Edition 3 · 29 June 2026
just tag @claude
Claude released its tag feature, so you can @Claude in Slack and it will proactively do work for you. Plus how we're redesigning our sales process.
Sorry I’m a few days late, been sick and went down to Albury Wodonga for basketball (we won).
What up y’all,
For the TLDR for today. Going to be running through:
how Claude lets you tag it in slack now + more AI news
how we’re redesigning our sales process
thoughts on building a horizontal AI startup, when it feels like the incumbents could ship your whole product as a feature.
Enjoy :)
💼 1. What’s new in the biz
We’re redesigning our sales process.
There is so much alpha to be gained from having simplistic visuals which portray how AI can actually be implemented for businesses.
As I touched on a couple of weeks ago, AI brains and Jarvis setups are all the rage. But they rarely actually move the needle for people. But the demand for visual illustrations of what AI can do is tantalisingly high.
The alpha is attributed to you when you can convey a complex idea, simply.
I’m reading the Steve Jobs bio at the moment and being able to see this phenomenon via a different lense, hammers home this message. His whole ethos for design was simplicity. Making things abundantly simple and clear - because people do judge a book by its cover.
As such, we’re not only redeveloping our onboarding process, but also ensuring the visual aesthetics of it are, simply delightful.
In doing so, we aim to be granular in the use cases of AI. Despite the overwhelming similarity of AI applications between commercial property and cleaning, for example, commercial cleaning prospects struggle to comprehend how AI can help them despite showing the exact same use case for commercial property.
So ensuring this is conveyed as simply as possible is absolutely vital for ensuring the best possible experience for the customer.
See Steve talking about simplicity in design idea here
🤖 2. What’s new in AI (and what I’m actually using)
You can tag @Claude in your slack.
Anthropic (parent company of Claude) released a cool feature. The ability to tag @Claude in your slack.
You tag it like you'd tag a teammate.
@Claude, summarise this thread.
@Claude, pull last week's numbers.
@Claude, open a PR for that fix.
It reads the channel, does the work, and replies in the thread.
I’ve been using it thus far and it keeps hitting rate limits - assuming its just an issue with its new release and a lot of people wanting to use it.
But the premise is super cool. What’s awesome is that it works like a coworker. So it shares memory across teams, departments and can be dropped in to do adhoc work, but then also be picked up by someone else whenever you’re ready. And it will go out of its way to do proactive work - e.g. if it thinks your lagging in sales or there is an that you have that is not good at the moment, it will flag and say, “CPL is up by 17%, want me to turn this ad off?”
It is only available on the team/enterprise plan, but for those stuck in Slack, it’s definitely worth trying out + its just a part of your subscription, no additional API costs.
SpaceX IPO'd, then bought Cursor for $60B in stock.
SpaceX went public on the Nasdaq on Jun 12 at $135 a share, raising about $75B, the largest IPO in history, and minting Musk as the first trillionaire. Four days later it agreed to buy Anysphere (Cursor) in an all-stock deal valued at a fixed $60B, the biggest acquisition of a venture-backed startup ever. The mechanics are the fun part. SpaceX stock whipsawed after the float, briefly past $225 (a near-$3T market cap) before sliding back under its $150 debut to about $153. Because the $60B is a fixed dollar value, and the share count is set by SpaceX's average price in the days before the deal closes in Q3, a falling SpaceX price actually works in Cursor's favour: the cheaper the stock, the more shares SpaceX has to hand over to make up the $60B, and the bigger Cursor's slice of the company if it later recovers. The dilution lands on SpaceX's existing holders, not on Cursor. Closes Q3 of this year.
Salesforce buys Fin for $3.6B
Largest agentic CX deal to date; Fin is an agentic AI customer service platform with 30k enterprise customers. The deal closes early 2027. I remember seeing Fin a while ago and thinking huh, neat product I wonder how it’ll go.
Lo and behold it’s sold for a few billion, markets really are massive.
Also a great move on behalf of Salesforce. CRMs and companies that store the data should have an inbuilt advantage in the AI space, as they should be able to combine the data they hold with the best AI tools and provide the best move. So bullish on companies like Salesforce, Hubspot, etc, despite the supposed SaaSpocalypse.
ChatGPT fell below 50% market share for the first time. It slipped to roughly 46% of the global assistant market, with Gemini climbing to ~27.7% and Claude to ~10.3% (Claude was 3% in December, note Claude has usage where it matters, in the enterprise). The one-model-rules-all era is ending, and the field is fragmenting fast.
🧐 3. Thoughts on building a horizontal AI startup where so much value accrues to the incumbents
So I mentioned the Claude tag feature earlier.
There is another startup Viktor, that both my co-founder and I have been monitoring, which had already built out Claude’s tag feature - it’s called Viktor and is positioned as the first “AI coworker that lives in your slack”.
They’d already reached a run rate of $10-15mil ARR in the last 10 weeks. But Claude shipped their whole product as a feature. Their CEO talks about their plans forward as a company here:
View @frydwia’s post on X
I’m still bullish on Viktor as a company but it made me think about how startups that are at the horizontal play should think about their growth.
The way I see horizontal AI-startups is that there are risks, plays and imperatives:
The imperatives:
Speed is of the utmost priority.
You need a wedge - something that makes your product uniquely identifiable with a pain point.
Distribution is key. If you can’t get your product out to enough people and get enough adoption early, doesn’t matter how good your product is.
The risks:
There is a general cap to their size/duration of the company, particularly when targeting teams and enterprises (since the AI tool usage isn’t limited to the individual and thus can’t be limited at the individual level) and if your product is super general, e.g. doesn’t solve one core problem really well.
You are literally going against the incumbents, so you could just be eaten.
The pros:
You have the positive of being able focus on the one core problem your users have.
You can be a model dispenser (e.g. switch between models) and provide the best one.
You can become an attractive acquisition target for the incumbents (if you get to scale quick enough).
For myself, I think horizontal AI plays are a lot more attractive (albeit scarier). Because if you can be niche enough (e.g. AI generated marketing) you’re differentiated from the incumbents, your tam is bigger and there is scope to become a wider platform play. The same can be said of vertical AI startups (in regards to becoming the wider platform play), but you’re often just limited to the industry you start in.
One of the more important things I’m finding for the horizontal play, is picking a wedge, or almost positioning your product as a tool in a vertical niche. E.g. Viktor was the “AI coworker for Slack.” That made it really easy for people, technical and non-technical alike, to understand the tool from day dot.
The plan from there would be to grow into the teams application (which they just launched) but it was that initial wedge which provided really good traction.
The scariest part about being a company like Viktor is that you constantly are watching over your back to see if the incumbents are going to ship your product as a feature.
I think there are examples where that has happened but it hasn’t been successful or that detrimental to the startup, e.g. lovable, higgsfield, etc. Claude design shipped and yet Lovable is at $300mil ARR. There are new video models every week but Higgsfield is the aggregator of these models.
Ultimately, I think the biggest part about this is having core messaging and moving fast to gain market share/distribution, so that when the incumbents do ship your product as a feature, it’s doesn’t matter all that much.
🎶 4. Carve outs
📚 Books I’m reading at the moment:
Jobs by Walter Isaacson (wasn’t aware how mercurial Jobs was, interesting guy). Almost at the end, love the arcs of his career, despite his odd nature.
Brother is reading Lucky Country which is quite good he says.
🎵 Songs:
Playlists from TV shows (the Succession one goes hard)
⚽ Sports:
Go the socceroos (this world cup has also sent me down a tailspin of Ronaldo highlights)
Until next time chaps,
Cheerio,
Alex