Thank you for bearing with our absence last week, as myself and a few other team members attended HubSpot’s inaugural UNBOUND conference. We had an amazing time learning, catching up with HubSpotters, clients, and partner colleagues and getting that team-level face to face time that remote work doesn’t always afford.
Now that we’ve had a week to settle back in and sort through everything we took away, it’s clear which features made the biggest impact and which ones have us most excited.
The biggest one is AI context. The value you get out of HubSpot’s AI tools depends almost entirely on what the CRM already knows about your business. You can’t generate good output from thin context. The tools that impressed us most were the ones that leaned on solid ICP definitions, buyer personas, and clean records underneath. That foundation is doing the hard work, and it’s the piece teams often skip.
Breeze Assistant also earned a real spot in daily operations. It’s genuinely good at answering the small, specific questions that used to mean opening a ticket or digging through the knowledge base. The kind of thing that would take 40 minutes is now sitting right there in the platform. The mindset shift is treating it as a first stop rather than a last resort.
CPQ and quoting got a meaningful upgrade, which matters for anyone whose sales process has outgrown manual proposals. We’re super excited to dive deep into how we can customize sales processes with the tools.
What ties everything together is a shift in what HubSpot is asking of us. The platform keeps getting more capable, but the payoff scales with the groundwork you put in. The teams that get the most out of these tools are the ones treating their CRM as something to invest in, not just something to fill.
If you weren’t able to attend and you’re curious what you missed, we’d love to hop on a call and discuss new features, focus areas, and developments that came out of the conference. Just reply to this email and we’ll set up a time!
-Sage Levene, VP of Marketing, Hypha HubSpot Development
Open Mic
Amazon Web Services (AWS) for Integrations
By Zachary Lyons, Integrations Developer
When working with clients to set up integrations to connect the applications they use every day, we often have to prioritize speed of implementation. There are many SaaS tools that offer out of the box connections to our apps, but costs can scale quickly with per-seat, per-event, and per-task pricing. Many integrations perform a simple task of triggering the code > transforming the data somehow > and sending that data to the next application. Building on AWS allows us to implement these data syncs ourselves without paying the middle-man (who likely is using AWS to power their SaaS for a fraction of what you pay them!).
Level 1: Scheduled Lambda
The simplest integration pattern is one we’ve talked about before in the open mic: Polling with Lambda. Using AWS Lambda (Lambda is just a place in the cloud where a bit of code can run) we can set up code to periodically reach out to your application (like an RSS feed, a CRM, or even your own internal systems). You set up the Lambda to run every hour, day, or interval of your choosing, by connecting AWS EventBridge (easy to connect via the UI). This is extremely cheap (I often see customers getting bills for pennies a month after we’ve set up their AWS account) and easy to implement with a developer.
Level 2: Lambda & API
Taking our hypothetical integration a step further we can switch from a basic polling system to an event-driven setup with AWS API Gateway. Connecting an API Gateway to your Lambda allows you to trigger your code when your apps send a webhook. HubSpot for example can send webhooks and trigger your off-platform code via Workflows! This allows you to see near-instant data synchronization between your applications, which is essential for some Sales teams. This setup is slightly more nuanced than the previous polling implementation, but is still easy for your developers to set up quickly.
Level 3: Step functions
Now we arrive at our “level 3” version of a simple AWS integration: orchestrated workflows. Using AWS Step functions, we can orchestrate many Lambdas to work together. Just like workflows in HubSpot, we can add branches for different conditions, decide how to handle errors (retry the workflow or just send a notification email to your dev?), and add delays if necessary. Now we’re building complex behavior for your integration that equals what we can accomplish with Zapier, Make and n8n for a fraction of the cost!
These solutions require a developer but make sense in high-volume integrations where SaaS pricing explodes. Consider taking a little time to save your team money in the long term, freeing up resources that you can invest back into the team. If you don’t have developer resources available, reach out to us at Hypha HubSpot Development and let’s talk about supercharging your team!
How many times can this possibly happen before it finally stops?
“It’s easy to assume that part of the problem lay in 2024’s mass corporate retreat from DEI initiatives, and Bakare did fault the ‘gutting’ of such programs. But all the diversity in the world will not help, she said, if employees with cultural sensitivity are not empowered to make decisions or feel free to speak up.
“‘Somewhere inside most of these companies is a person who would have flagged this in 30 seconds,’ Bakare said.
“‘The question is whether that person was anywhere near the approval chain, and whether saying no felt safe. I’ve watched marketers raise these concerns internally and come up against a lot of friction. When the people who can see the harm have no say in the decision, the outcome is predictable. Diversity only becomes a safeguard when it comes with decision rights.’”
AI mixed with pay data? What could possibly go wrong!
“More firms are using AI tools to comb job boards for competitors’ salary ranges in an effort to pinpoint market rates. It’s an exercise made easier by the expansion of pay-transparency laws. Third-party companies now offer AI salary benchmarking as a service, helping employers understand what it costs to hire and retain people in specific roles.
“This could be good news if an AI model indicates you’re underpaid. But if your pay is already on the high side, AI might dash your hope for a hefty raise even if you’re an all-star.”
Hypha Highlights
Sales wants better leads. Marketing wants clearer feedback after the handoff. Operations has workflows that need attention. Meanwhile, leadership wants to know why deals take so long to close.
Deciding what to fix to prioritize first means understanding where those problems originate—and what each one costs the team.
