cd /news/ai-agents/on-jev-and-athletics · home › topics › ai-agents › article
[ARTICLE · art-139649] src=dev.to ↗ pub= topic=ai-agents verified=true sentiment=· neutral

On Jev and Athletics

A developer has integrated Jev, a decision-making tool from TypeSafe, into The Pipeline Framework (TPF) to handle small, fast, cheap decisions over supplied data rather than relying on a single large model to automate an entire process. The developer argues that daily AI use gives a misleading impression of how easy full automation is, since reliability degrades across many steps and small problems compound. Jev returns structured, typed results from a defined set of possible decisions, which TPF uses to connect AI decisions, ordinary code, human input and real-world actions.

by read2 min views1 publishedSep 25, 2026

I think using AI every day has given us a slightly misleading idea of how easy automation is becoming.

Models have improved incredibly quickly. Give an AI a browser, some skills and a bit of context and it can do impressive things.

So it’s easy to think: surely automating the whole process is next.

I’m not so sure.

We’ve been running 100 metres with 30” hurdles — and still hitting a few — and now we’re talking about running 400 metres hurdles at the Olympics.

A skill that works while you’re watching is very different from something that has to work hundreds of times on its own.

Small problems multiply. A slow AI call repeated ten times becomes a slow process. Something that’s reliable 98% of the time doesn’t look nearly as good after twenty steps. Websites change, logins expire, bot protection gets tougher. And giving an AI access to everything it might need brings its own obvious problems.

This is one reason I’ve decided to integrate Jev into The Pipeline Framework.

Jev doesn’t try to solve all of that. It does something much narrower: make small decisions over supplied data very quickly and very cheaply.

And that compounds beautifully.

Data leads to a small decision, which produces more data, which informs another decision, and eventually the software does something.

That’s much more interesting to me than asking one enormous model to figure out the whole process.

Here’s the lovely coincidence I only noticed after integrating Jev.

The company behind it is called TypeSafe.

And strong typing is also one of the ideas at the heart of The Pipeline Framework (as is the case with many programming languages).

That’s not just a naming coincidence: Jev doesn’t return some prose and hope the application interprets it correctly. You give it a defined set of possible decisions and get a structured result back.

TPF takes the same idea further through the rest of the application: typed data goes into steps, typed data comes out, and those types connect AI decisions, ordinary code, human input and real-world actions.

Jev makes the decision faster and cheaper.

TPF makes the process around the decision something you can actually build software on.

── more in #ai-agents 4 stories · sorted by recency
── more on @jev 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/on-jev-and-athletics] indexed:0 read:2min 2026-09-25 · —