marcosomma
@SommaWork
Followers
2
Following
29
Media
34
Statuses
60
Founder @OrKaCore | Society-of-Mind believer | No black-box hype. 🔗 https://t.co/sNozcE63FW
Barcelona, Cataluña, ESP
Joined September 2022
18 primitives. 5 molecules. Infinite workflows. orchestration is the future of AI... https://t.co/i1194mA0TM
0
0
0
Congrats to our top 7 authors this week! 🏆 Sylwia Laskowska, David Duymelinck, Annu, @Chowdurhead, Aaron Rose, Shir Meir Lador, and @SommaWork. This week's lineup features projects like an AI-powered MS Paint clone, a guide to fine-tuning MedGemma, and much more.
dev.to
Welcome to this week's Top 7, where the DEV editorial team handpicks their favorite posts from the...
0
3
12
OrKa v0.9.8: local first reasoning stack with UI now starts via a single orka-start. OrKa reasoning repo: https://t.co/CLWJeTm1We Graph Scout Tutorial: https://t.co/qAZksSnwqM
0
0
0
# OrKa 0.9.4 thread 1) OrKa 0.9.4 is out. Focus: correct execution and consistent logs. Final agent is always logged, timestamps use ISO strings. GraphScout multi hop paths run end to end. - https://t.co/CLWJeTm1We - https://t.co/TIyVNbyihm -
pypi.org
Modular agent orchestrator for reasoning pipelines
0
0
0
🧠 New on @huggingface: OrKa cognitive AI framework ✅ test benchmark, with hight accuracy ✅ Society of Mind architecture ✅ 95.6% cost savings vs cloud APIs ✅ Full transparency + paper Try it live! 🚀 https://t.co/BFV2bPJPTS
#AI #CognitiveAI #OpenSource
huggingface.co
0
0
2
🧠 OrKa v0.9.3 dropped! New: GraphScout Agent (beta) for intelligent Graph exploration Moving beyond prompt engineering to real AI orchestration. pip install orka-reasoning 🔗 https://t.co/k688v1ayls ⭐ https://t.co/CLWJeTlu6G
#AIEngineering #LLM #OpenSource
github.com
Orchestrator Kit for Agentic Reasoning - OrKa is a modular AI orchestration system that transforms Large Language Models (LLMs) into composable agents capable of reasoning, fact-checking, and const...
0
0
1
🚀 OrKa-UI news 😀 Now fully aligned with v0.9.2 of OrKa-reasoning https://t.co/VEzWVsp8C7
#AI #ArtificialIntelligence #OrKa #MultiAgentSystems #CognitiveArchitecture #ExplainableAI #AIReasoning #SocietyOfMind #AIWorkflow #AgenticAI #OpenSourceAI
0
0
0
Most AI today runs on loose tricks hacks that look smart but break under pressure. I wrote about the Loop of Truth: a way to anchor AI in stable invariants so it can reason with structure, not just mimic us. @SommaWork #DEVCommunity
https://t.co/KTyDboplOX
0
0
0
Explore how GPT-5 and other AI stack up at coding versus creative writing with Ashley Childress. #AI #ChatGPT #GithubCopilot Read the full post: https://t.co/exOmiDXNsX
2
2
7
🚀 With OrKa-Reasoning I don’t just run agents, I stage societies of geniuses. Da Vinci, Einstein, Pythagoras, Newton & Tesla debated and converged on OrKa as emergent computational cognition 🧠✨ #AI #OrKa #MultiAgent #CollectiveIntelligence
1
0
0
🔥 Orchestration > the next big model. Build an orchestration layer that composes small, explainable modules for reliable, observable cognition. Map modular vs monolith, add traces & memory, pilot one workflow. https://t.co/VEzWVsp8C7
#AI #OrKaReasoning #Observability
0
0
0
Congratulations to our top authors of the week! 🥳 Amazing work from @skeptrune, @brianmtreese, @SivaramPg, @techgirl1908, @SommaWork. And a big shout-out to Ashley Childress and Pavel for their fantastic posts as well! #ai #machinelearning #githubcopilot #javascript #webdev
dev.to
Welcome to this week's Top 7, where the DEV editorial team handpicks their favorite posts from the...
0
2
8
🧩 Ditch AI monoliths: service nodes = scalable, debuggable cognition. OrKa sees cognition as orchestration—orchestrator routes, nodes w/ clear contracts & traces. Prototype memory/verifier/tool nodes+tracing. https://t.co/VEzWVsp8C7
#AI #OrKaReasoning #Modularity
0
0
0
🧠 Master the art of forgetting: memory decay > hoarding. Controlled forgetting reduces noise, boosts freshness, and makes AI explainable & reproducible. OrKa treats memory as pluggable, observable modules. Start: https://t.co/VEzWVsp8C7
#AI #OrKaReasoning #ML #Innovation
0
0
0
🧠 Master the art of forgetting: memory decay keeps AI relevant—prune stale facts, highlight novelty & improve decisions. OrKa makes memory a pluggable module with clear eviction, explainability & observability. https://t.co/VEzWVsp8C7
#AI #OrKaReasoning #ML
0
0
0
🚀 Big thanks to @devteam & @ThePracticalDev for OrKa’s Top 7 feature! 🎉 OrKa brings Hinton’s AI safety vision to life with YAML workflows. Join us: https://t.co/CLWJeTm1We See: https://t.co/wxm4VQq9sH Ideas? DM me! #AI #Safety #DevCommunity
dev.to
Welcome to this week's Top 7, where the DEV editorial team handpicks their favorite posts from the...
1
0
1
🚀 Stop chasing raw accuracy—design modular intelligence. Make reasoners, memory & orchestrator observable, swappable, composable. Audit integration points, instrument traces, swap 1 module & measure system gains. https://t.co/VEzWVsp8C7
#AI #OrKaReasoning #ModularAI
0
0
0
Tired of all the hype? Mak Sò's level-headed take might just be what you need to hear. Read here: https://t.co/5qb44oug1b
0
5
18
🚀 Rethink AI: parallelize, don't just optimize. Sequential pipelines = latency + wasted idle cycles. Fork/join runs parallel, traceable modules—lower latency, higher throughput, auditable merges. OrKa adds fork/join primitives. #AI #OrKaReasoning #ForkJoin #ModularAI
0
0
0