Show recap

Show recap — Wednesday, July 29, 2026

Wednesday, July 29, 2026 · with Casey Hawkins, Chris Carolan, Zach Hussion

From the show

Another Orange Morning Recap: July 29, 2026

Wednesday, July 29. Casey Hawkins was out with her home internet down, and the show opened with Chris Carolan wondering aloud how many days a person can survive without wifi before the body starts shutting down. Zach Hussion took the seat beside him. Six updates got worked through before the hour ran out, and most of them turned into a longer conversation than the release itself.

Here is what we got into: a color theme that finally applies to a whole dashboard at once, the new Actions Library making Customer Agent much harder to keep putting off, HubSpot shipping a branded connection to Epic and what that says about who they are trying to reach, the arrival of human-to-AI handoff in the inbox and the customer experience nobody has designed yet, AI suggesting your lead scoring rules and why the button is the risky part, and a summer of Starter changes that quietly move reporting and automation down a tier.

One Palette for the Whole Dashboard

The show started where the show always starts, with color. Dashboard color themes are in private beta, and they do what the name says: pick one theme, apply it to an entire dashboard, and every eligible report picks up the palette without anyone editing report settings one at a time. There is a brand kit palette option in there too, which Chris pointed out amounts to doing whatever you want.

Zach recognized the pain immediately, and it was not about pretty reports. It was about the hours that disappear making a dashboard look polished, and then someone adds one more report that does not match and the whole thing looks sloppy again. The deeper version of that problem showed up next: you do all the work to get the data right and tell an accurate story, and then one person in the room stops the meeting because red meant bad on the last slide and means something else on this one. Sometimes that pushback is fair, because consistency is the actual point. Sometimes people are just particular about their palettes.

Why it matters: This is a small update that removes a recurring tax on anyone who presents reporting to stakeholders. If nobody in your org has ever argued with you about a color, you can skip it and lose nothing. If they have, the beta is worth getting into, because the fix takes one setting instead of an afternoon.

For more details: read our full take on dashboard color themes.

The Actions Library, and the Question You Will Get Asked

Zach picked this one off Chris's list. The new Actions Library replaces the old single-form setup for Customer Agent with a searchable, filterable catalog of pre-built CRM and Shopify actions, sorted by category. The practical effect is that connecting an agent to an external system, checking an order status, resetting a password, pulling account details, no longer means building and hosting your own API proxy or hand-wiring every field. Those are the questions people actually ask a support team, and they are the ones that used to take three systems and four people to answer.

Chris took it somewhere bigger, and it was the strongest stretch of the morning. Now that Agent Hub has shipped, someone in your organization is going to open it, notice Customer Agent is not turned on, and ask what it is and whether you should be using it. That person may not work in support. They may not be someone you were expecting to hear from. And the answer you have ready matters. Saying yes, boss, we will get that set up sets an expectation you cannot meet. Saying you need a week, plus budget, plus credits, is honest but lands badly. The better answer is a short list of low-cost things you could turn on today, run as a small experiment, so that when you do go ask for budget you are carrying proof instead of a pitch.

He also walked through feeding the update to Breeze Assistant live on the show, which is now a habit rather than a demo. Breeze knows the business, so the ask is short and the answer comes back shaped for that portal, including an offer to pick the best three to five actions to start with. That is the difference between a tool that answers a question and one that knows the context of the question.

Zach closed the segment with the caution. As you give an agent access to more information, look hard at what that information actually contains. A note on a deal where someone said something unkind about a client. A thread where the team talked openly about margin. None of that should sit one integration away from a customer-facing conversation. Test, test, test. Crawl, walk, run.

Why it matters: The updates keep making Customer Agent easier to stand up, which means the conversation about it is coming to you whether you have prepared for it or not. Setting up Customer Agent also puts the wider AI layer in place for the portal, so even a small pilot has value beyond the pilot. Just audit what the agent can see before you connect anything to it.

For more details: read our full take on the new Actions Library.

