Very cool resource on security issues in models
huggingface.co/spaces/witnes…
3
755
We always us SGLang for agentic workloads because of the shared radix tree (prefix tree). It allows agents to “share the KV Cache”.
SGLang just adopted a few of the kernels from the new DeepInfer (Deepseeks inference engine) for an even more powerful set of tools.
- The fused many of the GPU actions into Mega kernels for optimization
- The prefix cache is cheaper because of the relay
The results are pretty good (these matter at scale with 1000s of agents running multiple workloads).
Me reading most of AI X now-a-days
"Of course that's your contention. You're a first-year agentic engineer. You just finished three Python tutorials and built your first LangChain agent.
You're gonna be convinced of that until next month, when you read the Go team's piece on cheaper cloud services. Then you'll be in here regurgitating benchmarks, telling everybody Python's too slow and Go's concurrency is the future of autonomous agents.
That's gonna last until you read Discord's engineering blog about switching from Go to Rust. Then suddenly you're gonna be talking about garbage collection, tail latency, memory safety, and how Rust's borrow checker makes agent-generated code safer.
And that's gonna last until next year, when you'll be in here quoting some obscure systems engineer explaining why none of this matters if your agent can't reliably call a tool.
Do you have any thoughts of your own on this matter? Or is that your thing? You read some engineering blog, repeat the benchmarks, and pawn it off as your own architecture?"
1
11
1,440
Real AI native... From the ground up
This isn't for most scenarios and companies... but when you have a unique relationships with "objects" your business - building a custom ERP can allow for some neat addons
for example
1. We have a custom data relationship we set that allows us to correctly connect multiple business, partners, contacts to a PO/SO (system of record problem - solved)
2. Once have that - we make everything an API
3. Then we give that api instructions on how to use, and business semantic layer
4. Then we can use agents to perform most of the work through chat
5. Then we can connect those agents to email/slack, etc and have things like
- auto update profiles
- add notes to orders or parts
- create/raise quality errors and exceptions. etc
6. We can also allow anyone to get a custom summary of reports, tables, transactions, either as notifications or emails.
7. Human in the loop through the agentic inbox
... and we're just getting started.
The goal is start with system of record, add in validation and rule sets, then stack workflows and if/then automations, and finally layer in llms where it makes sense.
Its pretty powerful
We are also building
- agentic workflows with multiple nodes on the backend
- generative UI components for custom canvases
I find Grok bot easy and great for consumer side
For business - I still need to control
- data/memories
- trace llm calls and tool traces
- tweak system prompts, etc
Been using this for close to 6 months and while the setup is a bit, once setup - its hard to beat.
... plus, with the security around signing each tool call and skill change... (and I run this on prem with all company data)
2
7
2,025
I think an organization issue that companies need to solve is good ole fashion
- coordination
- communication
in this ai age.
how do we continue to work together, and need each other - to create great companies, great products, and great work.
1
304
We've seen AI work best in organizations that are truly adopting it
(not just using it, but reconstructing work around it)...
... where they use AI to extend the speed, capability, time, or quality of an ALREADY exceptional individual.
AI makes things faster, cheaper, etc... it doesn't produce out of thin air... it exaggerates its inputs.
1
229
A side effect of AI i've seen is Distance.
Some call it loneliness, but its not just that. It's velocity creating distance.
We used to stop and wait on each other.
Backend waited on frontend.
Frontend waited on product.
Product waited on customers.
Now that backend dev can simulate the rest, or build outside their own skills, and keep going.
Faster. In their own direction.
No check-ins needed.
So everyone moves fast, in different directions.
That creates distance.
Loneliness is just one part of it.
And it's not only software. I see it in finance, exec work, manufacturing.
6
6
1,754
Quoting is one of the first and best places to start using AI
1. historical can add as
- database for pricing, exceptions, methods, ranges
2. its easy to use templates and rate sheets
3. combine with SOPs and you have a powerful way to quote.
if nervous - just keep a human in the loop for review
We've done this for
- Window and door manufacturer
- remodeling contractor
- cnc shops
- foundry
- aerospace MRO supplier
- computer/server distributor
- and many more
The point is, it can prob be done in your industry with some tweaks and the right work on the data
I think I just saw the future of small business tonight.
I connected job costing across our roofing company. Budgets, bills, and QuickBooks are finally talking. I’m using @bot to write estimates and work orders. Duplicate work is disappearing, and the data is getting more reliable.
Then it hit me: this can extend across the entire business, from the first customer inquiry to the final payment. The same system could support the next location or acquisition.
@bot, @Muse, @claudeai, @dots, APIs, and MCPs are putting that within reach of small business owners.
The capital and headcount needed to build a large service company may be about to change dramatically.
Holy smokes!
3
21
4,640
having fund making this sourdough platform automate demand from
- subscriptions
- drops
- wholesalers (cafes, etc)
and auto funnel it all to a demand page to easily see what you need to bake
... in short, taking the best of manufacturing workflows and over-engineering the hell out of this.
2
9
1,860
(I mean, we technically can make a plan too given we know the recipe and process with times, but this is really over kill)
the one useful alert is what literally can't be made given the due dates and time we said we want to work:
1
4
733
If you are running a bakery out of your home or microbakery, give tryproof.io a go -
- set up drops/preorders
- track wholesalers
- track inventory, recipes, unit costs, etc
2
2
566
Great progress on our AI learning platform for homeschoolers.
1. Put in the knowledge graph with sub topics and end states for all subjects and skills to 12th grade
2. Plus built out the 6 week custom curriculum builder a lot more (creates lessons, videos, and a large project to learn anything the student wants (my son choose 3d printing and isotopes)).
3. Beefed up the lesson plans with visuals, photos, diagrams, and key word highlighters.
This is becoming a really powerful, gamified way for Hud to learn. He is collecting points, passing assessments, and the tutor helps by pre-loading knowledge and using socratic questions to help him learn new things he needs help with.
I do not believe this is "the future of AI" - these are useful tools to bridge... but the future of MFG and Supply Chain AI looks very different (imo)
Worked with an Aerospace Manufacturing to build a realtime Supply Chain and Procurement tool backed by agentic workloads
Here is how it works and a few of the features we built
(data is real, some/most names & parts are switched in this environment)
1
20
3,066


















