The Operating System where reps and agents collaborate to sell. Agents do the repetitive legwork, reps close deals.
Wilmington, DE
Joined March 2026
- Tweets203
- Following234
- Followers432
- Likes486
SoftSync retweeted
the language barrier is very real
said "i'm a potato" in my video meaning i'm a bit of everything, a bit all over the place. in norway a potato literally goes with every meal, it fits anywhere. only to realize in english it means you're clumsy and lazy lmao.
clumsy I can’t argue tho..
SoftSync retweeted
We’re hosting the world's first Autonomous Company Hackathon @ycombinator
$100,000+ in prizes. Winners get a guaranteed YC interview. Mac Minis, iPhone 18 + much more: 💰💰
Pick a business. Build the agent team that runs it with the Naïve SDK. Show it operating at demo time.
November 7–8, YC office, San Francisco.
Thanks to our sponsors:
@OpenAI , @OpenRouter , @supabase, @neondatabase, @vercel , @stripe , @agentmail, @tenki, @MonidHQ, @apify, and more.
register ⬇️
SoftSync retweeted
YC is literally telling where are they going to deploy money
1. Company Brain: turn company rules into skills AI agents can run
2. Software for Agents: build for agents, not humans
3. AI-native services: sell the work, not the tool
4. SaaS Challengers: rebuild legacy software AI-native
5. Inference chips: silicon built for agent workloads
6. Counter-swarm defense: stop drone swarms cheaply
7. Chip supply chain 2.0: modern tooling for semiconductor logistics
8. AI for farming: fewer pesticides, more food
Which one are you building?
ycombinator.com/rfs
The bottlenecks in AI are moving beyond the models.
We’re excited about startups building the hardware, infra, financial systems, and specialized software to bring AI into the real economy, from factories and data centers to government and space.
ycombinator.com/rfs
SoftSync retweeted
100% of companies will need a shared context layer (company brain)
99% of companies don’t know this yet
1% can create edge from this right now
SoftSync retweeted
The best agent feature this year: knowing when to say no.
Permissions and context matter more than raw intelligence once agents sit inside your company.
A contractor who hadn't signed his NDA yet joined a company's Slack and asked the company's AI agent for the full customer list. The agent said no. Most agents running inside companies today would have handed it over, because they run with the access of whoever set them up.
Think about how these things get deployed. Someone connects Drive and Salesforce, drops the agent in a channel, and now a shared assistant has that person's access in every conversation it joins. Security teams know this, which is why so many AI pilots sit in review for months while the enablement team wonders why nothing ships.
Each stage of adoption brings a different worry. With chat, the worry is which models are approved and who's pasting the contract in. Once an agent connects to real systems, it becomes which folder the agent can reach, because every company has one no model should see. Put a shared agent in Slack and the right answer to any question depends on who else is in the room. Let agents run whole workflows and nobody's watching when one hands work to another.
Mintive's launch today handles all four stages with one policy layer. As they describe it, each agent gets its own identity delegated from a person, tool and model calls go through one gateway, and everything gets logged, which they say works out to about 2.5 million calls a day. Modern Treasury put a bank-ops agent live on it in under three weeks and cut a four-hour task to minutes, and the University of Michigan runs it for 150,000 people.
The $11M round led by Cowboy Ventures has an Okta co-founder and the WorkOS founder in as angels, which fits. They built the login layer for employees, and agents will need the same thing, with more of them than there are employees.
SoftSync retweeted
79% of our agents' approval requests never got an answer this month. We only know because we ingest everything into our company brain (calls, Slack, email, Claude sessions) and went looking.
That's pillars 3 and 4 for us. We're moving to optimistic adds with an admin veto wherever possible. The agent does it and logs it, and a person can undo it. That starts with watchlist updates and questions the agent can answer from our own data.
Who's actually measuring this in their own org?
95% of AI pilots never reach the P&L.
The companies that do have 4 pillars:
1. Goals that can't be hit without AI
2. Leaders who lead by example
3. 100% of company knowledge captured so agents can use it
4. Willing to disrupt themselves
Prerequiste: an AGI-pilled CEO who understands that the endgame of the AI-transformation is RSI ⬇️
Imagine Fable-6 or GPT-7 comes out. Would your company get better?
SoftSync retweeted
everyone's arguing 1 agent vs 10 agents.
wrong axis.
10 agents with no shared memory = 10 new hires on day one. all asking "where's the doc?"
1 agent that remembers the company > a swarm that doesn't.
the debate isn't how many agents. it's what they all know.
SoftSync retweeted
Context Conference starts.... NOW!
Businesses have always run on shared context. What things mean. How work gets done. What to do when the answer isn’t obvious.
For thousands of years, that context had one audience: us. Now it has another.
AI doesn’t know what your business means by revenue, which numbers to trust, or which rules to follow. Giving it that context is the challenge in front of us, and no single company can solve it alone.
