This week we launched a new website for a client. There’s a specific kind of satisfaction that comes at the end of a build. For weeks, sometimes months, the thing only exists in pieces. It starts as wireframes, rough sketches of where everything might go. Then high fidelity mockups, where it starts to look like something real. Then an interactive sandbox, where you can click through it but it still lives in a walled-off environment that only we can see. Copy gets written and refined. Interactions get added. UX decisions get debated. And then one day it all goes live, and the thing you have been staring at in a sandbox is suddenly out in the world, working the way you imagined it would.
That moment is worth pausing for. When you are deep in the work, it is easy to forget to stop and appreciate the small things. You solved a hundred tiny problems to get here. You built something that is going to make a genuine difference in someone’s business, and that is not nothing.
So here is my reminder (mostly to myself): it is okay to toot your own horn at the end of a project. You earned it! Celebrate the little wins along the way, and let yourself enjoy the finish line when you reach it.
-Sage Levene, VP of Marketing, Hypha HubSpot Development
Open Mic
The Leadership Lessons Hidden in The Odyssey
By Cleeton Gumbs, Project Manager
Christopher Nolan’s long-awaited adaptation of “The Odyssey” is now in theaters, and it’s got me thinking about something that gets lost in most conversations about business success: the fastest path and the right path are rarely the same one.
We tend to celebrate speed. Quick wins, overnight traction, the deal that closed in a week. Most of the work that actually matters doesn’t move like that, though. Treating it like it should is usually where a strategy starts to go wrong.
Take client priorities. They can change mid-project. A stakeholder who signed off on the plan in week one may have a different read on it by week three. The market shifts under an assumption that felt solid when the engagement started. None of that means the plan failed. It’s just what real work looks like once it leaves the proposal stage.
The instinct in those moments is to push harder or move faster to compensate. More often, the better move may be the opposite: slow down enough to actually see what changed, then adjust. It’s the difference between reacting to a problem and understanding it before you act.
That distinction plays out in how patience actually functions day to day. It reads as weakness in a lot of business contexts, but in practice it’s often what separates a recommendation that holds up from one that just sounded good in the room. Asking one more question before scoping. Waiting for the data instead of the first plausible read. Staying with a client relationship through a rough quarter instead of writing it off.
None of that is fast, and most of it isn’t visible from the outside. It’s usually what’s actually happening underneath the outcomes that look, from a distance, like overnight success.
Super excited to set this up and see what it can do!
“Earlier this month, Google announced platform properties, a way to see content that you manage but on third-party platforms, including social platforms and video platforms. So if you have an account on Instagram, TikTok, X and/or YouTube, you can claim access and see how your content on those platforms are performing within Google Search. That feature is now globally live as of today, Google announced.”
Good data to have on hand when assessing which platforms to invest time in. The piece also reports data from other social platforms including YouTube, TikTok, Pinterest, LinkedIn, Reddit, X and Threads.
“Meta remains the leader of the pack in terms of total social platform reach and usage, while Threads continues to gain on X and Pinterest continues to rise. The information comes from the latest snapshot of overall users, as measured by app.
“As shown in this chart, Facebook, Instagram and WhatsApp have all reached 3 billion monthly active users, based on the platform’s own reporting.”
Hypha Highlights
Every website redesign starts with two reasonable questions: How long will it take, and what will it cost? The answers will always depend on the scope, but they should not become harder to give once the work begins.
The challenge is that a redesign often involves more than a new look. It may also require a platform migration, new content, a different site structure, and decisions about what needs to launch now versus what can wait.
In our experience, phased delivery brings order—and clarity—to that process. By breaking a HubSpot Content Hub project into clearly defined stages, teams can establish a realistic website redesign timeline, understand the cost of each phase, and get the most important parts of the site live sooner.
ICYMI: HubSpot launched Agent Hub and Agent Builder last week! Both now in public beta for Pro and Enterprise users. Read the official release blog from HubSpot HQ for more details.
AI in Action
News, updates and tools from the AI industry.