HubSpot Connects to Epic

Chris flagged this one because he flags anything touching health data. HubSpot is opening a private beta for a native app connection with Epic, one of the most widely used electronic medical records systems. Health records sync from Epic into HubSpot and map automatically to contacts, appointments, and sensitive data properties. Customers working with health information have had to do that by hand, or not at all.

Zach had lived the gap. His firm works mostly with B2B manufacturers, but they had onboarded a chiropractor, a friend who went through school with his brother and who is also his own chiropractor. HubSpot did real work on the marketing side. The patient data never connected. The EHR systems were on premise, several had no API at all, and the data stayed locked down. The client eventually churned off most of the tools, still using some marketing pieces, but the original vision never happened. That is the exact shape of what this closes.

Chris read the strategic signal underneath it. Epic is effectively CRM for hospitals. This is a branded app, meaning HubSpot owns the integration, and that almost never happens without permission on the other side. Getting there means the compliance work, HIPAA and sensitive data handling, is done rather than pending. He connected it to a recent update letting Department of Defense email domains connect to HubSpot, and the pattern is the same: there is a real population of buyers who would use HubSpot but cannot, and as market share grows HubSpot has to keep working these corners.

Why it matters: Read this as market access, not as one more integration. The feature matters to healthcare-adjacent businesses on Epic, who should get on the interest form while spots last. The signal matters to everyone else, because it tells you which doors HubSpot is prying open next.

For more details: read our full take on the Epic app for medical records.

Handing the Conversation Back to the Agent

Zach was reaching for remote MCP connectivity when this one caught the room instead, mostly because of how open-ended the title is. Support agents can now manually assign any conversation to Customer Agent from the Help Desk or Inbox ticket owner dropdown, and AI takes over from there. Customer Agent shows up in the dropdown with an AI badge and a credits tooltip, and is greyed out when no credits remain. If the last message came from the customer, the agent replies right away. If the last message came from a human, it waits for the customer to speak again.

Chris named the reversal, and it is the thing worth sitting with. Every piece of handoff design up to now has run one direction, agent to human. This runs the other way. The mechanics are clear enough. What is not designed is what the customer experiences. A ticket that comes in and gets routed to an agent before a person touches it is an easy handoff. A live chat already in progress is not. The joke that landed the point: the customer notices their rep suddenly got less grumpy and their grammar got perfect, and starts wondering what happened.

The room worked through the failure modes. A silent swap reads as being brushed off. A visible one needs to be worth it to the customer, which means the message has to do real work, something closer to being connected with our AI assistant because of support volume than a blank change of name. And then there is the ping-pong problem: agent starts, human takes over, human gets tired of it and kicks it back, customer can tell and asks for a human again. Zach pointed at the second-order risk, which is that the humans now have a crutch. They are supposed to do the human thing, and they can bounce it instead.

Where both landed is that the label is not the point. It is customer experience either way. Seamless, helpful, efficient, fast, timely. Do those and the customer will not care much who is on the other end. Two things make the difference: tell people before it happens, because transparency is what builds the trust this depends on, and understand that the calculation changes fast. Roughly six months out, the odds are good the agent knows more than the human agent does.

Why it matters: This is private beta and it is worth testing, on your most repetitive and lowest-stakes queue. Write the handoff message before you build the workflow. HubSpot has shipped the mechanism and left the customer experience to you, and that part is not something to figure out live on a real customer.

For more details: read our full take on handing any conversation to Customer Agent.

AI Suggests Your Scoring Rules. Should You Click Add?

Both hosts wanted the scoring updates, and this was the one they gave time to. Discover Rules is live for Sales Hub and Marketing Hub Professional and Enterprise. It reads your portal's historical conversion data and suggests the scoring rules most likely to drive conversion, showing which actions actually moved contacts and companies to a goal stage. The pitch is a more accurate score built in minutes instead of from scratch.

Zach named the problem it is aiming at, with Casey's absence noted since this is her territory. Lead scoring is subjective in a way people do not admit. Who decided an email open is worth two points instead of five, or ten. Teams spend real time arguing about relative importance, then arguing about the scale, then about decay and whether an action counts once or every time. If this gets a team past that faster, that is worth something.

