Best of LinkedIn: Martech Insights CW 40/ 41

00:00:00: provided by Thomas Allgaier and Frennis, based on the most relevant LinkedIn posts about MarTech in calendar weeks.

00:00:05: forty-and-forty one.

00:00:07: Frennis is a B to B market research company that supports enterprise marketing teams and unlocking full potential of their customer data with help from AI.

00:00:14: You can find more info in description

00:00:16: Right.

00:00:17: so welcome everyone To today's deep dive.

00:00:20: If you're listening probably have tab open right now for your primary marketing platform.

00:00:25: Well

00:00:25: almost definitely

00:00:26: Exactly.

00:00:27: And if were to log into tomorrow morning You have a pretty set expectation of what you're going to see.

00:00:32: You expect the dashboard, those endless navigation menus maybe your journey builder canvas all those reporting tabs.

00:00:39: Yeah!

00:00:40: The visible architecture I mean...you know exactly what you are paying for because you literally clicking and dragging things around in it every single day

00:00:47: Right.

00:00:48: But imagine that you log-in tomorrow And All Of That Is Just Gone Like No Menus No Canvas.

00:00:56: Well A Single Chat Interface

00:00:58: Just a prompt box

00:01:00: just to prompt box and that's really what we're digging into today.

00:01:03: We've been tracking these top martech trends across LinkedIn And the biggest takeaway by far is that this interface, you know It's uh it's actively vanishing.

00:01:13: The traditional B-to-B marketing stack isn't just getting You know a fresh coat of AI paint?

00:01:18: It's being fundamentally dismantled and rebuilt from the ground up.

00:01:21: It Is A massive structural shift.

00:01:24: And you know, the clearest signal of this actually just came out of Salesforce's Dreamforce event.

00:01:28: Oh yeah.

00:01:29: Tom Burrell shared his really perfect observation about what happened there.

00:01:32: He pointed that Salesforce essentially handed its entire user interface over to AI.

00:01:38: Right.

00:01:38: So users are now running their activities through models like Claude or ChatGPT Or Gemini and Salesforce is sort-of sitting underneath it all right?

00:01:47: Supplying data structure, permissions, rules.

00:01:50: Yeah The screen itself, UI used to justify really massive premium price tag, it's just no longer the core product.

00:01:57: Wait let me push back on that for a second though because I mean if i am an enterprise and i'm paying a mass of premium for Salesforce...and i'm not even looking at salesforce anymore aren't they just kind of commoditizing themselves into like a glorified database?

00:02:12: That's a really fair question.

00:02:14: but uh why wouldn't you just build your own database right and hook an ai up to it exactly?

00:02:19: Why Not Just Do That?

00:02:20: Because The Value Was Actually Never In The Buttons.

00:02:23: The value is in the relational data and that really deep security layer.

00:02:27: The AI is just the operator.

00:02:30: think about it like Like the evolution of aviation,

00:02:34: okay?

00:02:34: We're moving from a pilot manually pulling levers on a cockpit to a full autopilot system The engine the aerodynamics all of that is still incredibly complex and powerful.

00:02:44: And that's Salesforce.

00:02:46: but you don't touch the manual controls anymore

00:02:49: You just punch in the destination

00:02:51: Exactly.

00:02:51: You give it a destination and the agent flies to plane.

00:02:54: And crazy thing is, its not just Salesforce doing this.

00:02:57: Emil Bjarnskar pointed out that platforms like Klaviyo & Braze they're shipping this exact same pattern right now.

00:03:03: Oh absolutely

00:03:04: Klavyo literally exposing I think he said more than two hundred sixty of his internal tools directly to Claude or chat GPT.

00:03:13: Then braze has operator connect in beta which lets you plan and build campaigns straight from a prompt.

00:03:20: So the AI isn't, you know virtually moving a mouse around your screen.

00:03:23: It's directly pinging the API endpoints of these platforms.

00:03:27: so The platform has officially become the engine

00:03:30: and the agent has become the operator.

00:03:32: Yes exactly.

00:03:33: And if you're a strategic marketer listening to this deep dive right now You really have to think about that downstream effects of that because Emil highlighted A massive implication here.

