Mickayla Williams

tech pr @VSCPR | CU Boulder alum #skobuffs

[Rune @Rune_Energy]

We've raised $40M to turn stranded solar power into AI compute, no grid connection required.

Meet RELIC, the drama-free data center, deployed in under 60 minutes and already live at a 200MW site in Texas.

More from @adele_peters at @FastCompany.

[vijaychattha @vijaychattha]

No summer slowdown for VSC companies. 🫡 💰 🚀

Clasp — Student debt is pushing healthcare students to Kalshi. (CNBC)

Reframe Systems — Bringing Amazon-style robotics to homebuilding. (The Information)

Deep Cogito — Former Google AI Search product team is giving companies more control over their own AI. (WSJ)

Inception — Making the case that the next AI breakthrough won't come from transformers. (MIT Technology Review)

Mayfield — Doubling down on early-stage AI and adding serious infra firepower with Adit Singh. (Bloomberg TV + TechCrunch)

Cognichip — Using AI to rethink how chips get designed. (Forbes)

Lumilens — Betting the future of AI data centers runs on light. (Reuters)

Latitude — Stripe and Uber alums making stablecoins work for everyday local payments. (Fortune)

TAR — Building off-grid power for AI data centers. (Bloomberg)

@IshansharmaDr, @StefanoErmon, @NavinChaddha, @karenlee, @FarajAalaei, @cyril_mathew@cyril_mathew @cyril_mathew, @maggieOphilbin

[Mickayla Williams @MickaylaWill]

@vijaychattha this summer has been hot all right 🔥

[vijaychattha @vijaychattha]

The next frontier in AI is turning powerful general models into specialized systems that reason better, perform reliably, and improve for each business.

That’s the opportunity @DeepCogito is going after.

Founded by @drishanarora and @drvdhruv—who helped build Google’s AI Search products—Deep Cogito just raised a $43M Series A led by @TQVentures, with @ericvishria at @benchmark, @nexusvp, @Atreidesmgmt, @spc, and @VSCVentures.

The round brings total funding to more than $56M—and @zscaler joined as both a customer and strategic investor.

Deep Cogito has already released open-weight models spanning 3B to 600B+ parameters. Now it’s bringing frontier-lab-level post-training to companies that want AI built around their own data, workflows, and competitive advantage.

Every major AI lab post-trains models for its own products.

Deep Cogito is giving the rest of the market that same power.

Proud to have @VSCVentures join the round and have @VSCPR help introduce the company to the world in today's @WSJ

[Mickayla Williams @MickaylaWill]

@EyeingAI Watched this come together, and @truefoundry nailed it. Bring your own models, bring your own infra, pay less doing it. That's not a small thing to ship.

[EyeingAI @EyeingAI]

OMG.

Someone just open-sourced the thing Claude wants you to pay for.

TrueFoundry just launched TrueForge.. a free, open-source framework for building serious AI agents on whatever stack you want.

Use OpenAI, Anthropic, GLM + 20 others.

And apparently, 50% cheaper too: 🧵

[MachGen AI @MachgenAI]

Advertising is entering its personalization era. The bottleneck was never creativity – it was compute.

MachGen removes it: up to 6x faster, 2–3x lower cost, with no quality tradeoff.

That takes an agency from one ad to 100 targeted variants by city, audience, or budget.

Trusted TV is already doing it in production: video generation down from minutes to ~10 seconds, with 10,000+ campaigns live.

Personalized, high-volume creative is becoming table stakes.

We're building the infrastructure underneath it.

Full conversation with Ian McCue → https://t.co/cJSUV9wEYj

[Mickayla Williams @MickaylaWill]

@ericrandall @MachgenAI @ViduAI_official pretty sick!

[MachGen AI @MachgenAI]

@ViduAI_official is now on MachGen: Generate a 5-second 720p video in ~6 seconds and a 5-second 540p video in under 3 seconds.

⚡ Vidu Q3 Turbo, now in real time. Available for free on MachGen for the next two weeks.

Technical blog → https://t.co/Q0kKI7WOc6

[MachGen AI @MachgenAI]

The biggest latency wins in AI inference don't come from the model itself. They come from the system around it. Here's why and how we built MachGen AI.

