Writing about and working on AI, DSPy, geo, and data. Trying to prevent prompt debt and help others build maintainable intelligent applications.
Bay Area
- Tweets 16,398
- Following 1,185
- Followers 9,962
- Likes 10,099
Drew Breunig retweeted
I'm heading to the GitHub HQ this coming Monday night for AI Show & Tell with @isaacbmiller1 from DSpy, @satyapd from Scalekit, @MathisJoffre1 from Blaxel AI, plus my colleague @cedricvidal.
RSVP: luma.com/wx7493ut
1
3
8
796
Been waiting for this… dbreunig.com/2023/12/27/open…
1
7
2,637
Pay attention to System One models if you are building custom harnesses and agentic workflows.
Great to see DSPy integrate Jev and now support System One models for calibrating agentic workflows.
I'm amazed at how naturally System One models like Jev fit into these frameworks.
I am using Jev to build my own custom harnesses like here: academy.dair.ai/resources/je…
So far, I've had great success with guardrails, routing, and verifiers.
However, I am starting to see other interesting applications in other parts of the agent harness, like:
> structuring skills and better optimizing them
> more efficient tool calling for agents
> dynamic workflows (i.e., dynamically generating harnesses with orchestrator + subagents patterns) that leverage Jev-like models for more optimally structuring outputs/information traveling between these workflows
> enhance coordination between agents
> dynamic and smarter loading (like tools and MCP) of system prompts, and really, all areas of context engineering that benefit from smarter and faster decision-making capabilities
> RLM + System One Models
And the list goes on and on. Just a wonderful time to explore new ways to work with agents.
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
17
11
82
11,152
Drew Breunig retweeted
whoa! so clean. the more the world changes, the more dspy stays the same at the interface level and very very different below the surface
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
9
13
154
11,881
Drew Breunig retweeted
DSPy methodology 🤝 System One
Program, don't prompt!
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
12
32
327
36,694
We’re considering hosting a short, casual @DSPyOSS webinar next week to run through Jev support and design patterns. Let me know if you’re interested, and what you’d like to see covered.
14
5
87
3,015
Drew Breunig retweeted
this is an honor! 🥹
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
3
13
233
32,701
Introducing ReAnchor, a DSPy optimizer built for Jev and other System One models. cmpnd.ai/blog/building-jev-p…
8
35
237
11,477
Drew Breunig retweeted
DSPy 🤝 Jev
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
6
34
2,014
Just released Jev support for DSPy, with a new optimizer that calibrate’s Je’s output thresholds against your metric.
Works with existing DSPy signatures and metrics, can switch between LMs and Jev with one line.
DSPy 3.4.0 was just released!
This release includes native support for Jev and System one models inside of DSPy! Use it with compatible signatures. This release also includes a brand new optimizer, ReAnchor, specifically for calibrating outputs with confidence.
4
7
92
6,243
Am really excited to build compound AI programs with this, it’s so easy. Here’s a lazy dungeon master program that delegates to Sol if Jev isn’t confident or decided the moment warrants a plot twist: gist.github.com/dbreunig/67d…
3
248
If you want to understand Meta’s camera-less glasses launch and Apple’s new always-listening Watch, read this piece I wrote in 2024 about Audio Augmented Reality (AAR), its history, future, and why it’s set for takeoff. dbreunig.com/2024/04/10/the_…
1
1
9
817
Drew Breunig retweeted
Terminal-Bench meetup tonight *with* swag. Register if you haven’t already! Benchmarks, RL environments, new features, and state of the union
luma.com/tbench
17
10
115
11,263
Appreciate this pushback from @JensenHuang, that this AI is just (revolutionary) software and labs need to transition their resource allocations.
1
2
12
688
AI engineering is more like law than board games. every.to/thesis-statements/d…
2
20
1,235
Anthropic's guidance on prompting Opus 5.5 is an excellent example of why you should be designing systems that can be continually be re-compiled with prompt optimization.
Old tricks that worked with Opus 4 or GPT-4o either aren't needed with Opus 5.5 or clash with its training.
15
37
768
60,440
Remember: natural language isn’t great for defining long-lived systems. dbreunig.com/2026/06/22/the-…
1
13
1,938
Drew Breunig retweeted
13 days until #CNG2026!
3 keynotes at Snowbird, Utah
▪️ @ianschuler – The Great Retooling
▪️ Katie Baynes – Boring at Scale: Consistency as Public Infrastructure in a Distributed Geospatial Ecosystem
▪️ @dbreunig – Using AI Without Losing Your Soul
🔗 2026.cloudnativegeo.org
1
1
289