00:03:43: This completely breaks the traditional CRM consulting model.

00:03:46: Oh for sure

00:03:47: I mean, for the last decade you've had these highly profitable agencies right?

00:03:51: And they built their entire business model on billing hourly rates for humans to manually click through journey builders.

00:03:56: Just setting up complex routing logic and configuring endless segments.

00:04:00: Right because it was incredibly tedious.

00:04:03: It was so tedious

00:04:04: An error prone too.

00:04:05: Basically paid someone Because we have a specific certification.

00:04:08: They knew exactly which nested menu A certain toggle is hidden

00:04:12: in Exactly.

00:04:13: But if an AI agent can execute all of that operational overhead in like two seconds via API call, those billblowers just evaporate overnight.

00:04:23: Poof!

00:04:24: Gone.

00:04:25: The real value for your team is rapidly shifting away from software operation so you're competitive advantage no longer knowing how to use the software it's strategy its data quality and governance.

00:04:38: Your teams' values basically scales with their judgment now not their headcount

00:04:42: which naturally brings us to the next really massive problem we saw across the feeds.

00:04:47: Because if that AI autopilot is flying the plane, I mean it needs flawless telemetry data right?

00:04:53: It does!

00:04:54: If the AI is the engine...the data is fuel.

00:04:57: and right now AI is pulling back the curtain on just how toxic that fuel actually is for most companies.

00:05:04: Oh yeah..it really is.

00:05:05: And Michael Notarbarto shared a brilliant takeaway from the HubSpot Unbound event perfectly.

00:05:12: He noted that AI isn't replacing marketers, it's exposing bad

00:05:16: systems.".

00:05:17: I love that!

00:05:18: It's so true and there was a quote he highlighted that i honestly think should be written on every marketing ops whiteboard...he said models depreciate.

00:05:27: context appreciates.

00:05:30: let's unpack that for a second.

00:05:31: because Because, I mean whatever LLM you are using today whether it's GPT-IV or Claude three point five It's gonna be obsolete.

00:05:43: in what?

00:05:44: six months?

00:05:45: Probably less.

00:05:45: Yeah

00:05:46: when a cheaper faster model drops

00:05:48: right.

00:05:48: so the AI itself is just a commodity.

00:05:51: but context appreciates Your proprietary customer data your historical interactions your specific business logic.

00:06:00: That is literally The only thing that actually differentiates your AI from your competitors AI

00:06:05: Exactly.

00:06:06: And you know, if your internal context is just a disaster then I mean the smartest agent in the world it's completely useless.

00:06:13: yeah Moria Fredrickson shared this highly specific and honestly kind of terrifying example at this inaction.

00:06:19: She talked about an AI agent that was deployed on a web property and he started confidently telling customers About discontinued products and totally outdated pricing.

00:06:28: now wait If i'm marketing director and hear my immediate instinct to say oh The AI hallucinated, it glitched.

00:06:35: Just turned the whole thing off right?

00:06:36: Like isn't that just a classic AI

00:06:38: failure?".

00:06:39: Well you'd think so but not at all!

00:06:41: It didn't hallucinate a single thing as Maurya pointed out... ...the AI actually did exactly what it was programmed to do.. ..it read precisely would have been published on their site.

00:06:50: .It found an old pricing PDF sitting in some forgotten directory and it just took it face value.

00:06:55: Wow So he proved that the problem wasn't a tech failure at all ,it was a governance failure.

00:07:00: So it was just confidently broadcasting their organizational mess to the entire world.

00:07:06: It just read the garbage data that they forgot to delete.

00:07:09: Precisely, it exposed this weak content operating model that frankly had no real ownership or life cycle management I mean in B-to-B.

00:07:18: we are incredibly good at operating content factories right?

00:07:21: We draft ,we design .We review ,we launch.

00:07:24: Push things out

00:07:25: Yeah, but we are terrible at maintaining or retiring that content.

00:07:29: So Moria really calls for treating content like a software product with an actual backlog strict ownership and a defined retirement point.

00:07:37: right if you're sitting there looking At your own CMS Right now just think about how many outdated product pages?

