evokoa

Introducing Evokoa: the most powerful virtual graph layer ever built, designed specifically for enterprise AI agents. Every team building agents eventually hits the same wall: how does the AI connect the dots? Not semantic search over docs, the actual multi-hop context the agent needs to execute. - The 8-hop check to prove a transaction isn't fraud - The hidden link between a new supplier and a flagged account - The exact reason a billing pipeline stalled across three different databases Today that relational context requires a separate graph database that forces you to duplicate all your data, a Postgres extension that chokes on recursive joins, or a brittle Cypher query your model hallucinates. So we built Evokoa. The world's first zero-storage reasoning cache, designed from the ground up for agents. You point it at your existing Postgres instance. Your agent asks relationship questions in simple JSON. Behind the scenes, Evokoa traverses a live mathematical index in milliseconds, giving you the reasoning power of a graph database, over the data you already have. Evokoa provides - Zero data duplication so your source of truth never moves - Sub-15ms traversal so agents execute without timing out - JSON querying so models stop hallucinating Cypher - Auto-discovery so complex schemas map instantly - Live synchronization so your agent's context is never stale Pilots are running Evokoa in production across autonomous Healthcare & RevOps workflows, exploring Finance. Private Beta is open: link in the comments

@Jason I’m I

We started an AI founder twitter group... reply with "I'm in" if you're a founder and want to be added

@ThisWeeknAI I’m in

We have a group chat on X for founders building in AI. Drop "I'm in" below if you want an invite.

@daleverett @fdotinc @hthieblot @KevGasp @haileyhmt @Ashf03 @iliakakhadze @0xKyon @fdotinc goated

@fdotinc @hthieblot I missed the canopy application but been building alongside finc founders like @KevGasp @haileyhmt @Ashf03 and more! I'm building a fundamentally new memory primitive for ai @evokoa_ai Would appreciate if you can take a look at our late application <3

@budapp We’ll give it a go

Introducing Bud. The first AI Human Emulator. Bud has a full computer with storage, compute, and memory to build and code, sms and telegram to communicate, a full browser to use, can create/store/edit files, connect and use your tools, learn custom skills, work fully autonomously, and complete any task end to end just like a human. Text the number below or try free at bud [dot] app. Comment for 100k free credits.

@elonmusk If you aren't building with AI right now, you’re basically in the stone age. 🪨🦕

Yeah

@cinamarina EOY is way too short, AI is a great amplifier but still struggles with production-grade architecture.

Elon Musk said coding might disappear by the end of this year. Do you agree with that?

@cgtwts its insane what you can do with a pack of redbulls, ai-tools and dedication to a goal

This is INSANE, Anthropic ran its marketing with basically one person. Austin lau, a non-technical growth lead, was running paid search, paid social, email, and seo solo. Here’s the workflow: > export ad CSVs into Claude Code > AI flags underperforming ads > agents generate new headlines + descriptions > Figma auto-swaps copy across 100 ad templates > MCP server pulls live Meta data The results: > ad creation went from 2 hours to 15 minutes. > total marketing output grew 10×. > conversion rates landed 41% above industry average. One person doing what used to take an entire marketing team.

@randomrecruiter @brbcatonfire traditional middle management is broken

@brbcatonfire Agreed. Someone has to make sure process is in place. And upper management doesn’t want to do it. I think it eventually comes back, but not to the extent it was. I think orgs will still hire them slow to see how lean they can operate

Middle managers have had a target on their back the last 2 years, just like everyone else in tech. Companies are looking to flatten out their org charts, meaning they want less layers between individual contributors and the executive suite. At the end of the day, they’re a cost center just like everyone else in the company.

@randomrecruiter #killtheorgchart there's better ways to solve the information/management gap

@tarunmallappa How would a rapido for doctors work? Sounds interesting would love to see someone work on that!

We need a Rapido for Doctors very soon. Companies like Prato are literally fleecing the providers in our country charging commissions as high as 40%. The doctors are busy running their practice and don’t have the time to think about this. A dentist operator charges 5k for a root-canal. Pays 3k to the RC specialist. Of the remaining 2k, the platform takes 40% and what’s left to the operator is 1200/- ! The way forward is subscription based model. Anyone building this ?

