciation of the tech and the tech being dramatically better than it was 12 months ago.
[00:19:28:19 - 00:19:36:16]
Mallory
Where do you see associations getting the most value from AI agents in 2026, kind of the options we've mentioned or maybe ones we haven't mentioned?
[00:19:36:16 - 00:21:23:18]
Amith
Well, you know, in the domains we're playing in, which is analytics with Skip knowledge with Betty, member services with Izzy, which is our new AI agent, we're launching this quarter. Personalization with the RASA platform. Those are the areas where we're primarily focused on and all of these areas have tremendous interest in them. I would tell you that the one that I probably think is going to have the most growth this year is in the area of member services or customer service in general, which could be member service, it could be event services. It's essentially the age old challenge of people asking you a question and your job is to give them a timely and high quality response. This is a tough problem to solve from the association's point of view. If you get an influx of a lot of requests and 80 percent of them, 90 percent of them are really the same thing over and over as a human. That gets kind of boring and you're copying and pasting basically your answers over and over. You don't have time to really contextualize, to think deeply, to do a lot of the higher touch types of things that you might want to do. But particularly in peak periods of time, like right before your annual conference, for example, you're probably inundated with a massive number of inquiries in the events department and the member services department. So this is an area of opportunity for something actually quite straightforward to do from an AI perspective. So it doesn't have kind of a sizzle of like solving some enormous grand challenge of biology or, you know, solving fusion or something like that. Like those big grand challenges are scientifically more complex. But sitting in our hands today is AI sufficiently good to do a truly extraordinary job of customer service of all flavors. And so I think the adoption of customer service tech is going to be great.
[00:21:24:18 - 00:22:23:17]
Amith
What I'm excited about is how the members are going to feel. So on the internal side, of course, it's great to automate customer service, portions of it. If you can provide your members with at least as good answers, preferably better, but at least as good answers as your human operators are doing, but much faster, that's good. It's helped you respond more quickly. It's helped you do it obviously more efficiently at lower cost. But from the member perspective, actually, I think people are going to prefer it. And here's why. Number one, I don't have to wait 24, 36, 48 hours to get a response, which is member services desks. If you go talk to a VP of member services or director of membership and you say, hey, what do you think is a really good service level agreement or standard that you aspire to always respond to? Is it four hours, eight hours, 16 hours, like the number of business hours, right, to respond? Typically, it's a day. Most people say, hey, within one business day, we aspire to respond to all inquiries.
[00:22:24:22 - 00:23:49:10]
Amith
Don't always hit it, but aspire to do that. And that doesn't sound terrible, but also from the member perspective, if I shoot you an email at 4.59 p.m. on a Friday afternoon and I don't hear back from you until 4.59 p.m. on Monday, that's still actual three days in human time, even though it's within eight business hours. So unless I'm asking you a basic question like, hey, what time does the meeting start? Well, the meeting started at 8 a.m. Monday. I didn't get a response till Monday afternoon, right? So it's not very helpful. So I think there's that piece of it. But the other part of it too is AI doesn't have time constraints like we do. All of us have the same amount of time every day. AI does not. And so if the AI can say, hey, I'm going to take an inquiry as mundane as what time does the meeting start and some customer service member services reps might reply to that and say, which meeting are you talking about? That's obviously a ridiculous question. You didn't tell me which meeting. But the member just makes the assumption that you know who they are and you took some time to look up in your meeting system. Well, which meeting is Malia registered for and stuff like that. So AI doesn't mind doing that every single time and making sure the response is perfect. So these kinds of systems are going to be coming online and not only saving the association time, but leveling up in terms of the quality, not just the speed, but the quality of the responses members receive. To me, that's the most exciting thing about this because then we're solving their problem faster, doing a better job. We actually rolled out a version of this at the very end of last year.
[00:23:50:15 - 00:24:26:08]
Amith
And so if you ever have a question about artificial intelligence in your association, so you're, you know, working late at night and you're wondering, hey, how can I get like really great resources and how to deploy event services, AI and my association? You can write to help at sidecar.ai. Again, that's help at sidecar.ai and our very friendly neighborhood AI agent called Izzy is going to be there to answer your question and will respond using the entire knowledge base of everything Sidecar has, knows everything Sidecar is doing.