00:07:42: Or old pricing tiers or conflicting white papers Are still live because That is exactly what Your AI agent Is going to read

00:07:50: Exactly.

00:07:50: And Gareth Osborne really echoed this exact sentiment across all the major platforms.

00:07:55: He was saying, it doesn't even matter if you're heavily invested in Adobe or AWS or Microsoft.

00:08:01: The choice of which AI agent you deploy is like almost irrelevant If your underlying data architecture and permissions are fundamentally broken.

00:08:09: Yeah...and look at a bigger picture.

00:08:11: This explains why that first major wave of AI pilots in life cycle marketing kinda stalled out over last year Because the bottleneck was never the AI's capability.

00:08:21: The bottleneck is that models couldn't understand the brand, they could not securely parse fragmented customer data and just didn t fit into how cross functional teams actually operate.

00:08:31: so context really is the ultimate bottleneck

00:08:33: It does.

00:08:34: So if Context is the bottlenec we have to look at where that context lives right?

00:08:42: And this a battle getting incredibly heated in the MarTech space

00:08:46: Very heated

00:08:47: But real quick, if you want to stay ahead of these rapid structural shifts like the ones we're unpacking today.

00:08:54: You should definitely hit subscribe so you don't miss our future deep dives.

00:08:57: We are tracking this evolution constantly.

00:09:00: Definitely do that because right?

00:09:02: So we have all this data sitting in massive data warehouses.

00:09:05: Right and we have delivery channels emails The ad networks the SMS platforms but historically The missing middle.

00:09:16: Yeah, Jonathan Moran brought this up or John Moran.

00:09:20: He pointed out that a lot of enterprise stacks invest heavily in the data layer and heavily on the channel layer But they completely lack a vital orchestration layer

00:09:29: right?

00:09:29: so Data is the fuel Channels are the delivery mechanism but without a dedicated Decisioning engine these AI agents just don't have a reliable framework to operate within.

00:09:41: Now, historically that brain was supposed to be the marketing automation platform.

00:09:44: The map but you know legacy maps were built for a totally different era.

00:09:48: they were built to execute these scheduled batch and blast campaigns based on static lists exactly.

00:09:54: They were never architected to remember an instantly process every single behavioral touch point a buyer has had across Every single channel in real time.

00:10:02: because of that architectural limitation Ahmed Dattou actually argues that data platforms, places like Snowflake or Databricks might just step in and replace marketing automation platforms entirely as the agent's brain.

00:10:14: Which makes perfect logical sense when you think about it.

00:10:17: if The AI Agent needs maximum real-time context to make the best possible decision

00:10:23: well

00:10:24: You don't move the data to the agent Right?

00:10:26: You put the agent directly inside the database.

00:10:29: Right, you bring the compute to the data

00:10:31: Exactly!

00:10:31: Christopher Marriott added this by pointing out newer platforms like MotherDuck and MotherDuc literally describes itself as a Data Warehouse built natively for agents.

00:10:41: Wow Yeah The mechanism here is that the LLM is writing SQL queries directly against the warehouse on-the-fly And that threatens the standalone customer data platform's entire role.

00:10:51: I mean, if you don't need a UI to manually build user segments anymore.

00:10:55: Why do even the CDP acting as middleman?

00:10:57: It's great point But you know the execution platforms and CDPs, they are not just rolling over in letting data warehouses win this.

00:11:03: They're fighting back really hard by building their own native real-time decisioning engines.

00:11:10: Saurav Agrawal broke down CleverTep's Tesseract Decision Engine as a perfect example of that.

00:11:15: It is designed to provide live behavioral context.

00:11:18: these agents need to act instantaneously like we aren't talking about batch processing what our customer did yesterday but streaming event data this exact second.

00:11:28: Which is absolutely critical for agentic marketing, because as Sarb noted life doesn't just wait For your next scheduled batch campaign.

00:11:36: No it really does not.

00:11:37: Customer intent changes infinitely faster than Your campaign calendar.

00:11:41: You need live sub-second intelligence to actually pull This off.

00:11:45: Yes and month in Patel shared this fantastic highly tangible example of this Inaction with post scripts agentic context platform.

