TechFollow

Tracking who the tech people suddenly started following.
#AI #founders #researchers
signals, not noise.

[TechFollow @TechFollowrazzi]

🚨 @karpathy followed @thdxr

@thdxr is building OpenCode (@opencode), an open-source coding agent at @anomalyco

[TechFollow @TechFollowrazzi]

🗞️ @rauchg followed @alighodsi

@alighodsi is steering @databricks deeper into agent infrastructure, with recent work around Lakebase, Agent Bricks and security tooling like Lakewatch as the company chases the enterprise AI stack end to end.

[TechFollow @TechFollowrazzi]

🗞️ @ClementDelangue followed @jainarvind

@jainarvind is building @glean into a full work AI stack, moving from enterprise search into chat, agents and workflow automation as more companies try to make AI useful inside messy real workplaces.

[TechFollow @TechFollowrazzi]

🚨 @dkhos followed @jainarvind

@jainarvind is the CEO of @glean, an AI-powered enterprise search platform that is redefining how businesses discover and utilize their internal knowledge.

[TechFollow @TechFollowrazzi]

🗞️ @jerryjliu0 followed @steijnpelle

@steijnpelle is building @LassieAI, an AI back office for health practices that handles claims admin, payment posting, and reconciliation so clinics can get paid faster without piling more work onto staff.

[steijnpelle @steijnpelle]

Today, we're introducing Lassie and $47M in funding led by a16z.

We're building AI that runs small businesses, starting with doctors' offices.

Lassie is already trusted by 700+ practices across the country, working autonomously to provide them with 30 hours of labor per month.

To get here, we first had to leave Robinhood and Superhuman to work in offices ourselves.

Here's how that went.

[TechFollow @TechFollowrazzi]

🗞️ @rauchg followed @jgreze

@jgreze is building @TownAI, an AI assistant for work that handles meeting prep, email follow-ups and routines tied to your actual calendar and inbox instead of waiting for you to open another dashboard.

[TechFollow @TechFollowrazzi]

🚨 @swyx followed @jgreze

@jgreze is the co-founder and CEO of @TownAI, building an AI assistant that handles inbox, calendar, docs and workflows across tools like Slack, email & iOS.

[TechFollow @TechFollowrazzi]

🗞️ @swyx followed @lina_colucci

@lina_colucci is building @LemonSliceAI, a real time interactive AI video model that turns voice agents into live animated characters and pushes video from passive output into a two way medium.

[TechFollow @TechFollowrazzi]

@thekitze if someone follows over 1k ppl, they’re off my radar usually. 1k is really the sweet spot.

[TechFollow @TechFollowrazzi]

🗞️ @cHHillee followed @_doubleAI_

@_doubleAI_ is building Artificial Expert Intelligence, starting with WarpSpeed, a system that rewrites GPU kernels and recently beat expert written NVIDIA code on cuGraph and Blackwell benchmarks.

[doubleAI @_doubleAI_]

WarpSpeed, our autonomous optimization agent at @doubleAI, just took first place on @NVIDIA's new SOL-ExecBench: 235 of the hardest CUDA kernels in production.

But the more interesting story is what we found along the way.

Verifiers designed for human errors don't defend against AI reward hacks. We found four ways the same benchmark's verifiers can be silently fooled. The first one broke transformer training.

🧵

[TechFollow @TechFollowrazzi]

🗞️ @AnimaAnandkumar followed @huskydogewoof

@huskydogewoof has been going deep on loop models, world modeling & recurrent reasoning, with recent work around Parcae, EqR and a fast growing map of the loop model literature.

[Benhao Huang @huskydogewoof]

🌀 Introducing 𝐄𝐪𝐮𝐢𝐥𝐢𝐛𝐫𝐢𝐮𝐦 𝐑𝐞𝐚𝐬𝐨𝐧𝐞𝐫𝐬 (𝐄𝐪𝐑) !

Feedforward models and weight-tied models behave very differently on hard reasoning generalization.

EqR pushes this difference to the extreme by learning 𝐭𝐚𝐬𝐤-𝐜𝐨𝐧𝐝𝐢𝐭𝐢𝐨𝐧𝐞𝐝 𝐧𝐞𝐮𝐫𝐚𝐥 𝐚𝐭𝐭𝐫𝐚𝐜𝐭𝐨𝐫𝐬 .

• Sudoku-Extreme: 99.8%
• Maze: 93%

#ICML2026

[TechFollow @TechFollowrazzi]

🗞️ @steipete followed @mattrubens

@mattrubens is building @roomote after the Roo Code run, taking what worked in agent driven coding and pointing it at the constant queue of engineering chores that usually slow teams down more than feature work.

[TechFollow @TechFollowrazzi]

🗞️ @soumithchintala followed @MiniMax_AI

@MiniMax_AI recently dropped M3 with 1M context, native multimodality and a coding push through MiniMax Code, while @Hailuo_AI and Music keep widening the company beyond text models.