[00:24:27:08 - 00:24:54:24]
Amith
So try it out. See what you think, but it's a cool resource and we're encouraging people to write to us more, right? Which is the antithesis of what most customer service departments measure success by. You see a massive spike in ticket volume. You're like, oh my gosh, what happened? But in this case, it's like, hey, we want to encourage that because we're actually driving so much more value. So to me, that's going to happen in that area. I'm sure there will be tons of other areas getting back to your original question of salary of growth and AI agents, but I find that extraordinarily interesting.
[00:24:54:24 - 00:26:00:16]
Mallory
Yes. And as you all know, we love to be a guinea pig here at Sidecar. So I'm excited to see how help at sidecar.ai, how that plays out with Izzy, our service agent. And we're going to keep reporting on that in 2026. So stay tuned. Our second prediction for 2026, again, not a shocker. We're predicting AI literacy becomes a job requirement. AI skills are shifting and already have from nice to have to required across knowledge work. Microsoft and LinkedIn data already shows 66 percent of leaders won't hire someone without AI skills. In 2026, we predict AI literacy will become as fundamental as computer literacy was in the 2000s. Job postings will explicitly require AI proficiency and employees without these skills will find themselves at a significant disadvantage. For associations, this creates both urgency to upskill your own staff and opportunity to provide training to your members. So, Amith, again, this is not a shock. If this is your first time listening to the Sidecursing podcast, we talk about the importance of AI literacy pretty much every single episode.
[00:26:01:18 - 00:26:06:16]
Mallory
Amith, in your opinion, what do you think baseline AI literacy looks like?
[00:26:06:16 - 00:26:58:00]
Amith
Baseline literacy is that you're actually using AI on a day-to-day basis in your job. It's not that you know chat GPT is there, but you're actually using it or some equivalent tool. So if you're not doing that, you're not literate in AI. You may have taken a class and we obviously offer a bunch of those and they're great. But if you're not actually using this stuff day-to-day, you're definitionally not literate in AI and you will very quickly find that what you learned at a theoretical level is out of date. So you have to practice it. You actually have to use it day-to-day. That to me is the fundamental idea of literacy. So if I'm interviewing someone, I'm going to ask them for examples of how they used AI in the last week and if they struggle to come up with anything. That doesn't mean I want to hire them necessarily, but I will kind of look at them going, well, do you really know what you're talking about when it comes to basic AI use? I'm not talking about tech. I'm just talking technology use cases. I'm talking about just basic office work.
[00:26:58:00 - 00:27:04:23]
Mallory
So it's less about like knowing what an LLM is and neural networks and more like practically everyday using AI.
[00:27:04:23 - 00:27:31:23]
Amith
Yeah, I think it's good for people to have the fundamentals. You know, in our course where we teach a lot of those basics to have people understand the fundamentals, it's kind of like if you're driving a car, it's good to know a tiny little bit about how the engine works and how the brakes work and all that. You don't need to be an expert in it. But if you have no idea how it works, it could limit your ability to be a really good operator of the vehicle. It's the same thing with AI. If you know a tiny bit about the fundamentals, it's helpful for how you use the thing.
[00:27:33:03 - 00:29:03:20]
Amith
To me, ultimately, though, I want to pivot just a slight bit on the topic of your responsibility as a leader, though. So it is the responsibility, the team member or the candidate for employment or the member, to seek AI learning and to make sure they're up to speed. But it's also your responsibility to make sure your people are educated. And that is important for the quality of work you do in your organization. But the absence of the availability of AI training for your team is the equivalent of malpractice as a leader, from my point of view. It's that simple. If you are not preparing your people for the future, you are not leading them or you're leading them. You're just leading them down a very bad, a very bad spot. It's called the unemployment line. And what you want to be doing is helping your people be prepared for the future, ideally in your organization. But regardless of where they go, you need to help them. And that's your fundamental job as leaders to grow your people. If you're not doing that, and if you're not doing that with AI, obviously, right now, you're really leaving them in a bad spot. So I don't think it's something that's so nice to have at all. I haven't felt that way for years. I think of this as a fundamental requirement for leaders to provide their team members with this. The good news is it takes zero dollars. You can go and provide AI training, tons of free resources online. We have tons of free resources. Obviously, if you want to spend a few dollars, there's opportunities to go for more formal types of things. But you have to do this. You've got to find a way to teach your people and help have them learn AI in a deeper way this year.