At Hypha, we approach sales process optimization with a simple question: What is preventing the next meaningful action, and what would it take to resolve it? The answer might point to missing qualification data, unclear ownership, or a problem with your HubSpot configuration.
Trace that obstacle through actual records, then compare potential fixes by business consequence, frequency, effort, and dependencies. That gives you a basis for choosing the first project and checking whether it helped.
There’s a new HubSpot beta we’ve been exploring called HubSpot Work, and I think this one is worth paying attention to.
The easiest way to think about it is project management inside HubSpot. Instead of having all of your customer information in HubSpot, but then jumping over to Asana, spreadsheets, docs, or another tool to actually manage the work, HubSpot Work brings those pieces together.
You can tell Work what you’re trying to accomplish in plain language, and it can help build out the project, tasks, tables, documents, forms, and automations for you.
One of the first use cases that came to mind for me was client onboarding. That’s where I think HubSpot Work could get really interesting.
Imagine a deal closes and your onboarding team already has access to the customer history, deal information, notes, and everything else living in HubSpot. From there, Work can help organize the onboarding process, assign next steps, and automate some of the busywork without your team having to bounce between multiple systems.
The HubSpot Work beta is available in your portal now, so if you’re curious about it, want to see what it could look like for your team, or want help testing out a use case, reach out to us. We’re happy to dig into it with you!
AI in Action
News, updates and tools from the AI industry.
The AI labs kept up their rapid release pace this week. Anthropic launched Opus 5.5, which it says beats the larger Fable model on many benchmarks while cutting prices. OpenAI released GPT-6 Sol and Luna, claiming roughly half the error rate of their predecessors and emphasizing 50% lower API pricing. SpaceXAI’s Grok 4.7 targets price-performance in coding with a new safeguard stack, while Google expanded its Gemini Audio family with two expressive text-to-speech models spanning 100+ languages. The common thread: cheaper, faster models and purported growing attention to safety.
Independent security researchers at Hacktron AI used a cybersecurity-focused version of Claude to exploit a bug in OpenAI’s community forum software, gaining access to an OpenAI employee’s ChatGPT account and read access to “Monorepo,” a repository of the company’s algorithmic secrets (though not its model weights). The team, which reported the flaws through OpenAI’s bug bounty program for a $6,500 payout, said the ease of the intrusion—“just three guys with Claude and Codex subscriptions”—shows how AI tools are lowering the bar for sophisticated hacking.
The State Department has ordered diplomats in its international organizations bureau to replace “artificial intelligence” with “super intelligence” (SI) in all communications, following Trump’s announcement at the U.N. General Assembly that he would “officially” rename the technology because the word “artificial” makes it sound “fake.” The attempted rebrand comes as Trump continues to reject calls to slow AI development through government oversight, citing competition with China, even as tech CEOs and others warn about the technology outpacing humans’ ability to control it.
At Meta Connect 2026, Meta positioned its personal AI agent Muse—recently the No. 1 app on Apple’s App Store—as the centerpiece of an AI strategy that differs from rivals, announcing new features including voice mode, interactive avatars, partner connectors, Mac computer use, a dedicated email address, hands-free access on AI glasses, and an AI pendant. The Deep View notes Meta’s ad platform is fueling Muse’s record downloads and that its advertising war chest lets it play a long game on consumer AI where OpenAI and Anthropic have pivoted toward enterprise, though its biggest challenge will be convincing users to trust it with private data given its track record.
“Kipp Bodnar and Kieran Flanagan built the marketing framework that millions of businesses adopted. Then they watched AI dismantle the only marketing model most businesses had ever known. So they threw out their own playbook and built a new one.
“They tested it on HubSpot’s marketing engine first. Within a year, it was outperforming the old one they’d ripped up.
“In Loop, they introduce Loop Marketing: a four-stage system (Express, Tailor, Amplify, Evolve) where the teams that win are the ones that learn and adapt fastest, not spend the most.”
How can we help you?
Case Study: Act! to HubSpot Migration
A commercial building products manufacturer was ready to leave Act!, but two things complicated the migration. Multiple locations shared the same corporate domain—HubSpot’s default matching would have collapsed distinct branches into single records. And the sales team was still actively quoting in Act! throughout the implementation.
Our data architect built parent/child company structures so each branch kept its own contacts and deals. For timing, the main dataset moved into HubSpot first for configuration and training, then a delta import at cutover captured everything created or changed since the initial load.
Deal pipelines followed the estimating and quoting process. Outlook logged email against relevant records. Gravity Forms connected web inquiries directly as contact records. Trade shows got dedicated lead capture using mobile scanning—contacts enter the system associated with the right event. Training was role-based throughout.
The team is out of Act!, users report the new system is faster and more reliable, and marketing has a repeatable process for trade-show leads instead of disconnected spreadsheets.
Customer data has quirks a standard import won’t handle, or your team can’t pause sales during the switch? Contact Hypha to talk through a clean migration.
![Screenshot of a "Content" table grouping example items by stage — Idea, Drafting, In Review and Approved — with columns for Title, Stage, Format, Owner, Reviewer and Target Publish Date. A side panel shows the draft "[Example] Launch checklist for AI-assisted content" with sections for Working title, Audience, Key points, Outline and Draft body.](https://www.hyphadev.io/hs-fs/hubfs/unnamed%20(2)%20copy%20(1).png?upscale=true&width=1120&upscale=true&name=unnamed%20(2)%20copy%20(1).png)
