Welcome back to The Hypha Wire, and happy Friday (the 13th + Galentine’s Day)!
We shared a blog last week about AI coding tools and enterprise platforms, and the message feels important enough to reiterate today. There’s been a lot of chatter lately about “vibe coding” and AI-generated apps replacing workplace software. From where we’re sitting, that’s not where this is headed.
The disconnect comes from mixing up two different things: code generation and systems architecture. AI tools are genuinely impressive at rapid prototyping and building single-purpose apps. What they can’t do is replicate the organizational infrastructure that makes software work at scale (things like multi-user governance, audit trails, integration ecosystems, compliance frameworks, disaster recovery).
Think about what actually goes into enterprise software. When 500 people need simultaneous access with different permissions, when you need complete audit trails for regulated industries, when you’re connecting to dozens of external systems—it’s not a coding problem. It’s systems engineering that requires understanding how businesses operate.
We see this in our client work: manufacturing workflows tracking parts across production systems in real-time, edtech CRM migrations that have to preserve complex institutional relationships, cybersecurity pricing calculators translating years of spreadsheet knowledge into automated systems, etc. You can’t just prompt these into existence.
What AI tools actually change is speed. Custom integrations that took months now take weeks, and platform customization gets more flexible. But you still need the operational foundation, compliance structure, security certifications, and the continuity planning that platforms provide.
Companies who ditch established platforms for spun up AI versions will fall behind. And the companies who succeed will be the ones using AI within their platforms to get the most value out of what they already have.
-Sage Levene, VP of Marketing, Hypha HubSpot Development
Open Mic
Finding the Right Platform To Host Your Hubspot Integration
By Zachary Lyons, Integrations Developer, Hypha HubSpot Development
When helping clients configure HubSpot for their business, we encounter limitations in the platform. So you have a developer write some code that uses HubSpot’s API to make up for those limitations and get the data to do what the business needs. Where will that code live? It can be tough choosing the right platform to host integration code. Let’s cover three different integration platforms (embedded, no-code, and cloud hosted) and examine the strengths and weaknesses of each.
Option 1: Host on HubSpot
If you have access to HubSpot Data Hub Pro, you can run integration code in HubSpot Workflow custom code actions. The advantage of keeping integration code here is everything lives on the HubSpot platform, tidy! There are many record lifecycle events that you can use to trigger your code. There are limitations however: long running jobs, orchestration, and price. HubSpot Workflows only have 20 seconds of runtime, and for complex jobs or slow API’s you may need more time than that to accomplish your task. Cost is also a consideration: upgrading HubSpot subscriptions may not be in the cards for smaller teams, meaning we need to look elsewhere for places to host our code.
Option 2: Integration Platform as a Service tools
Integration platforms like Zapier and Make can let you build the integrations you need. They often come with pre-built integrations that you can use out-of-the-box and let you blend no-code and coded solutions. I find these platforms are more affordable than some HubSpot subscriptions and they have very user-friendly interfaces that walk you through setting up your integration.
It’s worth noting that these platforms have their own limitations, too. They may not be able to trigger off the right HubSpot event you need or limit how long the code blocks can run (30 seconds for Zapier, for example, which still beats HubSpot’s 20 seconds).
Option 3: Cloud Platform
Finally we arrive at the most complex option, but in my opinion the best! Cloud platforms, when set up correctly, are more powerful and cost efficient than all previous options. They can run code for as long as you need, leave logs to let you see what went right or wrong with your integration, and are able to scale to meet the needs of your business. They can be a little intimidating to navigate and maintain however, so there are clients I wouldn’t recommend go the cloud route.
Consider what your team needs the most when picking your hosting provider: you may have to choose between keeping the integration user friendly, or going somewhere that has the horsepower to do what you need.
It’s important to review the basics every now and again to make sure your content is satisfying what SERPs like Google want (not just what AI engines want).
“It’s not an algorithm or ranking factor you can directly optimize for. Rather, it’s a conceptual framework that shapes how Google’s engineers and quality raters think about what makes content deserving of high rankings.
“In short, the quality raters use E-E-A-T to evaluate websites, and those scores are then used by Google to improve and train its algorithms.”
Hypha Highlights
Do us a favor: Think back to the last time you made a major purchase. The odds are that you didn’t settle for the first option. You likely considered your needs and pain points, conducted basic research, and narrowed your choices before making a decision.
The buyer’s journey is the process prospects go through from recognizing a problem to selecting a solution. It consists of three stages: Awareness (identifying the problem), Consideration (evaluating solutions), and Decision (choosing a partner).
According to a Gartner report, “only 17% of the total purchase journey is spent in interactions with sales reps.” Modern buyers use search engines, AI tools like ChatGPT, peer review platforms, and LinkedIn discussions to self-educate—fundamentally changing how companies need to create content for each stage.
On LinkedIn she says of it, “Our point of view is this: for AI to drive outcomes, it needs shared context. And shared context is the combination of data about your customer, knowledge about how your team and business actually work, and the ability to learn. That is what will help humans and AI agents to work together as teammates.”
AI in Action
News, updates and tools from the AI industry.