Chris's caution was sharper, and it was aimed at the interface rather than the AI. There is an add to score button. He tells every product manager he talks to the same thing: when AI is involved, do not make it that plug and play. One click and you now own a rule, and owning it means maintaining it, and nobody ever asked whether it should exist. The same suggestion routed through a conversation with Breeze Assistant goes somewhere better, because you can ask what would need to be true for this rule to hold up, and find out the property behind it is ten percent populated, which makes the signal mostly noise. Skip that and you have added tech debt you will have to dig back out of.

There is a second gap worth knowing about. The tool understands supported types. Anything you do differently in your portal, a custom event, a property used in a nonstandard way, sits outside what it can read, and it cannot tell you about a context gap it cannot see. Both hosts wanted the same fix, which is for HubSpot to put the conversation in front of the button rather than beside it, and to do it across far more of the product than scoring. Let Breeze ask the questions you did not know to ask, then set the thing up.

Why it matters: Use it if you are already scoring and want the rules grounded in what has actually converted rather than in a meeting. Do it in a conversation with Breeze rather than by clicking add. If your data is thin, fix the data first, because otherwise the AI will formalize your noise and hand it back to you looking official.

For more details: read our full take on AI-powered scoring insights.

Starter Gets Reporting and Automation

The last stretch went to HubSpot's summary of what changed in Starter this summer, which Chris always reads because how a vendor summarizes itself is its own kind of signal. The bundle is broad: per-user customizable navigation that is live now, a rebuilt HubSpot Home that surfaces the day's tasks, events, and meetings and offers a Breeze-drafted reply on an email that needs one, redesigned table and board views with lane-level controls, simplified two-column records that open in place, automations you can create straight from an index page, and full Analytics Suites for Marketing, Sales, and Service. The Home piece goes live July 30. Most of it applies automatically to enrolled Starter portals.

Clicking into it turned up a second signal. The navigation on the update page did not match the navigation Chris has in his own portal, because personalized navigation has been rolling out to client portals and the sections are different. New experiences often start on free and Starter, where exposure and risk are lower, and move up to Professional and Enterprise once they hold.

Zach brought the part that made it real. He sits on the board of the American Marketing Association chapter in Rochester. It is a nonprofit without much budget, so it runs on Starter, and his hands have been tied on anything automation related. That is the honest shape of the tier: a good interface over a contact and company database, cheap for what it is, and short of what a business actually needs the moment the work stops being manual. Reporting and automation were the two walls, and they were load-bearing walls, the reasons you eventually upgraded. Full analytics suites at Starter is a real change. Creating automations from an index page is the other one.

Chris added the counterweight. If you have watched any of the AI content, at some point you work out that you can ask Breeze to build the report and skip the report builder entirely. Which makes the analytics suite less of a prize than it looks, as long as your data is right. Get the data right and keep asking Breeze.

Why it matters: If you run a Starter portal, log in this week and look, because most of this arrived without you doing anything. Reporting was the capability HubSpot held tightest to drive upgrades, and it is now sitting in the tier below. Just do not let new automation run on top of data you have not cleaned up, because the automation will only make the mess arrive faster.

For more details: read our full take on what is new in Starter this summer.

The Sign-Off

A full board of releases and six of them worth stopping on, which Chris marked on the way out as the journey continuing toward Unbound 2026. The thread running under all of it was pace. There have been well over a hundred updates this month. Breeze went from something people wrote off to something worth reopening, and a lot of that shift happened in the last couple of months, which is why nobody should assume they know the current state of anything in HubSpot anymore. If you tried a tool three or four months ago and it was not good, that was true then and it is not true now. Try it again.

Thanks to Zach for covering the morning, and we hope Casey's wifi comes back before the body starts shutting down.

You probably already own the value you are looking for in HubSpot. You just need to wake up to it. We will see you tomorrow morning.

Produced by Value-First Media