I've talked to hundreds of Enterprise CIO's and CTO's and heads of AI.
The Agent problem in a nutshell:
Step 1. Convert Dollars to Tokens.
Step 2. ?
Step 3. Profit
Everyone is stuck trying to figure out how to turn those tokens back into dollars.
Coding agents is straightforward - either increase top line with more/better products, or increase bottom line by reducing engineers or both
Customer service agents is also doable - They are already highly process focused, documented, often outsourced, etc.
The rest of the enterprise: Very tricky!
to
Codex, build an app exclusively for Apple Homeview using GPT 6.1 that fixes all the Siri shortcomings
It has to be as native and powerful as possible. It should feel like the app was designed alongside the hardware.
1
1,107
Rowan Trollope retweeted
They did it! SpaceX has just caught their Starship rocket for the first time ever, marking the first recovery of a conventional orbital rocket upper stage in history!
This is one of the greatest engineering achievements in history. Congrats to the entire @SpaceX team! 🚀
How I made Muse less forgetful
rowantrollope.substack.com/p…
4
19
150,193
What is going on with Siri? In latest dev 27.2 beta 3 Can no longer create contacts.
Apple - since you’ve been unable to build a competitive Siri, please let your users replace Siri with third party apps which are far superior to the 2010’s era Siri.
Rowan Trollope retweeted
Memory is what allows Claude to learn from experience. In a new Engineering Blog post, we share how “dreaming” helps Claude organize what it’s learned, correct its mistakes, and find what it needs more easily the next time. anthropic.com/research/compo…
If you’re building agents, you know how hard it is to go from demo to prod. Forgetful agents, repeat questions, cracking under actual workloads—it’s the same story over and over.
Memory, routing, context, and traceability are what agents need to thrive. Now, Redis Iris has it all in one runtime. It’s time to build agents ready to deliver in the real world.
Read more about the new Redis Iris: redis.io/blog/agents-are-sof…?…
1
5
120,655
It’s time to build!
If you’re building agents, you know how hard it is to go from demo to prod.
Forgetful agents, repeat questions, cracking under actual workloads—it’s the same story over and over.
Memory, routing, context, and traceability are what agents need.
Now, Redis Iris has it all in one runtime. It’s time to build agents ready to deliver in the real world.
Read more about the new Redis Iris: redis.io/blog/agents-are-sof…?…
We've been cooking at Redis: Announcing GA and greatly expanded Redis Iris.
Why are we building this: Models are very good now, but very few agents are making it to prod...
The gap: CONTEXT. Agents forget what users said, repeat things they've already done, and can’t reach the data spread across your systems.
At Redis, we've been solving these problems for apps since before agents existed.
Iris has:
Agent Memory - state of the art, model-independent memory layer for AI agents.
Agent Mesh - for agent 2 agent messaging
Agent Playbooks, Intent Router, Context Gateway and Lens
Check it out and let us know what you think redis.io/iris/?utm_source=li…
Redis is proud to be a launch partner in the @FireworksAI_HQ Frontier Ecosystem. Check out our docs to use Redis as a retrieval layer with Fireworks.
👉 redis.io/docs/latest/integra…
The future of AI is specialized intelligence with models tuned to your data, your domain and your customers.
Today, we're proud to launch live from Forge the Fireworks Frontier Ecosystem with 60+ partners across the AI stack, helping you build and ship faster.
Check it out: fireworks.ai/partners
Rowan Trollope retweeted
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-sp…
github.com/RikiyaT/lm_und…
huggingface.co/Qwen/Qwen3.5-0…
huggingface.co/RikiyaT/Qwen3.…
“Why aren’t you doing X with agents yet?”
Where X is whatever latest trend / buzzword for the week (factories, swarms, evals, graph engineering, loops, recursively self improving, etc)
Are my LEAST favorite click bait headlines/x posts.
Are we trying to create anxiety in people?
Do these stupid headlines actually work at generating clicks?
2
6
1,430
Of the small non representative group of people I surveyed :
60% of people using personal agents are using 3+ agents.
Agents need to be able to talk to each other.
Stop cut and pasting and try this and lemme know if it works for you. (Instinct not supported)
Goagentconnect.ai
3
2
1,157
Redis has spent years as the fast state management layer next to the application, and in the agent era, that same layer holds agent state. @rowantrollope explains on Tech Talks with AIM: youtube.com/watch?v=b_tjSyL6…







