Al & Tech Educator . Sharing insights & practical ways to use Al & Tech Tools for you & your daily business 💌 [email protected]
Dm for promo
Joined November 2023
- Tweets43.1K
- Following3.2K
- Followers7.8K
- Likes38.7K
Ace retweeted
Your company's data systems were built for people who run a few reports a day and go home at six. The AI agent it shipped last week is still hammering that data right now, and it wants the newest row plus every row before it, every single time.
PMs rolling out agents keep telling me the same thing: every new agent needed its own pipeline, so the infrastructure bill climbed with each launch and the engineers who were supposed to build features spent their weeks keeping plumbing alive. Spark and Flink, the two workhorses under most of that plumbing, came out of university labs around 2010, long before anyone pointed an agent at them.
Eon just published a number on how badly that fit is breaking. At 1 million events per second, its new engine, Eon Yukon, ran on roughly a quarter of Spark's compute and a seventh of Flink's at the same freshness, in a benchmark it put on a public repo so anyone can rerun it.
Two details under the headline matter more than the ratio. The data ends up as open Iceberg tables in your own storage, so the engine has to keep earning the workload on cost and freshness every month. And it builds a description and stats for each table while ingesting, which is what lets an agent ask what changed last week without reading the whole table.
Data infrastructure is about to get graded on freshness per dollar. Any stack that needs a new pipeline per agent is going to have a rough time on that report card.
🤝 Paid partnership
What I'm seeing across the industry right now is that the biggest bottleneck to AI adoption isn't model capabilities. It's data infrastructure.
Today, we're expanding the Eon Data Foundation with the launch of Eon Yukon, a new serverless data engine purposely built to change that.
AI agents need fresh, contextualized data to act.
Specialized models need a continuous feed of data and feedback to improve. Both are still served by a stack, an architecture, and methods built for dashboards:
-Spark and Flink clusters sized for peak, plus compaction jobs to keep tables usable.
-A pipeline and a copy for every consumer, staged bronze to silver to gold.
-Hand-written jobs, wired together by an orchestrator and kept alive by specialists.
The result: shipping data is complicated, risky, and expensive. While coding agents ship apps in hours, they rarely build or run data pipelines. The data is too volatile, and a mistake costs too much.
With Yukon, data moves from ingestion to processing to serving in one continuous flow, without the infrastructure and pipelines traditionally required to connect them.
Events become compacted, sorted, deduplicated Iceberg tables within seconds.
Real-time aggregations run inline, with fully managed state. Agents and applications can access fresh data with sub-millisecond latency through a Redis-compatible API or through our SQL and spark compatible query engine.
Beyond moving and serving data, Yukon makes it continuously understandable - Live column-level statistics, accessible through MCP, give agents visibility into the data. Built-in monitoring, sampling, and data minimization help teams maintain control.
At 1M events/sec and equal freshness, Yukon used 4x less compute than Spark and 7x less than Flink. The benchmark is public: run it yourself with Eon’s ingest-bench.
This is another major step toward our vision of an Autonomous Data Foundation for the AI era, built to both protect enterprise data and unlock its full potential.
Ace retweeted
This is a desktop agent that connects video generation models with creative tools such as Blender, DaVinci Resolve, and Premiere Pro.
The idea is simple:
Tell it what you want, and the agent will generate footage, assemble it, and edit it using pro tools.
Ace retweeted
A lot of the AI video on your feed took 10, 20, sometimes 50 generations to get one usable shot, and the person who posted it paid full price for every miss. Nobody fighting the AI video price war is counting that number. One team just went after it.
Seedance 2.5 on Voyager runs at half of Higgsfield's price, and the more interesting move sits in front of the model. The desktop agent does what a film crew does before the camera rolls. It storyboards, blocks the shot out in Blender, previews it as a still or a low-res draft, and only calls the video model once you've signed off on what it's about to ask for. Their users are already making 3x fewer API calls, and the whole thing works out to about 5x cheaper than Higgsfield for the same finished video.
I've watched this exact movie play out in coding. Tokens got cheap for everybody within the same year, and the tools people kept using were the ones with a real harness around the model, the kind that plans the change, runs the tests, and checks its own work before you see it. Video is a year or two behind that curve, and @startvoyager is building the harness for it.
Per-generation prices will keep falling for everyone, so the lasting edge goes to whoever needs the fewest generations.
