Happy Friday! Thanks for joining me for this edition of The Hypha Wire.
I was talking to a client recently about podcasts and YouTube. He was reflecting to me the sea change he’s noticed in content consumption over the last five or six years: “Everyone wants to watch everything.”
We both admitted that YouTube wasn’t our most familiar territory. Sure, we know how to use it and find videos, but we don’t have a long list of creators we log onto the platform to watch.
I don’t know if I’m a luddite, but I still do listen more than I watch, even with podcasts I know have YouTube channels. I’m open to experiencing the other side, though. Maybe there’s something to be said for adding another sense into my podcasts. (How long until we’re doing 5D podcasts with haptics and smell-o-vision?)
I digress. I’d love to hear from you—what do you like the most about watching your podcasts, or favorite creators on YouTube? And who out there have I been missing since I’m siloed in my PocketCasts app?
Let me know your fave marketing, business, or tech channels! It’s time to broaden my horizons and get into my streamer era.
-Sage Levene, VP of Marketing, Hypha HubSpot Development
Open Mic
AI Finally Has a Killer Use Case in Sales
By Phil Stott, Head of Client Success, Hypha HubSpot Development
Since the dawn of AI (or, more specifically, the first iteration of ChatGPT), we’ve been looking for use cases that justify the hype. And we haven’t been alone there; it’s safe to say that the majority of businesses haven’t found a ton of value, despite an explosion of standalone tools and attempts to cram AI features into every tool in your stack.
However, a killer use case has finally arrived, courtesy of HubSpot: the Prospecting Agent.
If you’ve been following any of our recent emails, you may have already heard about it. So why flag it again? It’s simple: the more our team has been using it, the more convinced we are that it’s going to redefine the sales process, freeing up reps (and, in many cases, company founders) to spend less time sourcing contacts, and a lot more time doing actual selling.
Here’s the basic pitch:
Good sales reps tailor their outreach to the people they want to sell to. Personalized outreach nearly doubles open rates, drives significantly higher click-through rates, and can boost conversion rates by up to 60% compared to generic campaigns.
The problem? Doing it well takes serious time. You need to know who you’re reaching out to, what their specific pain points are, and how to frame your value proposition around their role and priorities.
That’s where Prospecting Agent comes in. Because it lives inside HubSpot, it works off your real engagement data, deal history, and intent signals—think funding rounds, hiring spikes, research behavior—and drafts personalized emails tailored to each prospect. Reps review and approve every send before anything goes out, which is what makes the auto-send feature actually worth trusting.
Where it gets really interesting, though, is in its ability to source actual human contacts for you to reach out to. Integrate it with Apollo (HubSpot’s preferred data partner since earlier this spring) or ZoomInfo, and you have a single workflow that surfaces in-market accounts, identifies the full buying committee, and gets relevant outreach in front of them, without your team ever leaving HubSpot.
We’ve been excited enough that we’ve started building it out for three very different use cases.
Our own portal—because the best proof of concept is always your own pipeline
A skincare brand with a sprawling, fragmented buyer base of med spas, dermatology offices, and licensed professionals—exactly the kind of long-tail TAM where manual prospecting has always been a non-starter
An early-stage B2B SaaS company running sales without BDR support. At roughly $1 per researched prospect (vs $60K+ for a single BDR hire) the math is hard to argue with.
In each of these cases, we’re not replacing a human: we’re leveraging AI to let the people responsible for sales spend more time selling.
Nobody got into sales to build prospect lists or write a dozen variations of an outreach email. Imagine what’s possible now they don’t have to.
The Reddit-ification of the internet continues! Just another reason to develop a Reddit strategy for your brand. But remember, don’t be to spammy or salesy.
“Google is updating its AI Search features to make it easier for users to find information from sources they know and trust. One of the more notable changes introduces ‘a preview of perspectives’ from firsthand sources like social media, Reddit, and other web forums, effectively linking your search queries with online conversations around similar topics.
“Google says this update aims to address that ‘people are increasingly seeking out advice from others’ when searching for information online. This will be relatable for anyone who’s added ‘Reddit’ to the end of Google Search terms to find experiences from real humans instead of SEO-optimized web results. It also backs up claims made by Reddit CEO Steve Huffman last year that ‘just about anybody using Google at this point will end up on Reddit.’”
Ah so maybe the deluge of LinkedIn posts claiming they vibe coded a replacement for their CRM weren’t onto something.
“Security researcher Dor Zvi and his team at the cybersecurity firm he cofounded, RedAccess, analyzed thousands of vibe-coded web applications created using the AI software development tools Lovable, Replit, Base44, and Netlify and found more than 5,000 of them that had virtually no security or authentication of any kind. Many of these web apps allowed anyone who merely finds their web URL to access the apps and their data. Others had only trivial barriers to that access, such as requiring that a visitor sign in with any email address. Around 40 percent of the apps exposed sensitive data, Zvi says, including medical information, financial data, corporate presentations, and strategy documents, as well as detailed logs of customer conversations with chatbots.”
