Hello hello,
I hope you’ve had a solid week and are making great moves.
Just a quick one from me on Arthur — it’s been a busy stretch tightening and lighting things up.
Before we get into the 😇 Good, 😈 Bad and 🤢 Ugly, a quick Arthur update.
ARTHUR – BUILD IN LIVE #4. Your startup mentor, co-founder and team.
This week wasn’t really about Arthur. It was about the founder on the other side of the screen.
Last week we wired in the basics — identity, payments, uploads and the early microservices. This week was the moment that plumbing turned into engines that move a founder forward: your 2-minute pitch on video, your deck, and reports you can edit, share and iterate with your team.
The shift was simple: stop building “cool AI demos” and start building your improvement loop.
Everything now follows one line:
Intake → Diagnosis → Founder Fix-First Value → Execution
Centred on a single question:
“How do we make this founder more market and investment ready, fast?”
🎯 The insight
Danger Room isn’t a feature — it’s your intake lane.
Once we treated the 2-minute pitch as structured signal, not “a video”, it unlocked the four things founders actually need:
clear, reusable scoring across every founder
investor-relevant signals, not emotional guesswork
a ruthless, practical fix-first list
a clean hand-off into Arthur to do the work with you
And now every run becomes part of your history — your pitches, scores and fixes over time — so you can literally see yourself getting sharper, not just collect another verdict.
It’s the first time a founder can measure how they show up, not just what they say — and then improve with an AI co-founder sitting next to them.
That’s the moment Arthur stopped being a demo, and started being a reusable engine for becoming investable.
Next week we have Pitch Deck and Website Intel being added, and these give a deep understanding of the business, turning hours of analysis into a 10 minute deepdive with actionable execution plans that change the game for founders.
⚙️ Why this matters
Arthur isn’t the story — you are.
Arthur is the invisible infrastructure that turns raw effort into investor-ready signal, on repeat.
ARTHUR - BUILD IN LIVE #4
Every week, I’m opening up the build of Arthur — what we’ve shipped, what we’ve fixed, and what we’ve learnt.
👉 Early Access is now open, if you want to be part of Arthur’s early access group, the founders who get the first features, the first tools, and the ability to shape what we build, sign up here and earn more Fuel units
* F42+ members will be automatically added with your subscription automatically translated into Arthur Fuel_units for the next 3 months.
The 😇 Good, 😈 Bad and 🤢 Ugly…
😇 The Good
⭐️ Obviously Arthur and here comes the 🚨 Danger Room
In two minutes, Arthur now pulls out the signals a real investor actually watches for: presence and framing, narrative beats, delivery patterns and, most importantly, whether your story would earn you a meeting. Under the hood it’s running Rekognition + Transcribe through our agent stack so it can read how you show up, not just what you say.
You get an overall grade, scores for content / delivery / presence, a ruthless fix-first list and the top investor questions your pitch will trigger — plus a history of every run so you can see yourself getting sharper, not just getting another verdict.
Do I always look this serious and grumpy?
I had a great call with our friends over at Google this week, and we’re lining up a deep dive on how to use their AI + data stack to actually compound growth, not just burn free credits. We’ll cover the practical bits founders care about – infra choices, AI services, data pipelines, with a light touch on Ads so you can see the whole stack, not just one product. Once dates are locked, Fusion founders will get first access, and we’ll turn the session into concrete playbooks you can plug straight into your own build.
Watch out for your invite to this session.
⭐️ AI for Startups this Tuesday
Missed it way too much, and there is just so much going on in the AI space not to keep informed.
😈 The Bad (and how we deal with it)
A few thoughts I’ve had as a builder, operator and investor in this new AI era.
1. Compound systems beat big swings
The mistake I see everywhere right now is founders chasing the next model release like it’s a strategy. There’s no strategy in “what’s the newest thing OpenAI or a Chinese Open source model can do this week?”
The real compounding comes from building a repeatable loop:
intake → diagnosis → fix → execution → data → better intake.
Do that 30 times and you’re unrecognisable.
Arthur is being built around that loop — not demos, not hype.
2. Founder-CEOs have never mattered more
A startup today isn’t product + distribution + capital; it’s context + conviction + community.
