Jump Capital

VC fund. Operator mindset. Backing founders modernizing industries.

[Jump Capital @jumpcapital]

Well, that escalated quickly!

4 months after we announced our investment in @vinylequity, @MorganStanley announced plans to acquire the company.

Huge congrats to Rob, Poornaprajna & team!

What did @yshkolnik see in Vinyl just four months ago?

Get her take on what made Vinyl worth backing just four months ago (and why it's an especially timely read):https://t.co/e36ecMIMDK

[Jump Capital @jumpcapital]

Inference has spent years getting better at generating tokens.

But agents don't get paid to generate tokens. They get paid to finish jobs.

Optimizing individual requests won't solve the whole workflow.

@pasha_aqil talks about why inference needs a rethink: https://t.co/k0xgLnHsbr

[Jump Capital @jumpcapital]

AI is changing the shelf life of security assumptions.

Researchers used AI to demonstrate that previously impractical PHP vulnerabilities could be exploited. Discovery is accelerating, but patching can't always keep pace.

Saaya Pal + Eliza Chamberlain on why exploitability matters more than vulnerability counts:

https://t.co/Tbd5A5tKbo

[Pluto Security @pluto_security]

The verdict is in: Pluto is officially out of stealth with $42M in funding! ☄️

Pluto isn’t just a planet - it’s where AI gets the freedom to thrive, safely. 🪐

We started with a simple belief: security should enable businesses to move faster, not hold them back. One year later, Pluto secures 1.5M+ endpoints and is trusted by dozens of enterprises.

That belief extended beyond our product; we’re committed to helping everyone deploy AI safely through cutting-edge research, open-source tools, and a people-first security community.

Huge thanks to our incredible team across Israel and the US, to Jump Capital, TLV Partners, and all the amazing investors backing us, and of course to the customers who trust us with their most critical environments.

[Jump Capital @jumpcapital]

Proud to back @trmlabs as they launch Ardence, bringing AI-powered, all-source investigations to law enforcement, defense, and the private sector.

[TRM Labs @trmlabs]

Introducing ✨Ardence✨, TRM's AI investigation platform for all-source intelligence.

Follow connections across sources, keep the source trail, and put investigator judgment at the center of the work.

Learn more 👉 https://t.co/vOLgkC3pVq

[Jump Capital @jumpcapital]

@pluto_security is out of stealth, and we led their Series A.

The endpoint is no longer just a laptop. Shahar, Gil, and Tal are building the next generation of endpoint security for what it’s become.

Congrats to the entire team and to @TLVPartners for backing them from the start!

More than 1 million endpoints and 30+ enterprise customers in under a year. @saayanath and the whole Jump team are proud to be along for this journey! To Pluto and beyond!

[Jump Capital @jumpcapital]

AI may not shrink product + engineering teams. It could make them bigger.

@nikesharora says @PaloAltoNtwks won’t cut its ~6,000 engineers even if AI doubles productivity. If AI also reduces the support required around building and shipping products, more resources could move toward the functions driving innovation.

@sachchitnis + Will Walker on what AI does to the org chart:
https://t.co/UxDljRMRhY

[Jump Capital @jumpcapital]

AI adoption isn’t the same as AI ROI.

9 in 10 CEOs @EYnews surveys can’t show where the return lands in the income statement, even as productivity rises and spending continues.

We think whoever connects AI usage to business outcomes could end up building something much bigger than an ROI dashboard: https://t.co/VTXg9KxQqO

[Jump Capital @jumpcapital]

Periodic pentesting made more sense when sophisticated attacks were expensive and relatively scarce.

AI is changing both sides of that equation.

The emerging opportunity: continuous offensive validation.

Saaya Pal + Eliza Chamberlain:
https://t.co/lLjiELqbTk

[Jump Capital @jumpcapital]

@GeminiApp had valid credentials. The actions it took made sense for the task it was given but it still reached the wrong companies.

That’s the problem with treating authorization as “does this actor have access?”

For agents, the question increasingly becomes: should it be allowed to take this action, against this resource, right now?

https://t.co/obC0WoGQC3

[Jump Capital @jumpcapital]

@Wipro is putting 1,500 FDEs into the field.