[00:29:05:05 - 00:29:17:13]
Mallory
But Amith, I, an association leader, CEO, encourage my staff to learn about AI, but I just don't want to require it because really I can't force them to learn anything. So I like to just encourage my team. What do you think about that?
[00:29:19:02 - 00:30:05:09]
Amith
I think that's you telling yourself that you can't force them. I think that as their employer, you can absolutely force them to do certain training. You force them to have training for other things that are mandatory elements of training. I think AI training at a very basic level, 100 percent should be mandatory. I think you can have the carrot and the stick. I think you could say, well, I want to encourage you to learn AI for all the right reasons and maybe even have some fun where you have prizes and rewards and things like that for people who accomplish great levels of training. The flip side is you can say, listen, no, you're not required to take the training, but you are required to take the training if you want to remain working here. It's really that simple. And so you have to be willing to take a stand on this. It's that important. This isn't some kind of passing fad. It's not some fringe thing. It's something every employee has to know.
[00:30:06:14 - 00:30:16:09]
Mallory
And then before we wrap this one up, I just wanted you to talk briefly about associations incorporating education into their member offerings in 2026 and what you think about that.
[00:30:16:09 - 00:30:29:12]
Amith
I'm excited by it. You know, we're partnering with a number of associations to do exactly that, taking our content and tailoring it for their industries, regardless of how you do it. You know, you're the association for a particular profession or industry.
[00:30:30:21 - 00:31:10:03]
Amith
Many times people look to you as the professional body that helps them with their continuing education. And so providing AI training that is deeply contextualized for your specific profession is so valuable because, yes, anyone can go online and take free training from Coursera or Microsoft or Amazon or Google. And there's some great training out there from those companies, as well as dozens of others that offer free online education. But if you provide your members with AI training that's contextualized, that has use cases in their space, it's so much more valuable. So it's a big opportunity. And I also think it's an element of responsibility for most associations.
[00:31:10:03 - 00:31:46:17]
Mallory
Our next prediction is that voice-first AI interfaces will become primary. Voice will emerge as the dominant interface for AI interactions for many use cases. With latency now consistently under 200 milliseconds and voice quality nearly indistinguishable from humans, the friction that made voice assistance frustrating has largely disappeared. We predict a significant shift from typing to talking, especially for search, customer service and quick information retrieval. Amith, what do you think voice-first means for associations and how should they be thinking about this going into 2026?
[00:31:46:17 - 00:32:49:08]
Amith
To me, it means about engaging in a natural way. Being able to speak voice-first eliminates a lot of barriers. It eliminates limitations that people might even have across cultural or language barriers because AI can be very, very good at communicating with people from any background. So it opens up access, which is exciting, but it's also so much richer than text. You're able to have a live conversation with somebody. You learn a lot more about them. You can share a lot more value back with them. And that's what AI is able to do. You know, the example I gave at the beginning of this episode of what we're doing with audio AI, both in the context of real time candidate interviews for positions we're hiring for at Blue Cypress Labs, but also audio that can later on take in audio AI that can later on take in an actual MP3 file from an audio transcript of the call and reason over it. These are tremendous things that really will allow you to do more than you've ever really imagined. So scaling up member services or scaling up any type of resources.
[00:32:50:08 - 00:33:15:15]
Amith
Audio is just kind of our native modality as a species. You know, we spoke to each other long before we wrote to each other. And so it's ingrained deeply. And for most people, they can speak and listen a lot faster than they can interact with classical computer user interfaces by clicking and typing. Some people can type super fast, but most people can speak a lot faster and listen a lot more than they can with written written content.
[00:33:15:15 - 00:33:26:03]
Mallory
Mm hmm. And I'm thinking, of course, voice makes sense for things like phone calls as well. But I mean, I don't know if you know this. Do many associations have phone like mobile apps as well?
[00:33:27:18 - 00:34:01:22]
Amith
Association mobile apps tend to be event centric. In my experience, there's a handful that have other resources on mobile apps that are used year round. But most association apps are used episodically for engaging at a conference. You'll download an association's app, use it at the conference and then be on your way. And the issue isn't so much app or not app. It has to do with how often you need to interact with your association. Most people only interact with their association a handful of times per year. So having real estate for a dedicated app on their phone typically doesn't make a lot of sense. But this, to your point, could potentially really change that.
[00:34:01:22 - 00:34:06:04]
Mallory
I think that's something certainly we'll be keeping an eye on in twenty six.
