Welcome back to The Hypha Wire! Spring has officially sprung, and so has HubSpot’s Spring 2026 Spotlight.
HubSpot dropped over 90 updates last week, but five of them really stand out: HubSpot AEO, a major Breeze Assistant upgrade, Smart Deal Progression, and expanded capabilities for Prospecting Agent and Customer Agent.
The headline is HubSpot AEO. It tracks how your brand shows up in AI-generated answers from ChatGPT, Gemini, and Perplexity. The pitch from HubSpot is straightforward: organic search traffic across their customer base is down 27% year-over-year, while AI referral traffic has tripled. And that AI traffic is converting at a higher rate.
Here’s where it gets interesting. Most AEO tools make you guess which prompts your buyers are actually using. Guess wrong, and you’re optimizing for queries that don’t matter. The embedded version in Marketing Hub Pro and Enterprise uses your CRM data to surface the prompts your actual customers are likely to use. That’s a gap standalone tools can’t close without access to your pipeline.
Smart Deal Progression caught my attention too. It’s not just summarizing calls. After every meeting, it analyzes the transcript alongside the full deal history and suggests CRM updates, drafts follow-up emails, and lists action items. What makes it different is that it applies your pipeline definitions when making recommendations. It’s not suggesting a stage change based solely on what was said in the last call. It’s using how your team actually defines progression.
The smaller updates matter too. Reddit integration, deal properties in forms, importing historical emails, custom reports on CRM records. These are the things that make the platform easier to use day to day.
The common thread across everything HubSpot announced is context. AEO uses your CRM to generate relevant prompts. Smart Deal Progression uses your pipeline definitions. Prospecting Agent uses your buying signals. Customer Agent draws on the full customer relationship. Every one of these features works better when your underlying data is clean, structured, and consistently maintained.
We’ve been tracking the AI visibility shift for a while now. Before HubSpot shipped a native AEO solution, we were building internal frameworks to monitor how clients appear in ChatGPT and Perplexity. This release gives every marketing team that visibility natively. If you want to understand whether your portal is set up to actually get value from what shipped this spring, let’s talk!
-Sage Levene, VP of Marketing, Hypha HubSpot Development
Open Mic
Your CRM Isn’t Messy, It Grew Without a Plan
By Deborah Mackey, Data Architect, Hypha HubSpot Development
If your CRM feels messy, hard to trust, or slightly unpredictable...it’s probably not because anything is “broken.” It’s because it grew without a plan. And to be fair, that’s how most systems evolve.
No one sets out to build a complicated or confusing CRM. It usually starts with a solid foundation, clean fields, a few workflows, maybe one integration. Everything makes sense in the beginning.
Then the business grows.
New campaigns are launched. New tools are introduced. Teams expand. Processes evolve. And with each of those changes, something new gets added to the system to support it: a field here, workflow there. An integration to connect another platform.
Individually, every decision is reasonable. In fact, most of them are necessary. But over time, those layers start to stack.
What was once a simple, easy-to-follow system becomes something more complex. Not because of one big mistake, but because of a hundred small ones that were never fully aligned.
Fields get reused for slightly different purposes depending on the team.
Workflows begin to overlap or unintentionally conflict with each other.
Integrations start writing data in formats that weren’t originally planned or documented.
At a certain point, the system still works...but it doesn’t feel predictable anymore. And that’s usually when we start hearing things like:
“We’re not sure which field is right.”
“I don’t want to touch that workflow in case it breaks something.”
“The data looks off, but we can’t tell why.”
Not exactly the kind of confidence you want in a system that’s supposed to support your entire business.
This is the difference between building a system and architecting one.
Building is additive. It solves immediate problems. It helps teams move quickly. Architecting, on the other hand, is about alignment. It ensures that everything being added fits into a clear, intentional structure.
Without that layer of alignment, complexity isn’t just possible, it’s inevitable.
A well-architected CRM doesn’t mean fewer tools, fewer workflows, or less automation. It means everything has a defined purpose and place.
Each field has a clear meaning and owner.
Each workflow has a specific role and doesn’t compete with others.
Each integration fits into the broader system instead of operating independently.
When those pieces are aligned, something subtle but powerful happens.
The system becomes easier to understand.
Changes feel safer to make.
And most importantly, people start trusting the data again.
It’s not the kind of work that immediately stands out. No one logs in and says, “Wow, this architecture is beautiful.” But they do notice when things just...work.
Reports make sense.
Automation behaves the way it should.
Teams stop second-guessing the system and start relying on it.
Most CRM issues aren’t caused by bad decisions. They’re caused by growth without alignment. And the fix usually isn’t adding more. It’s stepping back, understanding how everything connects, and making what’s already there make sense again.
Interesting update now being rolled out to all LinkedIn members. For discoverability, we can now think about writing our content with natural language.
From the first piece:
“LinkedIn’s search bar will now be able to provide results based on any criteria users enter into the prompt.