00:11:53: So think about a shopper named Amelia Okay.

00:11:56: She replies to a brand's automated text message saying, hey I love your foundation but i really need to travel size for a trip to Europe next month.

00:12:04: Think about the mechanics of how that used to work?

00:12:07: That text sits in an inbox.

00:12:09: A human operator eventually reads it Maybe manually tags her profile Goes into the CDP Builds a highly specific segment For europe travelers Next Month Write some copy And schedules campaign.

00:12:22: It literally takes days.

00:12:24: Best case scenario, it takes days.

00:12:26: But with an agentic context platform the AI parses the natural language of her text instantly.

00:12:32: It updates her live profile With a trip context Mathematically infers Her intent checks The product catalog for the travel size skew And then it autonomously censor Perfectly timed offer right before flight.

00:12:44: That is wild.

00:12:45: No human marketer ever built a segment, no one mapped to flowchart.

00:12:49: The agent just handled it based entirely on live streaming context

00:12:53: which completely obliterates the old way of doing things.

00:12:57: I mean if an agent can dynamically respond in real time intent like that then classic rigid rules-based marketing automation is officially dead.

00:13:05: yeah

00:13:05: and That includes the traditional MQL playbook.

00:13:07: this entire industry has run for almost two decades.

00:13:10: Oh absolutely Mike Pastore and Shai Atreya delivered very stark verdicts.

00:13:15: those massive complex decision trees you built.

00:13:17: You know the ones where if a user clicks an email, wait three days send white paper A. If they don't click Send Webinar B

00:13:24: The one that looks like a bowl of spaghetti?

00:13:25: Yes

00:13:26: That entire architecture is dead.

00:13:28: It's just too time consuming to build.

00:13:30: it ages out almost immediately and offers absolutely zero agility for the actual buyer.

00:13:35: And we are seeing direct replacement right now.

00:13:39: Vladimir Blagojevich posted about launch of Thave.

00:13:43: It's PHAVE.

00:13:45: This is a new AI native platform founded by John Miller, and you know the context here is super crucial because John Miller co-founded Marketo right

00:13:53: which is huge.

00:13:54: he essentially invented The Modern Marketing Automation Playbook And now HE is the one actively dismantling it.

00:14:00: It's a massive full circle moment for MarTech.

00:14:03: And to explain the mechanism here, let's upgrade an old analogy.

00:14:08: moving from traditional marketing automation to an AI native platform like Fave is kind of like Moving From A Train on Physical Tracks To Alive GPS Routing System Like Waze.

00:14:19: Okay I like that

00:14:20: Because a train can only go where you have manually laid the rails right?

00:14:24: If there's a blockage The train just stops.

00:14:27: That Is Traditional.

00:14:28: if Then Automation

00:14:30: Exactly, but an AI agent operates like ways.

00:14:33: It knows the ultimate destination say getting a meeting book.

00:14:37: But it looks at the live traffic It looks that the buyers real-time intense signals and it dynamically recalculates The route at every single intersection.

00:14:45: right.

00:14:46: so fave abandons those rigid flow charts entirely.

00:14:49: instead it Well, it reasons about buying groups using individual playlists for buyers.

00:14:56: So you just described the guardrails and goals in plain language, And the AI calculates the sequence of touches dynamically based on behavioral triggers.

00:15:04: And John Miller actually shared some really wild evaluation data to back up this architectural shift.

00:15:10: He had Claude The AI model objectively score eight different automation platforms against four hundred eighty one distinct enterprise requirements.

00:15:18: Wow!

00:15:19: And Fave scored a two point seven-four out three.

00:15:22: It vastly outperformed legacy systems like Salesforce, HubSpot and even his own previous creation, Marketo.

00:15:29: it's specifically dominated in AI decisioning and orchestration while the legacy incumbents basically all just tied with each other right in the middle of the pack.

00:15:37: But you know we have to be realistic here.

00:15:38: for a minute.

00:15:40: Autonomy and real-time AI reasoning that comes with really significant catch.

00:15:45: That dynamic GPS routing system is computationally heavy

00:15:50: Super Heavy.