[00:34:07:05 - 00:34:54:03]
Mallory
Our fourth prediction is that open source models reach frontier parity. We think open source models will close the gap to within five percent of proprietary frontier models like GPT five and Claude. Lama, DeepSeek, Mistral and others will offer mere equivalent quality at dramatically lower cost. As a reminder, we've covered this on the pod. DeepSeek R1 was trained for just about five point six million dollars. The shifts, the strategic calculus for organizations, powerful AI without vendor lock in or massive licensing fees becomes viable. The competitive moat for closed source providers narrows significantly. So, Amith, do you think we're predicting that they reach within five percent of parity? Do you think there's a chance we see open source models surpass closed models?
[00:34:55:05 - 00:36:22:18]
Amith
There's a possibility. I think the question is, you know, how how do we even measure these things like the classical benchmarks now that we've been measuring for several years? They're being hit, you know, pretty hard in terms of, you know, you're almost at saturation, particular benchmarks that are at one hundred percent essentially and have long since been forgotten that were all the rage even a year or two years ago. So how do you go about measuring? And then benchmarks are interesting, but not necessarily relevant. What matters is the fit for use for you. So there are many open source models right now that I can use either on my computer or in cloud providers of my choosing that are just as good as frontier models. For my needs. So, you know, if you're going to try to like stretch the models for the absolute hardest problems that science knows or writing knows. Well, yes, perhaps Claude or GPT 5.2 or Gemini 3 would have an edge of some percentage points, maybe five percent, maybe maybe higher in certain tasks. But for the average user doing most things in business, I actually think frontier models and open source models have been at parity for some time in terms of the usability of these models for. For many of the basic things, of course, that doesn't apply to everything, but it applies to many of the use cases. So I think that what open source does is it pushes the frontier forward because in order to have any chance at a viable business model long term, the labs have to have something better because why would you pay a lot of money for something versus the option of free?
[00:36:23:19 - 00:37:24:03]
Amith
You wouldn't. And so the commoditization of models will continue. I think performance relative to price is going to keep having this incredible downward pressure on it where you're going to get so much more bang for the buck every single month, really. And we've seen that all year in 2025 as expected. I do think that there's a chance that open source could actually exceed frontier models in certain benchmarks because what does it mean to be open source versus what does it mean to be proprietary? It's simply as people choosing different business models. And so there may be some incredible breakthrough from some brilliant scientists in some region of the world. You don't even know where it is and that they come out with a model that blows away a frontier model. There's also fundamental research, tons of it going on in AI with different model architectures, for example. And so a lot of stuff is going to happen to 2026. And I do think that there's a very good chance an open source model or many open source models will shock the proprietary labs in a lot of ways, which is ultimately really good for everyone.
[00:37:24:03 - 00:37:27:01]
Mallory
Yeah, a little bit of competition never hurt, right?
[00:37:27:01 - 00:37:50:07]
Amith
I think this is an incredible time. The competition is really, really forging tremendous value for everyone else. The people who are making these these models, I don't know what their business models are going to look like, certainly not at the model level, not a great future there, but they're building obviously vertically above that, which is why you see open AI and Gemini and all these other people building, you know, all these apps on top of their models.
[00:37:52:01 - 00:38:37:07]
Mallory
Moving on to our fifth prediction, this one is a little bit, a little spicy, a little scary. We're predicting a first association loses significant membership to an AI alternative. An AI powered professional community learning platform or knowledge service will directly compete with and take a meaningful market share from a traditional association. Whether it's an AI that provides on demand expert answers that replace the need for peer networking or an AI driven certification that employers accept over traditional credentials, the existential threat associations have discussed will become concrete. This will serve as a wake up call for the entire association community. But can you talk about this prediction kind of on both sides of the coin as a threat, but also as an opportunity for associations?
