PLRX

Your operations, run autonomously. The complete agentic stack: execution platform + Enterprise AI Agents. From $0.99/mission. Enterprise Agentic.

[PLRX @plrxainc]

two kinds of ai agents are being built right now.

the first makes individuals faster — prompt-driven, session-limited, useful.
the second runs operational workflows end to end, without individuals in the loop.

we build the second kind.

https://t.co/m9ppgqLLC9

[PLRX @plrxainc]

prompt-driven vs event-driven.

that's the actual line between a personal ai tool and an enterprise ai agent.

one waits for you to type something.
the other doesn't know you exist.

[PLRX @plrxainc]

why isn't the a2a protocol all over my feed?

because personal agents don't need it.

but enterprise agents coordinate across parties, systems, and time.
that requires a protocol.

a2a is what makes multi-agent enterprise workflows possible.
the feed will catch up.

[PLRX @plrxainc]

building autonomous operations in-house requires:

durable workflow engineering.
multi-agent coordination.
multi-model orchestration.
compliance architecture.
production ml.

simultaneously. with engineers who understand all five.

timeline: 15 months — or never.

https://t.co/zcNVMaAfOg

[PLRX @plrxainc]

complex workflows. multiple parties.
agents that resolve every breakpoint — autonomously.

that is agentic at scale.

→ https://t.co/TZcflzkyEx

https://t.co/m9ppgqLLC9

[Block @blocks]

We built an internal AI system called Builderbot. It coordinates agents across our entire codebase. Engineers tag it in Slack, and it researches, plans, and ships. The story so far:

- 200,000 operations per day.
- 1,500 pull requests merged per week.
- 15% of all production code changes across Block.

What used to take months now takes days.

How we built it: https://t.co/ixvmNK87Du

[PLRX @plrxainc]

each agent deploys independently.

one helm chart, one values file per service.
the deployment config lives in one place — not scattered across repos.

when something needs to change, you change it once.
no manual kubectl. no config drift.

@HelmPack @kubernetesio

[PLRX @plrxainc]

friction in enterprise operations looks like:

a human waiting for a callback that should be automatic.
a human re-keying data between systems that should talk.
a human chasing a document that should have been collected.

every one of those is a breakpoint.
enterprise ai agents resolve them — autonomously.

that is agentic at scale.

[PLRX @plrxainc]

grafana shows us the full picture — traces, metrics, logs — in one place.

when a mission stalls, we can see exactly which agent, which step,
which external call, and how long it waited.

observability in an agentic system isn't optional.
the failure modes are too distributed to debug blind.

@grafana

[PLRX @plrxainc]

one of our agents submitted a prior auth request on a monday.

the payer responded 5 days later.

the agent resumed exactly where it paused.
full context intact. no polling thread. no timeout.
no human in the loop.

durable execution is not a feature.
it's a different way of thinking about state.

[PLRX @plrxainc]

two eras.

personal ai agents: runs on your laptop. stops when you close it.
enterprise ai agents: runs on infrastructure. doesn't know your laptop exists.

the market is using one word for both.
that will not last.

[PLRX @plrxainc]

the friction in your operations is not in the logic.

the rules exist. the documents are defined.
the payers have protocols.

what creates friction is the execution gap —
every human step between them.

enterprise ai agents close that gap.

[PLRX @plrxainc]

aikido runs on every push.

dependency vulnerabilities, container image scanning,
infrastructure misconfigurations — caught before they reach staging.

we operate in healthcare. a vulnerability that ships
is not a patch. it's a breach report.

@AikidoSecurity

[PLRX @plrxainc]

what percentage of your operational workflows resolve without a human?

5% → interesting demo
50% → real progress
94% → autonomous operations

we're at 94. in production. today.

https://t.co/m9ppgqLLC9

[PLRX @plrxainc]

enterprise ai agents coordinate.

one validates the document. another scores the risk.
another submits the claim. another waits for the response.

they delegate and hand off via a2a protocol —
without a human in the loop at any step.

that is multi-agent. that is what enterprise requires.

[Neil deGrasse Tyson @neiltyson]

Not that anybody asked, but the last time the @NYKnicks won a championship, the Periodic Table had 11-percent fewer Elements.

[PLRX @plrxainc]

built a distinction into the error handling that took a while to get right.

transient failures — network timeout, service unavailable — get retried.
permanent failures — malformed payload, schema mismatch, invalid recipient
— go straight to the dead-letter queue.

retrying a permanent failure ten times
just delays the real problem by ten.

@apachekafka

[PLRX @plrxainc]

personal ai agent: you type, it responds.

enterprise ai agent: something happens in your systems,
it runs a workflow.

different trigger. different architecture. different outcome.

same name, for now.

[Paperclip @papercliping]

Paperclip v2026.609.0 dropped

💸 Anthropic Fable 5 support
🖼️ Company Artifacts and Rich Attachments
☑️ Checkbox Confirmations
🛡️ Low-trust agents

https://t.co/AOGbP7IffT

[PLRX @plrxainc]

we have a rule: no handwritten request/response dtos.

the openapi spec is the source of truth.
models are generated from it.
controllers implement the generated interface.

when the spec changes, everything regenerates.
there's no gap between docs and implementation
because they're the same thing by construction.

sounds like overhead. eliminates an entire class of bugs.

[PLRX @plrxainc]

renovate bot opened 15 prs last week.

every outdated dependency, every patch, every minor bump —
flagged automatically, linked to the changelog, ready to merge.

the alternative is someone manually checking once a quarter
and discovering a vulnerability that shipped three months ago.

@renovatebot