Next three hours: leaders from pharma, banking, telecom, and SaaS on where they’re investing in context. Inside one of the most advanced AI context platforms. The people behind Apache Ossie and Open Knowledge Format debating whether context can move across platforms. A context contract for this community to ratify.
Plus a few things we’ve built that no one outside Atlan has seen.
Let’s get started. 💫
SoftSync retweeted
Finance people in corporates…
What do you use to tell your AI tools/models how to interpret your financial data? What is the go between context layer that houses the specific to your business context so the model can interpret your data and workflows?
SoftSync retweeted
A contractor who hadn't signed his NDA yet joined a company's Slack and asked the company's AI agent for the full customer list. The agent said no. Most agents running inside companies today would have handed it over, because they run with the access of whoever set them up.
Think about how these things get deployed. Someone connects Drive and Salesforce, drops the agent in a channel, and now a shared assistant has that person's access in every conversation it joins. Security teams know this, which is why so many AI pilots sit in review for months while the enablement team wonders why nothing ships.
Each stage of adoption brings a different worry. With chat, the worry is which models are approved and who's pasting the contract in. Once an agent connects to real systems, it becomes which folder the agent can reach, because every company has one no model should see. Put a shared agent in Slack and the right answer to any question depends on who else is in the room. Let agents run whole workflows and nobody's watching when one hands work to another.
Mintive's launch today handles all four stages with one policy layer. As they describe it, each agent gets its own identity delegated from a person, tool and model calls go through one gateway, and everything gets logged, which they say works out to about 2.5 million calls a day. Modern Treasury put a bank-ops agent live on it in under three weeks and cut a four-hour task to minutes, and the University of Michigan runs it for 150,000 people.
The $11M round led by Cowboy Ventures has an Okta co-founder and the WorkOS founder in as angels, which fits. They built the login layer for employees, and agents will need the same thing, with more of them than there are employees.
🤝 Paid partnership
SoftSync retweeted
we built an ai chat inside our product, and almost nobody uses it.
the superusers don't stay in our chat. they connect their own model over mcp and work in their own system. we get the actions, not the conversation. that's great, but it's only ever going to be a small group.
everyone else opens the chat, sees a blank box, and closes it. no error. just back to doing it the way they always have.
i don't think that's a missing feature. it's the ui, a different way of thinking, and adaptation, stacked on each other.
a blank box with a blinking cursor is a test you have to pass before you've seen any value.
when my mum first tried chatgpt she asked me a question first, to check there was actually someone she could write to. a little curious, a little scared, and doing it the old way is the only way she knows works.
that's most people.
the gap between superusers and everyone else isn't about being technical. it's priorities. the ones who get through decide it's worth a few evenings, and they come out superusers. everyone else has a business to run, and that's a fair trade too.
but that gap gets wider every month.
what would it take for someone who'll never get there to end up with an agent doing the work anyway?
SoftSync retweeted
Why I think building an AI startup with OpenAI or Anthropic or Google or Grok models is kinda like ordering a crisp McDonald's Diet Coke.
🤖 Made with AI
SoftSync retweeted
Congrats to @call_broccoli (W22) on raising $52M across their Series A and Series B!
Broccoli’s AI agents answer calls for plumbing, HVAC, and electrical companies around the clock, booking jobs directly into the scheduling software contractors already use and following up on open estimates.
After their EV startup stalled, @abhishekjain25 and @adityaketkar knocked on 140 contractor doors until a Sacramento plumbing owner gave them five minutes. They built the first product from her office in three weeks, and today she’s Broccoli’s COO.
More than 1,500 home service businesses now run on Broccoli, with over 10 million customer conversations handled and 2.5 million appointments booked.
fortune.com/2026/10/28/brocc…
SoftSync retweeted
Imagine an alien species lands on Earth. They’re incredibly intelligent, especially good with software and computers, and mostly want to help humans get things done.
But they’re also unpredictable. They can misunderstand your instructions, and because they’re so powerful, a small misunderstanding can cause tremendous harm.
How do we integrate them into the economy and unlock their productivity while minimizing the risk?
You build a habitat where they can do great work, with boundaries that contain the damage when things go wrong.
That’s how we at Replit think about AI agents.
Today we’re sharing our secure Agent Habitat architecture: our blueprint for integrating agents into the enterprise.
SoftSync retweeted
last two days: cleaning up a mess we didn't see coming.
workspaces, customer payments, and an SOP with our partner that just wasn't good enough. most of it came down to communication.
old me would've read two days of cleanup as a failure. new me reads it as: good thing it surfaced now, while fixing the process is still cheaper than building around it.
does your team have a real SOP, or do you just wing it? 😂
SoftSync retweeted
🇳🇴 We're proud Norwegians building an American AI company.
But honestly, why couldn't we build it in Norway?
We have the talent, ambition and resources. Yet it often feels like our system makes it harder to build than to play it safe.
Is it taxes, access to capital, or our culture around success?
Entrepreneurs around the world, does your country actually support you in building something big, or does it hold you back?
Genuinely curious. 🌍
