@tryhighlight
Account based inUnited States
About this account
- Account based in
- United States
- Connected via
- Web
Account-level information from X, not a live location or the device used for a specific post.
The proactive AI assistant built for high-velocity teams.
Joined May 2024
- Tweets348
- Following29
- Followers4.2K
- Likes607
Was it an execution failure or an alignment failure?
A meeting can end with everyone nodding and still produce three completely different plans.
Researchers have called this the “illusion of alignment.”
In a small study of 18 meetings, their system surfaced an average of 2.89 hidden disagreements per meeting.
Something to remember as AI notetakers become standard: an accurate transcript can still document an ambiguous agreement.
AI meeting transcription and notes aren’t the finish line.
More value comes from turning conversation into explicit decisions, owners, deadlines, and actions in the systems where teams work.
How Highlight tackles the busywork surrounding meetings and keeps teams aligned:
highlightai.com/blog/invisib…
Highlight happy hour, but instead of beer it’s an 820,000 Scoville unit hot sauce.
No employees were harmed in the making of this (though some were definitely changed). Shout out @firstwefeast 🌶️
AI can plausibly get you to 100x productivity, but it’s just as important to talk about what that costs downstream:
100x more review
100x more routing
100x more rebuilding
We learned this ourselves. More suggested actions just meant more decisions to make. We’ve recalibrated to pull back and thoughtfully deploy actions, giving people room for meaningful work without adding to the coordination burden.
We’re going heads down this week, splitting into small pods to focus on the parts of Highlight that haven’t met our bar.
If there’s something you’d want from an AI tool dedicated to taking busywork off your plate, send it our way 👇
We’re listening while we build.
Chat has largely become the go-to AI interface for work.
But valuable AI needs to do more than provide a good response in a chat box. It needs to maintain context as work evolves, understand the state of an ongoing task, and coordinate the actions that follow.
At Highlight, we’re building toward AI that connects conversation to execution. A look under the hood: highlightai.com/blog/agentic…
You know the answer exists somewhere: Slack, a meeting note, a project doc, a weeks-old email thread.
So you spend 20 minutes searching, another 10 checking whether the info is current, then message someone to fill the gaps.
Constantly rebuilding context is exhausting. It leaves less time for the creative, rewarding, uniquely human work that matters.
We wrote about how Highlight is making more room for it: highlightai.com/blog/collect…
Highlight retweeted
My take... Models and harnesses are not nearly enough for mass adoption. There are two major blockers to appeal to the general population.
First is re; models + harnesses = not enough. Models are today essentially just knowledge stores. Becoming increasingly better + faster about accessing that knowledge is not enough. Knowledge =/= wisdom. Models + harnesses are not enough to develop wisdom. To do that, you need context (who/what/where/why/when) + memory (understanding of how the context evolved)
It doesn't matter how many things the models know how to do, or how many things they can do because of harnesses. What matters is what it *chooses* to do and *why*. The only way to do this is to transfer your wisdom; distill your thinking into signals that trigger the right thinking.
The reason why this matters: Trust
This is why to even leverage models/harnesses, we need an encyclopedia of skills + rules etc. We are trying to encode our wisdom into the loop to trust it will produce quality work. The effort that this takes is just completely unreasonable for anyone besides early adopters and builders.
Second blocker to mass adoption is: politics. I don't (only) mean government. I mean social politics. If you even remotely like AI, there are entire groups of people that will demonize and ostracize you because they often literally think AI is the antichrist.
There are unfortunately too many faces of AI companies who very recklessly talk about economic/job impact and security impact of AI, which only creates fear and distrust.
This is why I believe that any product that is going to succeed here is going to be model agnostic. If you are locked into a single model provider as a company, or as a person, you are risking many things between availability and reputation by association.
It will take a neutral party building AI products that earn your trust, and are never restricted by another company's optics or reliability.
Hot take… isn’t it kinda crazy that nobody is really using AI Agents? I don’t mean software engineers or AI early adopters. I mean “college friends talking about it in group chat,” the feeling you got when everyone started using Instagram or TikTok.
These frontier AI models are *insane* (as are the harnesses & tool calls & the like). And every large tech co has an AI agents platform, not to mention all the YC startups doing vertical agents. Yet all of your friends and family outside of tech — who spend all day staring at their iPhones and get paid to work in browser tabs — don’t really care or find themselves using any AI agents yet.
Yes ChatGPT, Claude, etc. are extremely popular… but if you look at the engagement data the vast majority of people are still using these aI chat tools like a glorified Google + Grammarly. That’s why the AGI labs are all pushing desktop apps for Codex, Cowork, etc. so hard to non-technical ppl. And yes exceptions for lawyers and customer service but even those have some asterisks and exceptions to rule.
Look I’m not saying the ChatGPT moment for AI Agents is not coming… it most definitely is! Remember we pivoted from Arc to Dia precisely because we believe computing is going to be radically reimagined around these AI primitives. No doubt. But that’s my point: it’s just so surprising it hasn’t happened yet because all of the tech you’d need is there.
Again if you stop for a second and think about it… for all the press and money and hype and models and crazy ARR numbers… this “AI Agent” moment does not *feel* like the other breakthrough tech moments we’ve lived through (e.g. think the shift to Stories via Snapchat & Instagram, or shift to on-demand via Uber/Airbnb/Doordash).
Which is a long way of saying: if you can figure out the answer to “why” most people don’t care about AI agents yet (and have no enduring interest in using them) — especially since the models and harnesses are here and ready — the answer to that question will allow you to capture a lot of marketshare and make a lot of money in 2027.
Theoretically, the tech is ready for AI Agents to totally transform how we work and live our lives… but alas the general public dgaf… that’s the generational puzzle to solve for the next 12 months for anyone not working on the models themselves.
In the business of making people more capable, never less necessary 👇
Your tools don't know each other.
Slack doesn't know what Linear knows. Linear doesn't know what's in your Notion roadmap.
You become the router, copying between windows and translating one tool's version of the truth into another's.
More on collective context and why the tools you use can't get there alone: highlightai.com/blog/collect…
A preemptive statement from our cofounder Josh (@Galkon):
"If I ever shill a product that operates on the premise of outsourcing your mindfulness, you can assume I've been replaced by a body double like Avril Lavigne."
A spreadsheet should decide what follow-up you owe after a meeting, not who you love. But according to @WSJ, “love hackers” are running dates and scoring partners in trackers.
AI should automate the spreadsheet-shaped work around our lives, not the judgment within them. That’s Highlight’s line in the sand.
We’re seeing more of this sentiment as we build.
The test for AI: does it pull you deeper into the work or turn you into a manager of machines?
Output ≠ progress if it costs immersion, joy, and creativity (the stuff people actually want to spend time on).
AI should absorb context switching. More from Highlight coming 🔜
Highlight retweeted
More of our time is shifting to defining what a great outcome looks like. This is where you work together as a team and iterate. Models do the work and deliver on this outcome.
You don't need graphs or loops to move faster, if you know where you're going.