“So now, users can describe what they’re looking for, and LinkedIn’s search system will surface contextually relevant results, based on a broader understanding of language.”
Hypha Highlights
PE firms that choose HubSpot for portfolio operations face a follow-up decision that’s just as consequential as the platform selection itself: how to deploy it across 5, 10, or 20 companies that were never built to operate the same way.
The platform decision gets most of the attention. The architecture decision determines whether it actually works.
In many PE implementations, a fund commits to HubSpot, rolls it out to three portfolio companies, and within six months discovers that inconsistent pipeline stages, mismatched lifecycle definitions, and incompatible reporting structures have recreated the exact visibility problem the CRM was supposed to solve. The tool is right. The deployment model is wrong.
This is a comparison of four standardization approaches PE firms use when deploying HubSpot across a portfolio—what each model enables, where each breaks down, and how to decide which fits your fund’s operating thesis.
Stanford HAI released the 2026 AI Index Report showing AI capability is accelerating rather than plateauing, with industry producing over 90% of notable frontier models in 2025. The report found the U.S.-China AI model performance gap has effectively closed with models trading the lead multiple times since early 2025, while documented AI incidents rose to 362 (up from 233 in 2024), responsible AI benchmarks lag behind capability benchmarks, and the number of AI researchers moving to the U.S. has dropped 89% since 2017 with an 80% decline in the last year alone.
Anthropic launched Claude for Word, a Microsoft Word add-in that works directly inside documents rather than in a separate window, allowing users to select text and receive edits as tracked changes, leave comments for Claude to address, draft content in existing templates with citations, check for consistency errors like broken cross-references, and save workflows as reusable “skills” for team-wide deployment.
Additionally, Anthropic has released Claude Opus 4.7, an upgrade to Opus 4.6 that, per the company, brings notable improvements in software engineering, instruction following, and vision capabilities, including support for images more than three times the resolution of previous models.
A leaked internal memo from OpenAI’s Chief Revenue Officer outlines the company’s Q2 strategy, including a new model codenamed “Spud,” an enterprise agent platform called “Frontier,” and an expanded Amazon partnership. The memo also takes direct aim at Anthropic, claiming its reported $30 billion run rate is overstated by roughly $8 billion due to how the company accounts for revenue share payments, according to OpenAI’s own analysis.
Google Chrome’s new “Skills” feature lets users run repeatable AI prompts through the Gemini sidebar with a simple keyboard shortcut, with a library of more than 50 presets covering tasks like summarizing YouTube videos or evaluating job listings. Users can also build their own custom Skills through Gemini.
Plus, Google has launched a native Gemini app for macOS, giving users AI assistance directly on their desktop via a global keyboard shortcut without needing to open a browser tab. The app, available for free to all Gemini users on macOS 15 and up, can read what’s on your screen for contextual help and supports image and video generation through Google’s Nano Banana and Veo tools, according to the company.
Allbirds, the wool sneaker company that once topped a $4 billion valuation before selling this month for $39 million, is planning to pivot into AI compute infrastructure and rebrand as “NewBird AI,” according to the Financial Times. The company says it intends to raise $50 million to pursue GPU-as-a-Service offerings, pending shareholder approval at a May 18 vote.
“Marlene Zuk gives us a new appreciation for the animals we often shun, explaining why these unpopular creatures have something special to teach us not only about the ways we deal with other species but about our own place in nature and what it means for an animal to belong somewhere.
“Writing with an infectious blend of humor and curiosity, Zuk invites us to reflect on our relationships with these close-to-home creatures and the ways our lives encroach on theirs, and to draw lessons from their behavior in all its fascinating complexity.”
How can we help you?
Case Study: Plastic Surgery Marketing with HubSpot
Three plastic surgery practices came to Hypha with different priorities:
One needed to be found: full website rebuild, stronger organic traffic.
One needed more consultations to support a multi-physician team.
One needed email to stay in front of patients during a long decision cycle.
Same vertical, similar patterns, but different pressure points.
Practice A rebuilt on HubSpot CMS, with interactive before-and-after sliders on every procedure page. Patients could see results directly instead of imagining them. Consultation forms routed into a HIPAA-compliant platform. Content mapped to how patients actually search at the procedure level. Organic traffic grew by roughly 30x over the engagement window.
Practice B focused on lead generation. Website, SEO, and content all aligned around consultation volume, with Marketing Hub configured to support it. Lead flow was steady and predictable.
Practice C already had traffic but needed to stay relevant over time. Email campaigns built around visual results kept the practice in front of patients who were taking months to decide.
Three practices. Three scopes. Each got what they needed.
The fundamentals were consistent: visual proof, clear consultation paths, and procedure-level SEO.
Practice thinking about a website rebuild, lead generation push, or email program? Contact Hypha to talk through where your digital infrastructure is working and where it isn’t.