00:15:51: And that brings us to what Bill Hobbett perfectly encapsulated as the platform tax.

00:15:56: Right, because if I'm a CFO and looking at this in thinking wait traditional automation was cheap?

00:16:02: If then statement costs like fractions of a cent in compute but having an LLM evaluate context generate natural language response decide action.

00:16:12: That means loading massive neural networks into memory, processing input tokens predicting output tokens and doing all of that for every single buyer interaction.

00:16:21: Exactly!

00:16:22: That fuel is not cheap

00:16:23: Not at all.

00:16:25: The software license is just the beginning.

00:16:27: I mean you always had to cost integration data maintenance administration but now AI token usage adds a totally new highly variable and potentially massive layer-of-cost.

00:16:39: We're paying from platform the human headcount to run it, the media to activate it.

00:16:44: And now we are paying a literal compute tax on how much AI we consume just to make it all work?

00:16:50: It adds up

00:16:52: fast does and CFOs are starting to ask much harder questions about How this specific technological overhead actually connects directly to pipeline and revenue

00:17:01: in.

00:17:01: The scary part is most marketing ops teams aren't remotely ready To answer those questions.

00:17:07: Amar Doshi shared a stat about Lean Data's new run tool that just captures the current anxiety perfectly.

00:17:14: He noted, that ninety three percent of go-to market teams have put an AI agent to work in some capacity but only thirty one per cent say their infrastructure is actually ready for it.

00:17:24: Yeah its classic operations backlog.

00:17:26: You've messy data undocumented processes and totally siloed systems.

00:17:31: Every single time you deploy a new AI agent into a broken system, You don't create efficiency.

00:17:37: No!

00:17:37: ...you just generate more support tickets and... More mess at much faster speed.

00:17:42: Lean Data built their RUN tool specifically to act as an Agentec Ops Tool To clean that backlog.

00:17:49: It uses AI to fix back-end issues And define context.

00:17:52: so human operators and AI agents are finally working from the exact same trusted facts

00:17:57: Which is so needed.

00:17:59: Which is because if your foundation isn't AI ready, Your high-speed dynamic routing system Is just going to confidently drive you straight into a ditch faster than ever before.

00:18:07: Faster Than Ever.

00:18:09: Okay, we have covered a massive amount of ground today.

00:18:11: We've moved from the vanishing user interface where platforms are literally becoming invisible engines for AI operators all The way through the absolute necessity of clean governance.

00:18:21: We explored the battle For them missing middle decisioning layer and finally the death Of those rigid marketing flow charts in favor of dynamic real-time orchestration.

00:18:30: It really is a complete foundational rebuild if how B to be marketing functions.

00:18:35: but you know before we wrap up I want to leave you with one final thought to mull

00:18:39: over.

00:18:40: Okay, let's hear it!

00:18:41: Throughout this whole deep dive we've been talking entirely about our own internal tech stacks becoming agent-driven right?

00:18:48: We are automating outbound motions Yeah But what happens when the buyers that we're targeting undergo the exact same transformation on their end?

00:18:57: What happens if our outbound AI driven marketing agents start interacting directly with our buyer?

00:19:03: inbound AI Driven procurement agents?

00:19:06: Oh wow Yeah,

00:19:08: we might be rapidly approaching an AI to AI dark funnel a scenario where our marketing software Negotiates shares gated content and argues value propositions with their procurement software Entirely behind the scenes before human buyer or human sales rep ever gets involved.

00:19:26: if The interface is vanishing for us as marketers.

00:19:28: It is going to vanish for the buyer too.

00:19:31: That is wild to think about.

00:19:33: The steering wheel isn't just gone from our car, it's gone from theirs too.

00:19:36: And the cars are negotiating with each other on the highway.

00:19:39: Definitely something to think about.

00:19:40: as you look at your own tech stack and your messy data this week If you enjoyed this episode new episodes drop every two weeks.

00:19:46: Also check out our other editions of Field Marketing Channel marketing & Partner Ecosystem AI in B-to-B Go To Market ABM & Social Selling.

00:19:55: Thank You so much for listening!

00:19:57: Remember to hit that subscribe button.