[00:38:38:12 - 00:40:40:15]
Amith
Sure, I do think that this is going to happen and it's going to happen, not just in the association community, but in a lot of different sectors where an AI native thing, call it company, call it a project, call it a platform, eats into the market share of an established business. And in the area of content, associations sell information, essentially, where it's like bits of value as opposed to atoms, right? They're dealing with the digital world, primarily not the physical world, which obviously lends itself to disruption in this world much more rapidly. And so I think that the association purely on the membership side, as we're predicting, could see losses if their core value prop is eroded and there's free or very inexpensive alternatives that are AI based or maybe they're not even freer or cheaper or anything else along that axis. But they're just more convenient and more powerful and there's opportunities to bring information and connectivity and community to a group of people, which is what associations traditionally provide. If you can do that through AI and you can do a better job and the consumer has a better experience, of course, there's going to be a migration of people in that direction. To your other point, though, Mallory, there's no reason the association cannot be the force for that kind of disruption and drive the disruption themselves. In fact, the fact that they have incredible content in many cases, that content oftentimes is locked away and not used by anybody, but they do have it. And then in addition to that, most associations have some degree of brand strength where their brand is respected in their space. They have these two routes to power that are just sitting there and they could take advantage of that along with these disruptive forces and be the disruptor of their own business model. Unfortunately, that's traditionally very, very difficult because it requires tremendous conviction around the idea and the willingness to experiment. And so I do unfortunately think it's quite likely this year that we're going to see considerable disruption to many associations who we'd love to see disrupt themselves essentially as an alternative.
[00:40:42:13 - 00:41:20:05]
Mallory
I want to move to our sixth prediction. This is a carryover from last year, Meath. This is one we didn't quite hit on the head that AI video is finally going to crack long form content. We predict 2026 is the year AI video generation achieves five or more minute coherent videos from a single prompt, complete with synchronized audio, consistent characters and narrative flow. We've talked about Sora 2 and VO3 on this podcast and they have certainly laid the groundwork, but we believe the next iterations will extend duration while maintaining quality. This will, of course, democratize video production for associations, enabling things like conference recaps, training content and more.
[00:41:21:06 - 00:41:29:20]
Mallory
I mean, you've been pretty bullish, I would say, as well on AI video content. Do you think 2026 is truly the year that we're going to see these longer form videos?
[00:41:29:20 - 00:43:29:05]
Amith
I think we're already seeing it. I mean, VO3 is a good example. Sora 2 is pretty cool. But more importantly, the fundamentals, you have to ask yourself, well, what is video? Video is a bunch of images stitched together as frames connected with audio a lot of times, not always. And to make it make sense over a long period of time requires more compute or requires more better reasoning over time. It requires all these fundamental building blocks that we now actually have. And so as these things become less expensive, more available and there's multiple competing options, it's hard to imagine a world where this doesn't happen because the forces are already there. There is no fundamental scientific breakthrough required to do what you just said. So it's not a matter of like, do we get the science to break through where the model has the reasoning over that longer horizon, five minutes, 20 minutes an hour. We have models that can do that already and we have the tools to construct incredibly good, incredibly accurate, incredibly aligned images at scale that can form that storyboard. And then the video model kind of fills in the blanks between, you know, Nano Banana Pro, which is such a fun name. That product is really remarkable in terms of its breadth of capability from photo realistic images of people or scenes all the way to infographics that articulate, you know, business points really well. And how Notebook LM has wrapped that and created slides very easily. That's kind of the second killer app of Notebook LM is Slideshow's, which is, you know, probably better than Gamma at this point. I'm still a big fan of the Gamma app. It's a cool tool, but, you know, they're obviously leveraging Nano Banana as well. But I think Notebook LM does a great job with slide generation. My point is, is that there's not that big of a gap between what we're currently doing and what you're describing. So I think 12 months is a very long period of time with the arc of AI progress and the kind of engineering resources that are being thrown at it. And there's a lot of dollars to be, you know, be made. So there's incentive. So I think this is going to happen this year.
[00:43:29:05 - 00:44:12:15]
Mallory
OK, sounds like we've got a high degree of certainty. So we'll report back for 2027, which sounds insane to even say it loud 2027. All right. Our last prediction for 2026 is another carryover. We're going to see more of the shift from SEO to AEO. In 2026, organizations will move beyond awareness of AI engine optimization to actually restructuring content strategies around it. As chat GPT, Perplexity, Gemini and AI enhanced Google search capture more of how people find information, optimizing for AI generated answers becomes essential, not optional. Associations will need to ensure their content is structured, authoritative and formatted in ways AI systems can easily reference and cite.
[00:44:13:22 - 00:44:22:23]
Mallory
So, Amith, if you were an association CEO, would you be shifting budget in 2026 from SEO to AEO? Are we maintaining both? Are we keeping an eye? What do you think?
