AI Projects - November 13, 2026

Week of: November 13, 2026

---

Overview

This week's AI Projects beat was dominated by builders shipping rather than theorizing: ArtCraft's AI-assisted rewrite of Photoshop as open source, engineers automating workflows with System One "Jev" models, and Zapier's guide to wiring Grok into existing tech stacks. On the business side, Mercor's $750M Series D at a $20B valuation and the leadership change at NVIDIA framed the commercial backdrop, while new open tooling like llama.cpp v1.0.0 and on-device agents lowered the floor for hands-on experimentation.

Stories

1. ArtCraft used AI to rewrite Photoshop as open source software

Source: Latent Space Link: https://latent.space/p/artcraft/

ArtCraft has used AI to rewrite Photoshop as open source software, according to Latent Space's profile of the company. CEO Brandon Thomas describes a vision that goes beyond a single application: in the piece's framing, he wants to "rebuild everything" and "take back the internet."

For builders, this is a concrete case study in using AI to reconstruct a large, mature commercial product as an open alternative — the kind of project that resets expectations for how fast a small team can reimplement complex software. The stated ambition beyond the clone is the part worth watching, since it hints at a broader platform play rather than a one-off port.

Impact Analysis: Treat ArtCraft as a reference point for AI-accelerated reimplementation of established commercial software, and watch whether the "rebuild everything" ambition materializes into shipped products.

2. Engineers show what they're building with System One "Jev" models

Source: Latent Space Link: https://latent.space/p/jev-revolution/

At the AI Engineer Code Summit, Jev lead creator Diogo Almeida called on developers to automate everything, Latent Space reports. The piece also covers how Glean and others are already using Jev in their work.

The value here is the shift from model demos to production usage: named companies describing how they apply a specific model family is exactly the kind of implementation evidence teams need before committing. Almeida's "automate everything" framing is a useful signal of where the tooling's creators think the leverage is.

Impact Analysis: Look to the Glean and AI Engineer Code Summit examples as templates if you're evaluating Jev for automation-heavy workflows.

3. Zapier publishes a playbook for automating Grok across a tech stack

Source: Zapier Link: https://zapier.com/blog/automate-grok

Zapier's guide covers two main ways to automate Grok with Zapier, spanning Grok 4.7, Grok Bot, and more. It highlights Grok's specific edge — live access to X — while noting that Grok models and agents remain solid alternatives to what OpenAI, Anthropic, and Google offer.

For teams already running work through Zapier, this is a practical integration path rather than a from-scratch build. The emphasis on connecting Grok to the rest of a tech stack is the recurring pattern in applied AI: model choice matters less than whether the output lands where the work actually happens.

Impact Analysis: If you're testing Grok for operational work, start with the Zapier integration rather than building custom plumbing.

4. Mercor raises $750M Series D at a $20B valuation on a $3B run rate

Source: Latent Space Link: https://latent.space/p/ainews-mercor-750m-series-d-at-20b-valuation-3b-run-rate/

Mercor closed a $750M Series D at a $20B valuation with a reported $3B run rate, per Latent Space's November 10 roundup. The same briefing notes that Google made Gemini 4 Carbon generally available.

The scale of the round and the run rate are the headline numbers for anyone tracking which AI businesses are converting demand into revenue. It's also a reminder that the funding environment for applied-AI companies with demonstrated revenue remains aggressive.

Impact Analysis: Use the disclosed run rate as a benchmark when sanity-checking revenue multiples for applied-AI startups in your own pipeline.

5. Consumer sentiment data shapes how brands approach AI influencers

Source: HubSpot Link: https://blog.hubspot.com/marketing/ai-influencers

HubSpot examines AI influencers — content creators that look like real people but are synthetic — and what consumer sentiment data says about how audiences respond to them. The piece frames the issue around the everyday feed experience on TikTok and Instagram, where such accounts are already appearing.

For creators and marketing teams, this is the content-production side of applied AI: synthetic personas as a scalable channel, with audience trust as the limiting factor. Brands weighing the approach need sentiment evidence, not just cost savings.

Impact Analysis: Any AI-generated creator strategy should be validated against consumer sentiment data before scaling spend.

6. llama.cpp ships v1.0.0 as a stronger foundation for local AI

Source: Hugging Face Link: https://huggingface.co/blog/ggml-org/llama-cpp-v1

llama.cpp reached version 1.0.0, with the project framing the release as a stronger foundation for the local AI ecosystem. Specific changes and migration details are covered in the post itself (see source).

A 1.0 milestone for a project this widely used in local inference matters for anyone running models on their own hardware. It signals API stability expectations for downstream tooling and projects built on top of it.

Impact Analysis: Re-check pinned versions and upgrade paths if your stack depends on llama.cpp for local inference.

7. BotPsy brings an agentic assistant on-device with a sub-billion-parameter model

Source: Hugging Face Link: https://huggingface.co/blog/qvac/botpsy

QVAC published a write-up on BotPsy, an on-device personal assistant built on a sub-billion-parameter agentic language model. Implementation and benchmark details are in the post (see source).

Small agentic models that run locally open up assistant projects without per-token costs or data leaving the device. That trade-off — capability versus privacy and operating cost — is central to most practical assistant builds right now.

Impact Analysis: Worth a prototype if your project needs a private, low-cost assistant without a hosted model dependency.

8. Jensen Huang resigns after 33 years leading NVIDIA; Colette Kress named interim CEO

Source: Latent Space Link: https://latent.space/p/ainews-jensen-huang-resigns-after-33-years-leading-nvidia/

Jensen Huang has resigned after 33 years leading NVIDIA, according to Latent Space's November 12 briefing. Colette Kress takes over as interim CEO, and the piece raises the question of what happens to open source and American AI leadership.

For anyone building on NVIDIA hardware or the open-weight ecosystem, a leadership transition at the company that supplies most AI compute is a planning variable — not an immediate technical change, but a reason to track roadmap and open-source posture signals.

Impact Analysis: Monitor near-term NVIDIA roadmap and open-source commitments under interim leadership before locking in long-horizon hardware assumptions.

Source Links

• Latent Space - AI Ambition: ArtCraft Cloned Photoshop and Now Wants to "Rebuild Everything"
• Latent Space - The Jev Revolution: What Engineers Are Already Building with System One Models
• Zapier - Grok automation: How to use the Zapier Grok integration (Grok 4.7, Grok Bot, and more)
• Latent Space - [AINews] Mercor $750m Series D at $20B valuation, $3B run rate
• HubSpot - AI influencers: What consumer sentiment data tells brands
• Hugging Face - llama.cpp v1.0.0 : A stronger foundation for the local AI ecosystem
• Hugging Face - BotPsy: On-Device Personal Assistant with a Sub-Billion-Parameter Agentic Language Model
• Latent Space - [AINews] Jensen Huang resigns after 33 years leading NVIDIA