In our first founder blog, we share our bet on diffusion and world models, what's been built so far, and why we think this is one of the biggest opportunities in AI infrastructure.
→ https://t.co/cY4p0Ar7py

[MachGen AI @MachgenAI]

Today we’re launching @MachGenAI out of stealth.

We build inference infrastructure purpose-built for image and video AI – the diffusion and world models the visual internet is being built on.

The models raced ahead. The infrastructure underneath them didn’t. (🧵)

[Michael Hahn @MichaelHahn26]

This is just the tip of the iceberg. If billing errors this large are hitting individuals, enterprises are absorbing them at scale. Every AI invoice deserves independent verification.

That's why @VauditApp exists. AI billing should be transparent and accountable.

[EyeingAI @EyeingAI]

The timing is interesting too.

TrueFoundry recently launched AI Gateway, MCP Gateway, and Agent Gateway.

Adding Seldon fills a major gap underneath that story:

production-ready ML infrastructure for enterprise customers.

Read more: https://t.co/bY1L1zsDQr

[EyeingAI @EyeingAI]

TrueFoundry just acquired Seldon AI.

The headline is simple.
The meaning is BIGGER.

Enterprises don’t just need LLM apps anymore.

They need one way to run ML models, LLMs, and AI agents together.

Here's what you need to know 🧵

[Rohan Chopra @rohanbchopra]

On the heels of @conveyAI's $38M Series A, I got to sit down with @julesrosenberg at MTS to talk about what it means for operators to move from AI assistants to full-blown teammates that actually own the work.

Thanks for having me on the show!

[Michael Hahn @MichaelHahn26]

AI billing has gotten too opaque for companies to take vendors at their word.

That’s where @Vauditapp comes in. We independently audit AI billing and usage so companies can see what they’re actually paying for.

Thank you @theinformation and @LauraBratton5 for the coverage.

🔗https://t.co/lrFLmXAizP

[vijaychattha @vijaychattha]

VSC companies crushed it in June. 💥 💰 🩺 🏚️

@PWRL_ (Nasdaq: PWRL) discussed public venture capital and providing mainstream retail investor exposure to @OpenAI , @SpaceX , Stripe, @databricks , @Waymo , @krakenfx , @Kalshi , and @Saronic (TITV, Bloomberg Business Week, and MTS)

Reframe Systems is building affordable & sustainable homes faster, cheaper, and smarter. (Forbes)

Mode Mobile is building a new AI stack of products owned by the people. (Business Insider)

@genspark_ai hits a $2.6B valuation and launches AgentBase to kill legacy SaaS.

@conveyAI is helping non-techies within enterprises put AI teammates to work, backed by Andreessen Horowitz, Khosla Ventures, and Pear VC. (Business Insider)

@helium is bringing wireless closer to where people actually use it. (Fierce Network)

Clasp discusses its ingenious approach to helping healthcare students turn debt into a career opportunity. (The Boston Globe)

@Immunaitech is on a dealmaking run with another powerful pharma deal. (Fierce Biotech)

CC: @rohanbchopra , @Mike @benjamindblack Black, @pejmannozad , @vkhosla, @illscience, @ajaykam @maggieOphilbin, @JayKapoorNYC

[Convey @conveyAI]

Enterprise AI cannot stop at adoption.

The next phase is outcome-driven: AI teammates that take recurring, critical work fully off operators’ plates.

Our CEO @rohanbchopra shares how our $38M Series A is a step toward that future.

Link in comments.

[MTS @MTSlive]

What did DoorDash's dispatch system look like in the early days?

@rohanbchopra, co-founder and CEO of @conveyAI:

"There's a guy named Steve. Steve sat in the corner of the office, and every DoorDash driver back then, if you can believe it, when they started doing their dashes, they would actually share their location with Steve on Find My Friends."

"Any time a new order came in, he would take a look at his phone, he'd see where the drivers are, and he'd shoot a text message to the closest driver, 'Can you go pick this thing up?'"

"What we saw is obviously Steve should not have been doing that. He was a really sharp guy in order to keep up with all this stuff, but we said, 'What if we could create a system that did that so Steve didn't have to?'"

"And so that's what we did. We sat next to him, and it wasn't the complex multifactorial optimization machine learning stuff... it was something very simple, just mimicking what Steve was doing. And with those changes, it was a massive unlock."
@julesrosenberg

[Mickayla Williams @MickaylaWill]

@julesrosenberg @MTSlive @rohanbchopra @conveyAI no actually