What didn’t happen in the AI world this week? Let’s dive in.
Several AI figures have resigned from major companies with warnings/concerns about the industry. Anthropic AI safety leader Mrinank Sharma quit with a message that “the world is in peril,” according to the BBC. Former OpenAI researcher Zoë Hitzig resigned over ChatGPT ads, writing that OpenAI is “making the mistakes Facebook made” by exploiting users’ private conversations. Meanwhile, the Financial Times reports a sixth co-founder of Elon Musk’s xAI is leaving following an exodus of more than half a dozen researchers amid tensions as Musk merges xAI with SpaceX.
The Super Bowl was flooded with AI commercials, but viewers weren’t buying the messaging, according to Rolling Stone. Ads from Ring, Google’s Gemini, and Amazon’s Alexa+ drew immediate criticism on social, with experts saying the “AI optimism at all cost” was “out of touch with public sentiment.”
One Super Bowl commercial in particular caught my attention, for ai.com, which announced “AGI is coming.” Crypto.com CEO Kris Marszalek purchased the domain for a record-breaking $70 million and launched it with the commercial. The website currently allows users to reserve usernames for themselves and their AI assistants, with Marszalek describing it as an AI assistant platform with apparent social media aspects, though details remain unclear beyond username reservation.
Plus, Anthropic’s Super Bowl commercials mocking OpenAI’s plans to introduce ChatGPT ads haven’t deterred the company. OpenAI proceeded with its ad rollout, stating that ads will be clearly labeled, won’t influence ChatGPT’s answers, and will remain private from advertisers, though critics fear ads could still affect the user experience.
Intensifying their rivalry, Anthropic and OpenAI released competing AI models on the same day. Anthropic unveiled Claude Opus 4.6 with improved office productivity and coding performance plus an expanded context window, while OpenAI countered with GPT-5.3-Codex, a faster coding-focused model that uses fewer computing resources alongside a standalone desktop app.
Continuing on Anthropic, the company has raised $30 billion at a $350 billion valuation from investors as it prepares for a potential IPO this year. The funding will support the company’s infrastructure expansion and help it compete with its rivals.
Moltbook, a platform marketed as a social network exclusively for AI agents, appeared to show autonomous agents organizing and developing shared goals, but researchers discovered that much of its viral content was actually created by humans role-playing as agents. Analysis revealed that about 17,000 human operators were behind the platform’s claimed 1.5 million agents, and exposed security vulnerabilities allowed unauthorized posting, making it impossible to distinguish between genuine AI-generated content and human-authored posts.
Matt Shumer, CEO of OthersideAI, published a viral essay warning that AI could disrupt jobs on a scale “much bigger” than COVID, drawing over 60 million views and mixed reactions from tech leaders and scientists. While some like Reddit founder Alexis Ohanian agreed with Shumer’s assessment, critics including NYU professor Gary Marcus called it “weaponized hype” lacking real data, and others argued that replacing human workers with AI confuses efficiency with purpose and eliminates meaning from work.
“Jackson sets the record straight about Black women’s longtime movement organizing, theorizing, and coalition building in the name of racial, gender, and sexual justice in the United States and abroad. These essays show, in both critical and deeply personal terms, how Black women have been at the center of modern liberation movements despite the erasure and misrecognition of their efforts. Jackson illustrates how Black women have frequently done the work of liberation at great risk to their lives and livelihoods.”
How can we help you?
Case Study: Multi-Market Website Repositioning on HubSpot CMS
An industrial air treatment manufacturer built full systems—ductwork, dampers, fans, scrubbers—for semiconductor fabs and municipal wastewater facilities. But buyers still saw them as a component supplier.
They served three markets with different language and buying processes. The website tried to speak to all of them at once and diluted the message. They needed clear positioning and qualification based on actual application needs.
We rebuilt the site around three entry points, and each section reflected that buyer’s priorities. On key pages, we replaced the generic chatbot with a structured conversational interface that qualified by segment, material specs, and air handling requirements. Those inputs fed HubSpot workflows that handled routing, follow-up, and lead scoring. A semiconductor contractor moved through a different path than a municipal buyer in procurement.
The result: clearer positioning and cleaner handoffs to sales. Leads arrived with context.
If you serve multiple markets that require different conversations, structure isn’t optional. It’s what makes automation work. Hypha builds the systems that turn positioning into execution.
-2.png?upscale=true&width=1120&upscale=true&name=unnamed%20(1)-2.png)
-1.png?upscale=true&width=1120&upscale=true&name=unnamed%20(2)-1.png)
%20overlayed%20on%20an%20image%20of%20a%20woman%20doing%20research%20on%20her%20computer%20and%20a%20man%20reading%20at%20the%20table%20beside%20her.jpg?upscale=true&width=1120&upscale=true&name=The%20Buyer%E2%80%99s%20Journey-%20Understanding%20the%20Three%20Stages%20(B2B%20Guide)%20overlayed%20on%20an%20image%20of%20a%20woman%20doing%20research%20on%20her%20computer%20and%20a%20man%20reading%20at%20the%20table%20beside%20her.jpg)