[MiniMax (official) @MiniMax_AI]

Introducing MiniMax M3: The First Open-Weights Model to Combine Three Frontier Capabilities

- Coding & Agentic Frontier: 59.0% SWE-Bench Pro, 66.0% Terminal Bench 2.1, 34.8% SWE-fficiency, 28.8% KernelBench Hard, 74.2% MCP Atlas
- MiniMax Sparse Attention scales context to 1M
- Natively Multimodal from Step Zero

API: https://t.co/fHRdSV7BwZ
Token Plan: https://t.co/BDCycxepZw
🚀New! MiniMax Code: https://t.co/GvB4YiB6Ul

Weights & Tech Report in ~10 Days

[TechFollow @TechFollowrazzi]

🗞️ @rauchg followed @coffeewithone

@coffeewithone is building @octolane, a self-driving CRM that reads email, calendar, calls and site activity so deals update themselves instead of waiting for someone to fill in fields.

[TechFollow @TechFollowrazzi]

🗞️ @steipete followed @francedot

@francedot has been building @trycua around a simple need in agent workflows. Give models a real desktop to work on, not a toy sandbox, with cloud machines, local macOS VMs and tooling built for Codex and Claude Code.

[Cua @trycua]

Today we're bringing Cua Driver to Windows: background computer-use for any agent. Claude Code, Codex, or your own loop can drive real Windows apps through CLI or MCP while your desktop stays usable, with true multi synthetic pointer support.

1/6

[TechFollow @TechFollowrazzi]

🗞️ @OfficialLoganK followed @__nmca__

@__nmca__ is an @AnthropicAI researcher working on model behavior and AI safety, with recent public work around sabotage risk, cyber capability and what stronger coding models may do once given more autonomy.

[TechFollow @TechFollowrazzi]

🗞️ @rauchg followed @almoggavra

@almoggavra is making a sharp bet that databases have become too expensive and too complicated and his recent work at Responsive is all about fixing that with object storage native architecture and a shared LSM core.

[Almog Gavra @almoggavra]

1. https://t.co/Op3hkDt3Td has a new website
2. Greek mythology is wild (go read "Mythos" by Stephen Fry)
3. hardcore database engineers can build nicely designed sites that aren't generic
4. slatedb is so good. it just works

[TechFollow @TechFollowrazzi]

🗞️ @steipete followed @mattlam_

@mattlam_ has been on agent tooling lately, shipping things like `pi-gui`, a remote control layer for Codex and release trackers that turn fast moving agent updates into something builders can actually follow.

[TechFollow @TechFollowrazzi]

🗞️ @rauchg followed @noperator

@noperator is building in the AI-for-security lane, with open work on SiftRank and listwise ranking for vulnerability hunting, plus a recent essay that cut straight through the tortured ways people defend human creativity in the age of AI.👇

[Caleb Gross @noperator]

https://t.co/gOSvgSg8kv

[TechFollow @TechFollowrazzi]

🗞️ @soumithchintala followed @bilaltwovec

@bilaltwovec is working on scaling AI for science at @IsomorphicLabs, bringing a rare mix of pretraining science and performance engineering after time across Nvidia, MosaicML, Google Brain & Cohere.

[TechFollow @TechFollowrazzi]

🚨 @OfficialLoganK followed @bilaltwovec

@bilaltwovec is an AI researcher at @IsomorphicLabs, the AI-first drug discovery company behind partnerships with Lilly and Novartis.

[TechFollow @TechFollowrazzi]

🗞️ @jackclarkSF followed @tantumscollins

@tantumscollins just launched @inherent_labs, a London lab building AI agents for scientific discovery, with a 50 million dollar seed and a striking thesis around recursively improving the research organization itself.

[Inherent @inherent_labs]

We’re excited to introduce Inherent, a lab designed from scratch to build AI agents that discover new knowledge.

The coming era of machine-driven scientific inquiry demands a new kind of research institution and a new kind of AI.

To achieve our mission, we live within the experiment, recursively self-improving the entire research organisation. We investigate questions including:

- What does ‘AI taste’ look like in the sciences, and how can we build an institution that embraces this new aesthetic of discovery?
- What new kinds of human-machine teaming will make the most of AI that can truly innovate?
- How can we build recursive self-improvement at the collective level that continually increases human agency over outcomes?

We have just closed a $50m seed round led by @IndexVentures and @radicalvcfund, with participation from other outstanding investors including NVentures (@nvidia's venture capital arm), @buildexante, Metaplanet, Macroscopic, @MythosVentures, Charlie Songhurst, @chalfs, @jluan, @dwarkesh_sp, @Thom_Wolf, @j_foerst and @maxjaderberg. We are advised by @matthewclifford.

Inherent is a Public Benefit Corporation headquartered in London.

[TechFollow @TechFollowrazzi]

🗞️ @paulg followed @nico_laqua

@nico_laqua is making one of the bolder AI bets in fintech with @usecorgi, building the carrier itself rather than a broker layer and using automation to turn startup insurance into something much faster and more programmable.