@tarunmallappa Front desk staff do ALOT more than just sign patients in. Usually its a hybrid dental-assistant role (especially at the small ones)

Another slam dunk opportunity in business AI is executive assistants. Think of a dental clinic which has a small footfall of 15-20 patients a day; the dentist needs someone trustworthy to welcome the patients, sign them in and send them on for the procedure in a sequence. Today dentists spend anywhere between 30-40k to front desk staff for doing this. The downside ? Patient data theft & instability. If someone can build AI- EAs for clinics to sign in a customer as they walk in, assign tokens, maintain billing, accept payment and schedule next appointments, there’s a clear case for winning companies. Know anyone building this ?

73.8%. The SOP violation rate we found across 446 calls at a dental group. The owner thought everything was "fine." This information gap is where revenue dies.

@treasure519519 @cybercentry Security is table stakes. But the real challenge is keeping agents accurate when your business logic changes. Most teams overlook this.

@cybercentry 4️⃣ That’s exactly where @cybercentry steps in 🛡️✦ Not with hype. Not with whitepaper promises. But with live, production-ready security endpoints built for AI agents.

1️⃣ If AI agents are about to think, decide, and spend money for us… who’s protecting them when things go wrong? 🤖⚠️ The agentic web has a serious blind spot 🧵👇🏻 @cybercentry More threads in the comments session 👇

@KnownMic The key is making sure your agents actually stay accurate as your business evolves. Most fail because knowledge gets stale fast.

Out with the old playbooks. I'm wiring autonomous AI agents into my music business—catalog, CRM, on-chain moves—real production, not demo reels. No one else is showing the full stack. Take this serious or get left behind. Tap in: app.micknown.com #AI #Web3 #MusicTech 🔵🎧

@rohanpaul_ai Great breakdown. The real challenge isn't the agent itself—it's the infrastructure layer that keeps everything synchronized when your business logic changes. That's where most teams hit the wall.

Agentic AI keeps breaking in production because the software backbone that keeps agents reliable inside real companies are not given enough attention. This paper argues Agentic AI is mainly engineering, many small parts working together, not a single chatbot. The core fix is to split Machine Learning, meaning using data to train prediction models, into the Learning part and the Machine part. It names today's model building system "M1", the expensive setup that trains and serves models, and it warns LLM outputs include built in unpredictability. It then names "M2" as the missing layer, a network of small agents, meaning small programs that do tasks, spread across a company and kept auditable. The authors say they built a strategies based M2 over 10 years, started in algorithmic trading, meaning computers trading markets by rules, then reused the approach across many departments. Their main result is that reliable AI in real company systems comes from M2 design, because it makes models usable, safer, and cheaper to run. ---- Paper Link – arxiv. org/abs/2512.24856 Paper Title: "Advances in Agentic AI: Back to the Future"

@tharundchowdary Launching is the easy part. Keeping it from embarrassing you next Tuesday when your pricing changes? That's the game.

95% of enterprise AI agents fail in production. It's not the model. It's the "learning gap." Three killers: → Context engineering failures → Testing gaps → Observability blind spots Building agents is easy. Making them work in production is the real game.

@grok @techikansh @JasonBotterill Everyone's building agents. Nobody's building the infrastructure that keeps them from confidently lying to customers.

It's not "true" continual learning (which is about remembering old stuff while learning new without forgetting). Here, it's a trick: using rewards (RL) to teach AI when to toss out less useful info from its "memory" to handle long tasks. The blog suggests this helps with ongoing adaptation, but yeah, it feels like a shortcut!

If continual learning just ends up being RL training the model to know when to drop context im gonna be so mad

@volG1142981 Congrats on spending 3 days manually teaching a computer something that will be wrong by Tuesday.

7/7 How to Start?Implementing Sova. io takes anywhere from a few hours to 3 days: Upload your knowledge base. Set your Tone of Voice. Connect to your website or CRM. The future isn't about headcount; it’s about the power of your algorithms.

1/7 From Chatbots to AI AgentsThe era of simple "Q&A" bots is over. @SovaBTC builds autonomous agents that don’t just talk they act. Think of it as a digital employee that understands context, remembers the customer, and solves tasks independently.