Your agent wants permission to run a migration and you’re not at your desk.
Cuttlefish, our iPhone app for Managed Agents, puts the approve button in your pocket with your agent's requests on screen. Send a correction mid-run, or stop it if it goes sideways. https://t.co/GbEKlRQTlY
We're sorry for any inconvenience you've experienced. Did you get a chance to create a ticket for this? If so, please share the ticket number so we can have the relevant team review your request further. If you haven't created one yet, you can create a ticket here: https://t.co/PMRMacNYS2
If you can sign in to your account, please submit a ticket here: https://t.co/2JkPWT0lK8
Let us know if you need any help creating it!
Own your weights, your data, your model and your agent.
Sam Herring of @NousResearch on the principle behind Hermes Agent, and Malcolm Ah Kun of DigitalOcean on running agents safely at scale: API keys never enter the VM. https://t.co/ptJL1V3m9y
@MaksudMansuri15 Hey Maksud, thanks for the feedback. We'd love to hear more! Could you share the details on our Ideas portal? Our team actively reviews submissions to help improve the platform for everyone: https://t.co/YTrwsQPmDn
v5 Droplets served up to 50% more microservice requests than same-size Dedicated Premium, with half the latency at 1,000 req/s.
v5 also had superior performance in every CPU, memory and database benchmark we ran, by more than 2x on memory-bound work. https://t.co/Rb9I7kb6sd
The hyperscalers aren’t losing AI-native startups on price or performance. They’re losing them on onboarding.
AWS and GCP were built for the CIO. The AI developer, and increasingly their agent, never gets past the signup flow.
That’s a distribution problem. No feature fixes it.
Quick answer, it depends on how much you use it. Serverless Inference bills per token, so it’s usually cheaper for light or spiky traffic. A GPU Droplet bills by the hour whether you’re using it or not, so self-hosting only wins at steady, high volume. You also manage the model yourself.
Per-model prices are here: https://t.co/1h4AwXaQhV
Hello there! Thanks for reaching out. Did you get a chance to create ticket in reference to this? If yes, could you please share the ticket number so we can have the relevant team review this request further? In case you were unable to create one, you can create ticket using this link below if you're unable to login to your account:
https://t.co/cfyuSGxqfO
If you want us to create a ticket for you, we request you to share the email address you've used to register on our platform. We'll create a ticket for the relevant team and have them follow up on your request.
We built DigitalOcean as an AI infrastructure company that’s different by design: a full software platform, nearly 700K paying customers with our top 25 at 27% of revenue, and a balance sheet that’s an asset in a noisy market.
@paddix at @TomorrowXSummit: https://t.co/iV9Wxlz5dN
Had a great time at @TomorrowXSummit today talking about how we built our AI cloud.
A lot of the industry went scale first: grab the GPUs and the power, then figure out the software. We did it the other way. Software first, then scale.
Why does it matter? Because nobody running a real AI app wants just a GPU. They want the model, the agent, and the database, storage, and networking around it, all in one place. That's where the value is, and it's a lot harder to build than it is to buy.
Scale can be bought. A platform has to be built.
Wrote more about it here:
https://t.co/JAIVe56lgF
GPT-6.2 Astra and Sol from @OpenAI are live on DigitalOcean Serverless Inference.
OpenAI reports Astra sets new professional & research eval highs at under ⅓ of GPT-6 Astra’s cost per task, and Sol handles everyday and agentic work.
One DigitalOcean account, one bill. https://t.co/GaY4ZgvQYU
The future of AI security might be microVMs.
Unlike traditional VMs, microVMs are lightweight, spin up in milliseconds, and can be destroyed after each task.
If an AI agent’s sandbox is compromised, the damage can be contained rather than exposing the entire environment but microVM is also easier to be compromised. So catch-22.
Agree with @paddix . I expect microVMs to become the default isolation layer not just for AI agents, but eventually for consumer AI applications by foundation models.
Most AI clouds were built around capacity. We built ours around software.
@paddix is live at @TomorrowXSummit on why that makes DigitalOcean Different by Design. Tune into the livestream on X. https://t.co/hbjrO5D8L5