Show recap — Friday, June 12, 2026

Another Orange Morning Recap: June 12, 2026

It is Friday, and somehow already halfway through June. The show opened where a lot of us actually are this time of year, half-joking about second-half plans that are supposedly documented and in place, and the marketing leadership baggage that builds up when they are not. The fix, of course, is somewhere in HubSpot. Madelyn Donovan joined Casey and Chris to go find it.

Here is what we got into: a navigation change one of us is still a little nervous about, AI-written next steps showing up inside task emails, meetings finally usable as a workflow trigger, a hard look at what an email engagement score is actually measuring, an astrology data model someone built before seven this morning, and a closing argument about why the words you hand an AI change the answer you get back.

Personalized Navigation and the Five Percent

The headline, tagged as a major update, is that the left navigation is going personal. Each user gets a shortened nav tailored to how they actually work, with the ability to add their own groupings, reorder items, and add or remove what they do not need. It is rolling out across Professional and Enterprise tiers, with Marketing Hub Starter included, and it is still in development, so the usual forward-looking caveat applies. Word is the team behind it may come on the show in a couple of weeks, once it moves into private beta.

This one split the room, and the split is the interesting part. Casey sits in the small slice of users this makes more anxious. When your job is helping clients and supporting portals, a standardized interface is the thing every playbook quietly depends on. Once the nav itself can change, along with where objects live and what they are named, "go to your nav and turn this on, then this" stops being a reliable instruction from one portal to the next. Chris named the counterweight, which is also the official stance: this is better for something like ninety-five percent of HubSpot users. Most people use a handful of tools, not eighty, and the current nav asks them to carry all of it.

Why it matters: The discomfort here is the discomfort of designing for people who do not live in HubSpot all day, which is most people. The one caution worth holding onto is intentionality. The same flexibility that trims a nav down to what you use can rebuild an eighty-item nav if every request to pin one more thing gets honored.

This came up on Thursday's show as well. For the fuller walkthrough of how the new nav restructures CRM objects and tucks the rest under a More section, that recap has it: Another Orange Morning recap, June 11.

Breeze CTAs and Keeping the Human in the Loop

Next up, and live now across all hubs and tiers, is Breeze AI-powered CTAs in assign notification emails. When a task gets assigned to someone, the notification email now carries AI-generated next-step buttons that link straight to the relevant record and open a sidebar with context-specific prompts. The idea is that the assignee opens the email already knowing what they are supposed to do, instead of staring at a task titled "follow up" with no idea which follow-up, for whom, or why.

Madelyn had mixed feelings, and they were the useful kind. Her hesitation came from notifying team members who are not steeped in HubSpot, where a task email can land as noise, why am I getting this, what is this for, even when you took care to write it clearly.

Casey took it deeper, to the human in the loop. As AI picks up more context and more capability, it gets easier for the person to quietly step out of the loop, and once you are out, it is hard to get back in, sometimes after the damage is done. That pressure is worst exactly when bandwidth is tight, which is when you are most likely to trust whatever the AI surfaced first without checking it.

Chris put it in front of the camera. Working in the new universal record view, he created a task and watched HubSpot auto-associate him with a few deals, listed under the Opportunities label, that were not the most recent, the largest, or the most obvious. That raised the real question under this feature: what is the default logic, and do you trust it? When Breeze surfaced the contextual brief, it offered up genuine signal, challenges around workload prioritization and capacity, the kind of thing an assignee would actually want before they act.

Why it matters: This feature is only as good as the data sitting under it. The show's caution to admins and managers was direct: before notifications go out, understand exactly what your users are now receiving. With clean data, this is a quiet power move. With messy data, the first experience is a context-free button, and the fast path from there is notifications turned off and a team that decides Breeze is not for them.

For more details: Breeze AI-powered CTAs in assign notification emails.

Meetings in Workflows Goes to Work

Meeting properties and automations are now available inside workflows, in public beta for Professional and Enterprise tiers across the hubs. It has been around in some form for a bit and looks to be moving toward live, so expect to keep seeing it. The short version: you can now build workflow automation off meeting bookings and meeting properties, using the meeting itself as the trigger.

