Replicated

@polyaivoice congrats on the raise!! Now use some of that money to scale your self-hosted deployments.

PolyAI has raised $200M from Nvidia, Khosla Ventures, and multiple top VCs. We're one of the fastest-growing companies in the UK, and we handle 500M+ calls for: • Marriott • PG&E • Gordon Ramsay's restaurants • And 3,000 more real deployments Which means that if you've ever called them, chances are you've talked to our voice agents. Every restaurant we onboard books thousands in revenue within 30 days. But how? Because PolyAI works 24/7, answering every call in <2 seconds, and we also: • switch between 45+ languages • handle payments & cancellations • verify identities • and even upsell your services If you want to try creating an agent with PolyAI, we built Agent Studio Lite to make it easy. Just enter any URL, and in 5 minutes it will analyze your website and build a working agent. We're opening early access to a limited number of people. Comment "PolyAI" and we'll add you to the waitlist and give you 3 months for free!

@getbluejay_ai not seeing that you're available on-prem. Want to change that?

@TakoViz I could see a use case for your customers wanting to control their own data here -- are you thinking at all about offering a self-hosted version?

Tako now supports the Machine Payments Protocol (MPP), so your agents can now pay for real research. It's simple: just set your agent up to pay with Stripe or stablecoins via Tempo and now your agent can autonomously discover and pay for Tako’s APIs, unlocking real-time, authoritative answers across markets, startups, web + app analytics, public opinion, geopolitics, sports, weather, and even your own data. MPP removes even more of the friction separating your questions from great answers. Check it out!

@sonalgoyal yup, gotta deploy on-prem!!!

4/Think about what that means in practice: → A procurement agent that can't read your ERP → A sales agent that can't query your CRM → A support agent that can't see your ticketing history Powerful AI. Crippled by locked data.

1/Vibe coding has everyone debating whether SaaS is dead. Both sides — "just build it yourself" vs "a POC isn't a product" — have a point. But they're both arguing about the wrong thing. 🧵

@reggitales “every datapoint from your workspace” is exactly where SaaS-only starts to break. move beyond SaaS — use Replicated to ship Dex into customer-owned environments

Introducing Dex: the self-driving workspace for operators. Dex is the first agent system with full operational context and a self-updating knowledge base. Every datapoint from your workspace is ingested, synced, and structured into compounding context for agents to take action. Comment "DEX" or tag @dexbythirdlayer access. First 1,000 sign-ups get 7 days free. After, join our rolling waitlist. Sign up at joindex [dot] com for a fun surprise. How dex works (threads)

@dexbythirdlayer agents reasoning over every Slack thread, email, meeting, and deal update is exactly where SaaS-only starts to break. move beyond SaaS — use Replicated to ship Dex into customer-owned environments

Karpathy was right about personal knowledge bases for LLMs. But markdown files alone won't cut it. Dex captures raw operational events - every Slack thread, email, meeting, deal update - and structures them into a live context graph that agents can reason over at scale. Personal knowledge bases are for you to read. Operational knowledge bases are for agents to act. Get early access at joindex[dot]com

@cleric_io an AI SRE agent building memory of your production environment is exactly the kind of thing enterprises will want self-hosted. Replicated makes that deployment motion a lot easier

How fast does an AI SRE agent learn your environment? Background discovery starts immediately. Episodic memory improves with each investigation. Skills take a week or two. Shahram and Willem break it down. youtu.be/8iDfsiAnt2M

@raindrop_ai AI agents debugging production infrastructure inside shared SaaS environments is gonna create some very interesting security conversations. Replicated saw this coming

Introducing Raindrop Triage.

@GoZen_Tech AI agents are accelerating the shift back toward customer-owned infrastructure. that’s why we built Replicated

Gartner just confirmed it. 40% of enterprise apps will embed AI agents by the end of 2026. That's not a prediction anymore. That's a deadline. And the question isn't whether your organisation deploys AI agents. It's the workflow you start with.

@OptroHQ Agree completely! Humans can easily create security vulnerabilities when working with these new tools.

Many people think that the biggest AI risk surface is the technology, but it’s not. Our recent survey of more than 800 GRC leaders confirms that, really, it’s human behavior. Learn more about these findings in our new report: optro.ai/resources/eboo…

@IraLHorowitz the more autonomous enterprise agents become, the more customers are gonna want them running inside their own environments. good thing Replicated exists 👀

Ever wonder what it looks like when conversational AI, autonomous agents, and enterprise search actually work together? ServiceNow Otto is making that a reality across the entire enterprise—employees ask, Otto handles the rest. 🚀 sa.servicenow.com/fud46mCLpHpw

@leendotdev security teams are increasingly gonna want those connections happening inside their own environment. good thing Replicated exists 👀

Connecting two souls can take years. Connecting your security data shouldn't.

@dedaluslabs @Techweek_ BYOA but eventually BYOC too. Enterprises are gonna want those agents running in their own infra

We’re heading to NY Tech Week in June! Here’s our full event lineup:

Hearing a new trend around the industry - enterprise AI companies are worried about their self-hosted software being reverse engineered by customers. As more vendors ship AI into customer-owned environments, protecting model IP is becoming just as important as supporting self-hosted deployments. That's why we're here.

@stackgenai gotta make sure that data is secure, gotta make sure that data is on-prem!

Incidents don't fail due to lack of data. They fail due to lack of shared reality. We’ve built incredible observability stacks, yet when things break, chaos still wins. Not because systems are complex, but because every team sees a different truth. Infra is green. Apps are failing. Network says “all good.” And suddenly, alignment not technology becomes the bottleneck. Bridge calls turn into debates. RACIs turn into confusion. Ownership becomes unclear. This is the gap AI SRE is closing: → From more data → to connected context → From alerts → to ranked hypotheses → From war rooms → to shared truth Because incidents aren't resolved with more dashboards. They're resolved when everyone debugs from the same reality. That's when DevOps finally works the way it was meant to. Full blog → stackgen.com/blog/managing-… #SRE #IncidentManagement #MultiCloud #DevOps #AISRE #StackGen

@dip_ak and saying yes to big on-prem contracts 😉

Scaled a CIAM platform to 1B+ users. The skill that mattered most wasn't coding or selling. It was saying no. No to bad customers. No to wrong hires. No to shiny features. No to fast money. Every yes costs you a hundred better opportunities.

@logicballs_ai love this! Hopefully also offered on-prem so your customers control their data?

Most AI tools confidently make things up. We built one that doesn't. LogicBalls AI asks clarifying questions before answering—so every response is accurate, grounded, and trustworthy. No more fabricated sources. No more fact-checking. Just AI you can actually depend on.

@juliusai see that you all offer VPC deployments. Ever considered air-gapped?

@aionxyz some of this is because enterprises aren't owning their data -- deploying on-prem unlocks a lot of opportunities

Enterprise AI spending is projected to hit $1.5T in 2025, yet 95% of orgs report $0 ROI🤯 The culprit? A massive diagnostic failure. Most companies select the tool before they define the problem. In 2026, AI success = Data Readiness + Governance. Our blog documents why most enterprises (incorrectly) create an AI strategy that is a "solution in search of a problem"-- and how to course correct.

@Spezzer this is very cool. Are your on-prem customers happy with their experience? That's where we come in 👀

MCP Apps are a better replacement for elicitation We build an enterprise MCP gateway, and on first use of a service, users have to authenticate with the underlying MCP service. MCP apps can be a rich, in context experience that supplants elicitation.