That mix doesn’t live in a playbook or a hired CEO.
It lives in the founder who can:
see the customer clearly
hold a non-consensus view long enough to win
design a system around a real pain
Most of the value in the earliest phase is belief paired with evidence.
You can’t outsource that.
3. AI is leverage, the most powerful we’ve seen
The quiet revolution isn’t “AI replacing jobs” right now.
It’s AI deleting the administrative drag that kills momentum:
reconciling data
documenting decisions
moving work between teams
rewriting the same narrative 20 times
The teams that win use agents to buy back their own attention and redeploy it into sales, product and story, not to run leaner and louder.
4. The builder has changed and that’s a massive deal
The next wave of products won’t be built by a single technical founder in a dark room.
They’ll be assembled by operators who know the customer, stitching:
workflows
agents
micro-apps
into something that solves one painful, expensive problem.
This is the first generation where you don’t need a codebase to have a product — you need taste, empathy and clarity.
Arthur exists for these people — the ones who aren’t “AI engineers”, but design better answers.
5. Narrow beats loud
It’s never been easier to launch something that looks impressive.
The irony is that distribution now rewards the opposite of noise.
The winners will choose a very narrow slice of reality —
“fix this thing, for these people, in this moment” —
and become the default agent for that job.
If Danger Room taught me anything, it’s that a two-minute pitch can be a better intake tool than any “unified platform”. It is so powerful even as just a video in our telegram channel.
Solve one sharp pain → extend from there, don’t build feature build products.
6. The interface is becoming conversation-first (voice will follow)
We’re not a “voice product” today, but it’s obvious where things are heading. The real shift isn’t text vs audio, it’s forms and clicks vs conversation and context.
Founders won’t want to learn yet another tool – they’ll want to talk through a fundraise, a deck, a hiring plan and have the system keep up. As voice, video and agents converge, creators won’t just own content, they’ll own small pieces of software and tuned workflows that sit directly in that conversation with the user.
The leverage flows to whoever owns that relationship; AI quietly does the rest in the background. Arthur is being built for that world, even if today most of it still looks like “chat on a screen”.
If you have not checked out the new fusion website, take a visit and check it out, you can literally ask it anything, about the arthur build, latest content, event calendar, even what size shoe I wear.
🤢 The Ugly
Investors don’t know how to read agentic products
The strangest thing I’m seeing right now is this: founders are building real agentic systems… and then having to pitch them like it’s a 2020 SaaS product.
The positioning gets flattened into feature lists, screenshots and a token “copilot” slide. On the surface it looks like another platform with an AI add-on. Under the hood, it’s a living system.
The real value in agentic work sits in places most investors can’t pattern-match to (yet). They still want to “see the platform” — a big shiny UI — when in reality that layer doesn’t need to be flashy anymore, so they quietly downgrade it and don’t invest.
The real questions to ask are:
How good is the loop?
Intake → signal → decision → action → data → better next run.How much drudge work actually disappears?
Not demo theatre, but hours and hand-offs removed from the workflow.How fast does the system learn?
Does each run make the agent sharper, or is it just re-prompting the same model?What does “good” look like in this agent’s world?
Clear before/after states, not vanity dashboards.
Because none of that is standard in decks or conversations yet, a lot of serious work looks “unclear”, while shallow wrappers look “fundable” simply because they fit the old pattern.
And yet you’re still seeing eight-figure cheques go into thin products with a bit of AI on top — or essentially an old SaaS product stitched together with a few automations calling LLM APIs. It feels like the electric scooter wave all over again: plenty of capital, not much systems thinking.
If you have not joined the Fusion42 Community on Telegram —
it is probably time to do so.
For the ❤️ of Startups
Enjoy and hope to see you soon.
❤️ & ✌🏼
Thank you for reading. If you liked it, share it with your friends, colleagues and everyone interested in the startup Investor ecosystem.
If you've got suggestions, an article, research, your tech stack, or a job listing you want featured, just let me know! I'm keen to include it in the upcoming edition.
Please let me know what you think of it, love a feedback loop 🙏🏼
🛑 Get a different job.
Subscribe below and follow me on LinkedIn or Twitter to never miss an update.