Casey had a real one from this week. A client is rolling out call-recording software with a GDPR feature that skips recording when a specific tag appears in the meeting title. So she built a workflow: when a contact has already said they do not want to be recorded, that gets added to all of their future meetings automatically. The line that came with it was the best kind of consultant satisfaction, glad you didn't ask me last year. Chris pushed the same thread further, toward building an entire day around the calendar meeting as the triggering record, what happens before, during, and after, with the transcript and what actually occurred feeding straight back into the CRM.

Why it matters: Meetings as a workflow trigger closes a gap that used to mean manual handling or no handling at all. Paired with the wave of AI transcription, the meeting stops being a dead end in your data and becomes a moment your automation can build on.

For more details: Meetings in Workflows public beta.

When the Email Metric Doesn't Tell the Whole Story

Asked what else from the week had caught her eye, Madelyn brought up predicted email engagement in segments, and that opened the most useful teaching beat of the morning. The feature compares a contact's past engagement to the average engagement across all contacts in your portal, which sounds clean until you remember how many different kinds of email a single contact receives.

Chris framed his hesitation well: he is not anti-email, he is anti not knowing what kind of email a number is even about. Madelyn, who works on email for a lot of clients and cares deeply about suppression-list logic, added the wrinkle that some of what you would need to segment on is not visible to you at all, like a contact who globally bounced in another portal but not yours. HubSpot knows. You do not.

Madelyn brought it back to the structural problem the show keeps circling. Marketing engagement, sales engagement, and transactional engagement are not the same thing, and a low score on one can sit next to a person who is replying to one-to-one emails every day. The cautionary tale landed hard: a quarterly attribution report, marketing's credibility riding on it, undone the moment someone clicks into a list and finds a contact the whole company knows is engaged sitting there flagged as cold. The fix is not a better dashboard. It is marketing and sales in the same room, and the reminder that a lot of the time you do not need a lead score at all, you need a list or a segment.

Why it matters: A prediction is only as honest as the question you ask of it. Before anyone gets excited about a new engagement signal, the move is to understand what it measures and say so out loud to the team, rather than turning a quarterly report into a credibility test you did not know you were taking.

An Astrology Data Model, Built Since Seven A.M.

Then the fun. Madelyn has been quietly arguing that astrology and HubSpot have something in common, a pile of objects and the associations between them, and this morning she stopped tinkering with it in Notion and built it in HubSpot instead. Three custom objects, signs, planets, and HubSpot product updates, associated to each other, with each sign loosely mapped to a hub. Gemini, all communication and duality, reads like Marketing Hub and Content Hub. She tied in where the planets actually sit today and a few of her favorite updates from the week, then asked Breeze Assistant for a reading.

This is the part worth sitting with. The demo portal had no knowledge base and no configured assistant. Breeze still pulled the two product updates associated with Gemini, treated them as planetary transits, and produced a themed reading that pointed back to exactly what was linked on the record. Chris asked it to write the reading into a note and add a super compelling visual, and it returned something that looked like an editorial horoscope poster. None of which anyone had to specify in advance.

Why it matters: The data model is where the real power sits. AI stepped up off nothing but keywords and associations, because the structure underneath it was good. That is the argument the segment was quietly making, and it is the reason to make room for this kind of play. You cannot spec your way to results like these, so the only way to find them is to experiment, ideally before seven in the morning if you want to be this ready.

Tribal Knowledge or Human Domain Expertise

The astrology demo set up the closing thought, and it tied back to the worry from the Breeze CTA segment about people removing themselves from the loop. This was the opposite. It used AI to keep the human in the loop, by speaking directly to the human-built structure she gave it.

From there the conversation found its real point, which is about the words you choose. Ask an AI to solve a tribal knowledge problem, and it does what its training tells it to do: extract the knowledge, record it, systematize it, and reduce the dependency on the person. Ask the same AI to make use of human domain expertise, and the answer flips to how you could build something with that person. Same situation, opposite direction, decided entirely by the framing and the namespace those words carry. It is the old HubSpot adoption truth, that everyone reads the same word on the screen differently, scaled up by a thousand. For more on that thread, the Value-First Data episodes from earlier in the week, Tuesday with Klemen and Casey and Wednesday with Trisha and Aaron, sit right alongside it.

The June 11 Updates

All three of Thursday's releases got airtime today, so there is nothing waiting in the corner. Full breakdowns of every June 11 update are on the updates blog: HubSpot Updates: June 11, 2026.

The Sign-Off

You probably already own the value you are looking for in HubSpot. You just need to wake up to it. Have a great weekend, and we will see you next week.