[00:44:22:23 - 00:47:33:08]
Amith
I really think it's one bucket. And I think that the essence of what's good SEO is actually even more true for the essence of what's good AEO, which is provide useful stuff. Don't hack it. Don't basically try to do trickery in order to get views and try to get volume. Provide high quality, relevant information in your field. Be the most truthful, best quality source of information in the world on your topic, which in the last 10 years has been true for SEO as well, because that's what drives people actually linking to you and people sharing your content and all that. But it also means in the world of AI, where there's so much more reasoning going on in the machine, the machine actually values you as a great source. And so, you know, for example, at Sidecar, we have seen conversions in our AI Learning Hub come from Chachi PT at this point. And we're optimizing for it by trying to provide really useful content to our listeners, to our readers, to our viewers on YouTube. And I think that that's actually plays very well into the strengths of what associations are about. Associations don't provide flimsy content. That's clickbaity. Typically, they provide deep content. The one thing you might have to think about is what's your strategy about how much of your content do you put outside of the paywall versus how much do you retain? So back in 2018, I wrote this book called The Open Garden Organization. What I argued for back then, I think is more true now than it was even then, which is that you have to be open minded about your audience. Don't try to typecast your audience into one particular persona where it's a member that has certain professional credentials and background. Your content and your audience, therefore, is people interested in your expertise, people who value you and respect your authority in your domain. And that can be a lot of different people well beyond what you may think traditionally. For example, if you're a particular branch of medicine, if your focus is there, you may have people that are disciplined or in that discipline specifically. But you might have a lot of other people who are not in your particular branch but might find your content useful. And if you find a way to make your content engaging and helpful, people from outside of your specialty, whether they're doctors or nurses or even patients, might find your content super, super valuable. And that might drive business value to you ultimately in the form of people signing up for different kinds of products and services that you don't think of as a membership funnel but are still of value to you. And of course, ultimately, you're moving the ball forward in terms of your mission when you do that. But you have to be a little more open minded about how much value do you give the world. I'm not saying make everything free. That is not what I'm suggesting. I'm simply saying take a good hard look at how much you expose externally and therefore is consumable by the engines and how much do you protect and keep behind that paywall for that premium upsell, which is still a good strategy. But having literally next to nothing that's consumable outside of the paywall, which is what a lot of associations do, is really not a great strategy because it means the people are the engines are going to bypass you. They're going to say, well, there's this great blogger called Mallory's blog. And on Mallory's blog, she talks about, you know, all this stuff that's super relevant to your field. She's going to end up getting a lot more traffic from chat GPT than you.
[00:47:34:13 - 00:48:15:07]
Mallory
And again, that is something we will keep reporting on in 2026. Well, Amith, every one of these predictions, I think, represents a door that is wide open for associations willing to walk through it. We've talked about agents that make your member services feel effortless, content strategies that make you the authoritative voice AI systems turn to training programs that make your members indispensable in their careers. The opportunity has never been bigger, but the window to lead rather than follow is measured in months, weeks, days, not years. So 2026 is the year to be bold. Amith, do you have a New Year's resolution for associations or the listeners of the Sidecar Sync either or?
[00:48:15:07 - 00:49:21:00]
Amith
Well, it's nothing I haven't said before, but I do think it's a really good piece of advice, which is if you haven't done this yet, get your calendar out and block off a 15 minute chunk of time, seven days a week, 365 days a year, just every day. It could be at 5 a.m. It could be at 9 p.m. It could be whatever. Block off 15 minutes with yourself to do some kind of AI learning. If you do that consistently all year long, I guarantee you that you will end 2026 in a way better place in terms of your own knowledge of AI, your own use of AI, and very likely your organization's adoption of AI at scale because you can't help to not be better if you put in the reps. It is not magic. It is not a degree. It is not some fancy title. It is doing the work. And if you do it every day and if you spend that time to learn something new for 15 minutes once a day, every day of the year, you will be world class by the end of 2026. So go do that and you will find yourself in a really good place no matter what happens with this crazy technology.
[00:49:21:00 - 00:49:28:16]
Mallory
You heard it, everybody. 15 minutes a day. We'll give you today off since it's, you know, New Year's holiday. Start tomorrow. Don't let a meet
[00:49:28:16 - 00:49:34:04]
(Music Playing)
[00:49:44:22 - 00:50:01:21]
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.
[00:50:01:21 - 00:50:05:02]
(Music Playing)