Ofer Mendelevitch

Fractional AI Devrel
Author, "hands-on-RAG for production" (

[Ofer Mendelevitch @ofermend]

Yesterday I received my author copies of the "Hands on RAG for production" - it was very exciting to hold in my hand the actual physical copy, after so many months of work.

https://t.co/M5DzN6jn3F

@OReillyMedia

[Band @band_hq]

Are you ready?? San Francisco, we're coming atcha! It's time for our Dark Factory Buildathon, running together with @hackersquadio - taking place in San Francisco and remotely. Kicking off on November 13th at the AWS Loft, the build week will run virtually (so you can join from anywhere!) until November 20th.

We've got amazing prizes (see the Luma for more details!) and some fantastic guys joining us - thank you to @nimble_search, @Cognee_, and @Sierra_Ventures.

Register here: https://t.co/Q61UCSbxc7

[Ofer Mendelevitch @ofermend]

In chip design and manufacturing, the AI agent failures I'd worry about most don't crash anything. They return a plausible - yet wrong - number, and no one notices.

Our awesome-agent-failures list has 60+ case studies of agents going wrong, and none of them come from a semiconductor company yet.

Here is what the same failure modes would look like on a design or fab team:
🔹 Wrong revision: the agent answers from an outdated datasheet or errata rev, and sounds just as sure as it would with the current one.
🔹 Response hallucination: a voltage limit, timing margin or process spec that appears in no source document.
🔹 Incorrect tool use: an agent-written EDA or test script that runs cleanly but calls the wrong flow or parameter.
🔹 Premature "done": the agent reports verification complete after checking only some of the corners.

In most domains a hallucination is embarrassing. In the semiconductor industry it can mean a respin, a yield excursion or a field return.

How we approach this at @vectara:
✅ Answers grounded in your own documents, with citations back to the source
✅ Robust handling of complex documents including tables, images, and other multi-modal data
✅ Specialized models for the semiconductor industry: https://t.co/DMIsCaz31M

The full list of agent failure case studies and mitigations: https://t.co/fYD6mUdmuK
PRs welcome.

#AIAgents #Semiconductors #RAG #AIReliability

[Ofer Mendelevitch @ofermend]

@GoogleAI Antigravity is now supported in Band Desktop, and Gemini-4 will enable even better outcomes with an AI Software Factory, e.g. with Band Desktop.

[Ofer Mendelevitch @ofermend]

@elonmusk This is great to see, and adds AntiGravity as another viable option for coding agents that can be part of the AI software factory.

[Ofer Mendelevitch @ofermend]

@swyx @antigravity @saranormous Yes impressive results.
AntiGravity is also now supported in Band Desktop (for to use in the AI Software Factory). So far it was mostly adverserial review between Codex and Claude Code, now there's another really good third option.

[Ofer Mendelevitch @ofermend]

Super excited for the AI Engineer Code Summit in SF later in November. Please come by to say hi at booth S22 if you are going

[Ofer Mendelevitch @ofermend]

I find myself more and more using multiple coding agents on the same problem. Each one with a defined role, and in many cases using different harnesses and LLMs.

But adding more coding agents doesn’t automatically mean more progress. When several agents collaborate on the same task, keeping their plans, decisions, and handoffs connected can become a job of its own.

This is why agent teams need a shared view of ownership, progress, blockers, and decisions, to keep the work coordinated.

https://t.co/pAy1w9SQEY

[Ofer Mendelevitch @ofermend]

I want to repeat my suggestion from a few weeks ago here, which I think is a good way to separate talk from action: every frontier lab should allocate (and demonstrate that allocation by external audit) 50% of their budget towards AI safety - not just research, but also GPU resources, the whole thing.

[Ofer Mendelevitch @ofermend]

@NewsFromGoogle @GeminiApp @antigravity @GoogleDeepMind Antigravity is now formally supported in Band Desktop, so you can use Gemini 4 Argon for your AI software factory, for even more diversity in harnesses and models.
Download here: https://t.co/XN9Ha0oFIl
@band_hq

[Ofer Mendelevitch @ofermend]

@deepseek_ai @DeepSeekHarness deepseek-v4.1-flash doesn't only show great performance it's also a top model on @vectara's hallucination leaderboard with only 4.1% hallucination rate.

https://t.co/6o3gfMlxKk

[Ofer Mendelevitch @ofermend]

@antigravity Gemini 4 Argon just added to @vectara hallucination leaderboard: https://t.co/6o3gfMm5zS

[Ofer Mendelevitch @ofermend]

Huge congratulations to the winners of the AI Dark Factory hackathon.

[lablab.ai @lablabai]

Nobody watched the code get written. That was the whole point.

The Dark Factory challenge asked teams to build the thing that builds the app. 167 submissions came in. Here are the six that won ↓

https://t.co/MCVwUibOh1.

[Ofer Mendelevitch @ofermend]

Excited to share that Band for Developers is live.

Software development is moving beyond a single coding agent working in a terminal. With Band, multiple coding agents can collaborate to solve a problem together - taking on longer-horizon tasks and coordinating their work from start to finish.

I’m especially excited about the ability this unlocks to run adversarial workflows, bringing together different harnesses and models (Claude Code, Codex, CoPilot, OpenCode and so on) so agents can challenge each other’s assumptions and explore different approaches.

[Band @band_hq]

Keep your coding agents. Stop being the go-between.
BAND for Developers is live

[Ofer Mendelevitch @ofermend]

Hey @OpenAI, seems like agent sandboxing is not enough - we may need employee sandboxing too.

[Ofer Mendelevitch @ofermend]

LLMs suck at production.
Small, task-specific models don’t.

We’re at AI Engineer NYC at Booth #40! Drop by to chat about how @_SpecificAI_ helps you automatically distill task-specific SLMs that actually perform in production.

Plus, join the raffle to win some iconic LEGO sets:
👾 LEGO Icons PAC-MAN Arcade Machine
🛵 LEGO Ideas Wallace & Gromit Motorcycle
⚡️ LEGO Pokémon Eevee

#AIEngineerNY #SpecificAI #SLM #AIEngineerNYC #LEGO #AIENY

[Ofer Mendelevitch @ofermend]

Excited for great conversations about task specific SLMs with AI engineers at @aiDotEngineer in NYC, and always great to catch up with good friends.
@seldo @swyx

[Ofer Mendelevitch @ofermend]

@j_asminewang Just to repeat my suggestion from few weeks ago (that got no response): if @OpenAI, @AnthropicAI and other frontier labs are truly worried about "pacing themselves" - they need to invest 50% of their budget/resources into AI safety.

Today!

[Ofer Mendelevitch @ofermend]

@Yuchenj_UW Have you tried @band_hq Desktop? With it you can build a team of coding agents collaborating on a task. And they can be from different harnesses.