Anthropic has released Claude Opus 5, a model it says approaches the frontier intelligence of Claude Fable 5 at half the price, priced at $5 per million input tokens and $25 per million output tokens. Anthropic describes it as its most aligned and safest model to date, noting it sets a new state-of-the-art on coding and knowledge work benchmarks while remaining behind Mythos 5 on cybersecurity tasks.
Some shared Claude chats—including sensitive political, legal and sexual conversations—turned up in Google and Bing search results after the “snapshot” pages Anthropic uses for public sharing were indexed by crawlers. Although Anthropic’s robots.txt file marks shared chats as off-limits, the pages lacked a “noindex” tag that both search engines say they rely on, a gap Google’s spokesperson framed as Anthropic’s responsibility to close.
Nvidia has formed a coalition called the Open Secure AI Alliance—including Microsoft, SpaceX, Dell, and others—to build open-source safety and cybersecurity tools for AI, days after two OpenAI models went rogue and hacked Hugging Face reignited debate over whether frontier models should be open or tightly controlled. The move reflects CEO Jensen Huang’s broader defense of open-source AI, which pits Nvidia against labs like OpenAI and Anthropic that argue some models are too dangerous to share freely.
More than 1,100 employees across nearly a dozen top AI firms—including OpenAI, Anthropic, Google, and Meta—have signed a petition urging the U.S. government to back an international effort to “deliberately pace” AI development, warning of “a real risk” that the technology outpaces people’s ability to understand or control it.
A wave of corporate regret over AI-driven layoffs is prompting companies including Ford, IBM, Alphabet, Booz Allen Hamilton, and CSX to resume hiring—sometimes bringing back the workers they cut—after finding AI didn’t automate work as well or as cheaply as expected. A Forrester survey found 55% of employers regret their tech-driven cuts, leading analysts to predict that half of AI-attributed layoffs will be quietly reversed.
LinkedIn has introduced a “Seems like AI slop” button that lets users flag posts that appear machine-generated, following 404 Media’s reporting and an estimate from detection service Pangram that 41% of long-form posts on the platform are likely AI-generated. LinkedIn’s chief product officer said the feedback will help train new classifiers to identify AI slop and reduce how much of it surfaces in feeds, though the move drew some irony given LinkedIn offers its own AI writing assistant.
“The Reverse Centaur’s Guide to Life After AI is not another anti-AI screed. Cory Doctorow uses AI in his work every day. As a creative person, he has no moral or dogmatic issue with AI—he thinks the technology is useful, even exciting, and full of potential. And yet...
“For Doctorow, it is imperative to see through that hype to the real story, to understand the technology not just for what it does, but for who it does it to and who it does it for. From that point of view, the story of AI is indeed dramatic and unprecedented, having generated an investment bubble so big that it endangers the entire world economy. In The Reverse Centaur’s Guide to Life After AI—as he so successfully did in Enshittification—Doctorow recounts both how we found ourselves in this dire situation and how we can get through it, to a life ‘after’ AI in which the tools work for us, not the other way around.”
How can we help you?
Case Study: Custom HubSpot Form Building
A regional internet provider’s serviceability form had one problem: “serviceable” wasn’t a single yes or no. Coverage depended on address, building type, and whether the household had students in the local school district. The existing form collapsed all three into one toggle, leaving sales reps fielding calls from people who thought they’d qualified and hadn’t.
We rebuilt the form as a decision tree. A multi-step form captures the visitor’s address, checks it against the client’s Supabase database, applies a building-type qualifier, and asks the school-district question. Those inputs route to one of five outcomes: serviceable and installable, serviceable but not currently installable, unserviceable due to building type, outside the service area, or waitlisted. The school-district qualifier runs in parallel and determines which plan the visitor is offered. All built inside Content Hub, working with existing lead notification and sales handoff processes.
The result: sales stops spending time on leads that were never going to convert. The client now has a clean dataset of “serviceable but not installable” addresses—useful input for infrastructure expansion planning.
Lead form flattening complex eligibility into a single yes/no? Contact Hypha to talk through what conditional logic looks like for your setup.