Will Walker recently wrote about why AI could create a new generation of systems integrators: more software means more complexity to implement, integrate and operate.

Wipro just put a real number behind that shift, and a reminder that incumbents are coming for the opportunity too 👉
https://t.co/BzqtB6sKs7

[Jump Capital @jumpcapital]

Living rent-free in @yshkolnik's head: the codebase is starting to move faster than the organization’s understanding of it. Her latest blog argues that we need a new layer for what comes next: codebase intelligence.

Everyone downstream of the merge needs to catch up:
https://t.co/U80MUguMsv

[Jump Capital @jumpcapital]

The latest Jump Off Point is out.

This month: what AI does to the product org chart, why the core may not be the real core banking problem, where security moves as agents gain autonomy, what’s happening with cyber funding, plus portfolio news, events + who’s hiring.

Catch up with us↓
https://t.co/TPIovhDlVb

[Jump Capital @jumpcapital]

@OpenAI's @huggingface report is a good example of why “who has access?” is becoming the wrong question for agents.

An agent can start with legitimate access, discover new credentials, tools and paths, and end up somewhere its original permissions never anticipated.

Access is becoming a chain of decisions, not a single one.

@saayanath, @pasha_aqil + Eliza Chamberlain on why the control layer has to move closer to execution: https://t.co/obC0WoHorB

[Jump Capital @jumpcapital]

80% of people in @McKinsey's latest State of AI survey say AI has made them more productive. Only 37% of organizations attribute any EBIT impact to it.

The difference? Nearly 3/4 of AI high performers have fundamentally redesigned workflows around AI, vs. about 1/4 of everyone else.

Giving people better tools isn’t the same as changing how a company works.

Will Walker + @sachchitnis on why AI transformation may be less of a tooling problem and more of an inertia problem: https://t.co/VTXg9KxQqO

[Jump Capital @jumpcapital]

@CaterpillarInc has roughly 1.6 million connected assets generating more than 16 petabytes of structured data. Those numbers tell us more about where physical AI actually is than another robotics funding announcement.

CTO Jaime Mineart told @TechCrunch that the hard part isn't the machine itself. It's integrating the technology into customer jobsites and workflows. That's the shift @pasha_aqil has been watching.

As robotics moves from demos to deployment, the useful signals change. Fleet size matters. Utilization matters. Production data matters. Because once machines are operating in the real world at this kind of scale, the problem becomes keeping them reliable there.

That changes where the infrastructure opportunity sits. Aqil's argument: robotics is entering its operations era, and the tooling around reliability, observability and production data hasn't caught up to the machines.

Give it a read 👉 https://t.co/ueAdmi2lFt

[Jump Capital @jumpcapital]

Cybersecurity companies are raising bigger rounds earlier. But more capital also comes with a higher bar. Or as, Jump Capital Partner @saayanath put it in a @SPGMarketIntel article, “The bigger rounds buy you the scope to actually build a platform, but they also compress the window to prove it.”

We unpacked what’s behind the shift, how AI is changing what security companies need to build, and why capital is increasingly concentrating around fewer, higher-conviction bets: https://t.co/Mdzu2wmUqy

[Jump Capital @jumpcapital]

Congrats to the @TrueFoundry team on the agent harness launch!

TrueForge is:
- Vendor-neutral: any model, any MCP server
- Open source (MIT license)
- Benchmarked at up to 50% lower cost than Claude Managed Agents, with the same accuracy

Worth a look if you're building with agents.

[TrueFoundry @truefoundry]

https://t.co/10IB5MzXJl

[Jump Capital @jumpcapital]

Identity authorizes what an agent can do. Runtime monitors and enforces what it actually does. Endpoint sees where it's doing it.

Do you need all three?

Our read from @BlackHatEvents: not one vendor absorbing the other two, but consolidation organized around the moment of action rather than the moment of access.

Execution is the only moment all three are adjudicating the same event: https://t.co/WSobsyirSb

[Jump Capital @jumpcapital]

"Agents will replace engineers" and "agents just make slop" are both wrong.

The real shift: execution goes to zero, judgment concentrates.

Sach & Will explore the shift: https://t.co/mGj335tzpa