Hypha Highlights
CRM implementations go over budget when scope expands across four predictable project phases: discovery, configuration, data migration, and adoption. Each phase has a distinct failure mechanism—undocumented requirements surfacing late, customization debt compounding silently, data quality issues forcing a binary choice between timeline and cleanliness, and adoption resistance defeating technical success.
Research from the 2025 CRM Failure Report shows only 25% of CRM implementations achieve their planned objectives, timeline, and budget simultaneously. The 75% that miss are not random—they fail along the same four phases, in the same order, for the same structural reasons.
Anthropic is launching a $1.5 billion joint venture with Blackstone, Goldman Sachs, Hellman & Friedman, and several other Wall Street firms, designed to help businesses, including private equity portfolio companies, integrate AI tools across their operations, according to the Wall Street Journal. Anthropic, Blackstone, and Hellman & Friedman are each expected to contribute roughly $300 million, with Goldman Sachs putting in around $150 million.
Additionally, Anthropic has outlined the research agenda for The Anthropic Institute, its internal think tank focused on studying AI’s real-world impacts across four areas: economic diffusion, threats and resilience, AI systems in the wild, and AI-driven R&D. The institute will draw on Anthropic’s position inside a frontier lab to publish findings on topics ranging from labor market disruption and surveillance risks to the possibility of AI systems recursively improving themselves, according to the company.
Two days before his lawsuit against OpenAI went to trial, Elon Musk texted OpenAI president Greg Brockman to explore a settlement, and when Brockman suggested both sides drop their claims, Musk reportedly responded: “By the end of this week, you and Sam will be the most hated men in America,” according to a court filing. The judge overseeing the case declined to admit the text into evidence, though OpenAI’s lawyers argued it demonstrated Musk’s motivation was to attack a competitor rather than pursue a legitimate legal claim.
OpenAI has finalized The Deployment Company, a $10 billion joint venture anchored by TPG and 18 other investors that will embed OpenAI’s tools directly inside private equity portfolio companies, with the company guaranteeing backers a 17.5% annual return over five years. The deal mirrors Anthropic’s $1.5 billion venture referenced above.
In further OpenAI news, the company has updated ChatGPT’s default model to GPT-5.5 Instant, which the company says produces 52.5% fewer hallucinated claims than its predecessor on high-stakes prompts in areas like medicine, law, and finance, while delivering tighter, less verbose responses. The update also brings improved personalization, with the model drawing more effectively on past conversations, uploaded files, and connected Gmail accounts to tailor responses, according to the company.
A new survey of 900 CEOs by Dataiku and The Harris Poll found that 80% of respondents believe their job is at risk if AI fails to deliver this year, with 72% of U.S. CEOs feeling direct pressure from their boards to prove AI-driven ROI. In a notable shift, 65% of CEOs now report feeling more stressed about over-investing in AI than about falling behind on it.
Google, Microsoft, and xAI have agreed to give the U.S. Commerce Department’s Center for AI Standards and Innovation early access to their AI models for pre-release security evaluations, joining existing arrangements with OpenAI and Anthropic. The move comes as concerns over Anthropic’s Mythos model have accelerated the Trump administration’s efforts to establish a formal government review process for AI tools.
DeepSeek is seeking to raise $3 to $4 billion in its first-ever outside funding round at a valuation of up to $50 billion, reversing its longstanding strategy of self-funding through founder Liang Wenfeng’s quant hedge fund. China’s national AI fund is in talks to lead the round, with Tencent also reportedly in discussions to invest, as DeepSeek faces growing competitive pressure from better-funded domestic rivals.
Newsletter Neighbors
The newsletters Team Hypha subscribes to.
I’m not unearthing any indie darlings with this recommendation, but the recently launched CMO newsletter has been a nice resource to read about what’s happenin’ in the greater marketing ecosystem.
A stone fabrication equipment manufacturer needed accurate regional sales tax on multi-SKU quotes before they reached customers. HubSpot’s tax calculation beta wasn’t reliable, and manual state lookups created friction between sales and finance. The CFO was closely involved because quote accuracy directly affected approvals and customer trust.
We built a custom QuickBooks Online integration from scratch. To work around Intuit’s production credential dependency, we deployed the OAuth server and token manager on AWS Lambda before production access was available. The token manager refreshes credentials automatically to keep requests active without manual intervention.
When a rep checks “Bundle Now” on a deal, a HubSpot workflow triggers a webhook to Lambda. The service pulls line items and customer address data, maps HubSpot SKUs to QuickBooks product IDs, calls the Estimates API, and returns the calculated tax amount to HubSpot as a visible reference line item without affecting deal totals.
The result: reps check a box and estimated tax appears automatically. No manual lookup, no finance handoff, no quoting delay.
HubSpot quoting depends on financial data the platform can’t calculate natively? Contact Hypha to scope a custom integration for your setup.
