job; applying SI to real world problems @Samsara otherwise; 🚸🐕⛳⛷️⛰️
Joined September 2011
- Tweets1.8K
- Following1.2K
- Followers1.2K
- Likes7.5K
Pinned Tweet
My favorite part about building product is that I can be told my work is "unstoppably great" and "the 667th layer of hell" in the same day
Dylan Sellberg retweeted
Our Q2 FY27 results: $IOT crossed $2.1B in ARR and added 242 $100K+ ARR customers and 20 $1M+ ARR customers, both quarterly records.
More large enterprises are choosing Samsara to connect and improve their physical operations at scale.
Read more: samsara.com/company/news/pre…
Dylan Sellberg retweeted
shout out to any parent here that shifts from a crazy llm dopamine rollercoaster for 8hrs to the complete opposite of it and trying to play calmly with play doh and magnatiles
Dylan Sellberg retweeted
That’s the load-bearing insight. The quiet part is: this isn’t a bug, it’s a feature.
Zoom out. The takeaway? It just works. Seamlessly. Effortlessly. No notes. Rest is noise.
And honestly? That’s the whole ballgame. Let that sink in.
It’s not X—it’s Y.
Dylan Sellberg retweeted
devastating comeback
We’re sorry to see that OpenAI put out a note saying they plan to block Cursor users from accessing OpenAI models in three months.
OpenAI models serve about 5% of Cursor user traffic, and we’re speaking with the OpenAI team to resolve this.
Cursor was one of the very first users of OpenAI, we’ve worked closely with their team for years, and we’ve trusted their platform to be neutral infrastructure for our business.
Dylan Sellberg retweeted
This is just the beginning of what's coming. Not paid to say this, but I think Harvey provides a glimpse into the future of what successful AI-native companies will look like and how they operate.
Successful companies will need to think of how to build and own their entire intelligence stack. They own the models, agents, and everything in between.
Introducing Tenet, our first model post-trained for legal.
Tenet is a Kimi K3 base that we post-trained with @FireworksAI_HQ on a corpus of publicly available legal data, synthetic data, and human expert data simulating long-horizon legal work.
Training increases Tenet's all-pass rate by 82% on LAB and 22% on LAB Contracts relative to the Kimi K3 base model. It achieves state-of-the-art performance on LAB Contracts and places second on LAB.
These gains generalize to other leading agentic benchmarks including @mercor's Apex Agents - Corporate Law, @crosbylegal's Redline Bench, and @scale_AI's Professional Reasoning Bench.
Tenet is also optimized for token efficiency, operating at less than a fourth the cost of leading foundation models.
We additionally post-trained three specialist models for Tenet to use as subagents:
1) M&A Diligence: post-trained with @baseten on our LAB Diligence environment in an RLM harness, this model is optimized for high-scale, long-horizon tasks.
2) Review Tables: trained with @appliedcompute on our Review Table environment, this model is state-of-the-art and cost-effective at high-volume document review and structured data extraction.
3) Firm Knowledge: trained with @EngramLab on our synthetic law firm environment, this model is optimized to learn and search over a firm's knowledge via memory and structured notes.
More details on model training, environment design, benchmarking, results, and more in the article by @gabepereyra below.
What's next for Harvey’s research?
- Scaling LAB to more jurisdictions, practice areas and workflows
- Scaling compute to bring new generalist models and capabilities to Harvey
More to come soon.
Dylan Sellberg retweeted
AI agents are leaving the browser. 🤖🚚
@Samsara's Beyond 2026 keynote wasn't about writing emails or generating code.
It was about AI helping fleets, warehouses, job sites, and supply chains make better decisions in the physical world.
Here's what stood out. 👇
Dylan Sellberg retweeted
New SOTA public submission to ARC-AGI:
- V1: 94.5%, $11.4/task
- V2: 72.9%, $38.9/task
Based on GPT 5.2, this bespoke refinement submission by @LandJohan ensembles many approaches together
Dylan Sellberg retweeted
This paper from Stanford and Harvard explains why most “agentic AI” systems feel impressive in demos and then completely fall apart in real use.
The core argument is simple and uncomfortable: agents don’t fail because they lack intelligence. They fail because they don’t adapt.
The research shows that most agents are built to execute plans, not revise them. They assume the world stays stable. Tools work as expected. Goals remain valid. Once any of that changes, the agent keeps going anyway, confidently making the wrong move over and over.
The authors draw a clear line between execution and adaptation.
Execution is following a plan.
Adaptation is noticing the plan is wrong and changing behavior mid-flight.
Most agents today only do the first.
A few key insights stood out.
Adaptation is not fine-tuning. These agents are not retrained. They adapt by monitoring outcomes, recognizing failure patterns, and updating strategies while the task is still running.
Rigid tool use is a hidden failure mode. Agents that treat tools as fixed options get stuck. Agents that can re-rank, abandon, or switch tools based on feedback perform far better.
Memory beats raw reasoning. Agents that store short, structured lessons from past successes and failures outperform agents that rely on longer chains of reasoning. Remembering what worked matters more than thinking harder.
The takeaway is blunt.
Scaling agentic AI is not about larger models or more complex prompts. It’s about systems that can detect when reality diverges from their assumptions and respond intelligently instead of pushing forward blindly.
Most “autonomous agents” today don’t adapt.
They execute.
And execution without adaptation is just automation with better marketing.
Dylan Sellberg retweeted
My favorite rebrands of 2025:
Gambling - Prediction Markets
Service Business - Full-stack startup
Sales engineers - Forward deployed engineer
PED's - Peptides/HRT
Private Equity - AI Rollup
GPU Provider - Neocloud
Docker container - RL environment
Series A - Pre-seed
Automation - AI agent
[startup] - AI [startup]
Dylan Sellberg retweeted
Building AI agents right now is a process of:
1. Build scaffolding to address a limitation of the AI model so your agent works
2. The AI model gets upgraded and solves the very problem you were trying to mitigate, rendering your scaffolding obsolete
3. Identify new, harder use case you want to solve and go back to step 1
There’s basically no way to avoid this process because if you don’t mitigate the model’s limitations then you’re dead on arrival, *and* you don’t know which of your mitigations will be surpassed and when.
So the reality is that you just have to accept that you are going to be writing a lot of throwaway code for the next few years, and you have to be very unsentimental to the path you’ve pursued. Just do whatever it takes to make the agents work.
Dylan Sellberg retweeted
One of the most fun parts of OpenAI is watching people here level up so fast and do such excellent work.
We are operating at a high level across many different disciplines and many of the people doing it have never done it before, and joined us at the beginning of their career.
If you believe in people and give them a lot of responsibility and support (and pick the right people to bet on) you will be surprised on the upside more often than you think.
I would love to see more companies operate this way and think we would all benefit.
(This was also one of the most fun parts of startup investing.)
Dylan Sellberg retweeted
The worst mistakes in life are made when you try to do fast what’s meant to be done slow. Real, durable things take a long time to build. Careers. Businesses. Relationships. Health. There are no hacks. There are no shortcuts. The long way is the right way.
Reiterating for those who may live under a rock. @Tesla FSD is ridiculously good and constantly getting better. Most valuable product I own, hands down