🤝 Paid partnership
Ace retweeted
You've explained your job to ChatGPT more times than to your own mother, and every new chat starts you from zero. One of Twitter's early international hires just built a memory that follows you across AI apps, and her team says it gets 92.2% on long-conversation questions while using up to 22.6x fewer tokens.
An app that holds your memory has a pretty good reason to keep you around, which is why ChatGPT, Claude and the rest each ship memory inside their own walls. The benchmark she's quoting is LoCoMo, which asks a system questions about conversations that ran for weeks. When Mem0 published its paper, the full-context baseline, where you dump the whole history into the model, came in around 73%. Setups differ, so treat the 92.2% as their run. Still, anyone who's pasted a 100K-token thread into a model and watched it get dumber knows more context stops helping at some point, and you pay per token for it.
Living Brain scores each memory on whether it's worth keeping, where it belongs, what it connects to, and whether it matters right now. The team says each brain also tracks freshness, readiness and strain, so a stale fact gets flagged before it's reused. A memory that only ever adds is a bigger context window with extra steps.
Write something in ChatGPT, then read it back in Claude or hand it to a Cursor or OpenClaw agent. Once that works, you own the memory and can swap the model whenever you feel like it. By their count the team has pushed 100B+ tokens through OpenAI's models, so they built this to fix their own context tax, and it starts at $5.
Models get swapped out every six months now, so the thing to hold onto is the record of how you work. @ladyxtel @livingbrain
🤝 Paid partnership
Ace retweeted
This is a shared memory service where you can connect multiple agents and AI tools to reuse knowledge.
For example, you can save information using Codex and retrieve it through Claude Code.
Two scenarios where this is very helpful:
• To keep a portable knowledge base not tied to any agent.
• To have the flexibility of using multiple agents without duplicating the knowledge.
Ace retweeted
Five weeks a year. That's how much time the average freelancer spends on work nobody pays for, like proposals, invoices and the "just checking in" email. It comes to 204 hours, per a Smallpdf survey of 397 freelancers this year, and 70% of them already use AI for that stuff. Nearly half say it has done little to cut it.
I think I know why. A chatbot can write you a scope email in ten seconds, but it has no clue what you actually agreed to back in March, because that deal lives in a Gmail thread, a Notion page and a PDF the client never opened. So when "can we also add a video?" shows up at 11:47pm the night before launch, you end up being the only database in the business, and you're tired.
The fix seems obvious once you say it out loud: the proposal should be the record. MonoDesk builds the project board and the client portal straight off the scope you priced, so the agreement stops living in three places. In their demo a late request comes in, the agent checks it against that scope, flags it, and drafts the pushback with a phase 2 quote attached. If you've ever put off that exact email for days, you get why watching it drafted in seconds got my attention.
Their launch video tracks the whole thing in heart rate. A designer hits 152 BPM when the scope creep lands and drops to 62 once the admin is off her plate, and anyone who has done client work knows which part of the week that spike comes from.
Five weeks is a lot of time to hand back to the part you actually like.
🤝 Paid partnership
Ace retweeted
You've yelled "representative" at a phone three times, a little louder each time, and heard "I can help with that!" three times back. On that company's QA dashboard, the call scored green. The bot was reading a transcript, and a transcript can't hear you losing it.
The same blind spot is everywhere. You say "I'm fine" in the tone that means the opposite and the bot takes you at your word. A hesitant "...yes" to a payment confirmation gets booked as a confident one, and if you pause to think, the bot decides you're done and talks right over you.
Speech-to-text can transcribe almost anything now and synthetic voices sound strikingly human, yet the agent sitting in between still runs on text. A furious caller and a calm one can look identical on the page. I think nobody fixed this because the tooling can only read the transcript, so teams optimize the transcript, and the angry customers who hang up live in the part nobody kept.
@hume_ai just put out an API that goes after exactly this. You ask it a plain-English question about the audio ("does this caller sound unsure?") and get back a probability per answer. It's audio-native, so it listens to the recording itself and nobody has to build a separate classifier for each use case. Hume says it's trained on millions of human judgments, and in their initial evals it matched a panel of human listeners more closely than a single human rater did, which would let a lab score every model checkpoint instead of waiting on a panel.
Voice agents have spent years getting graded on the transcript, which is the one file that throws away how the caller actually sounded.
🤝 Paid partnership
Introducing the Hume Audio Decision API.
Ask anything: How many speakers? Is anyone laughing? Is the caller frustrated?
What used to take 10+ models is now one call.
Trained on 10M+ human judgments, so it answers the way human listeners would.
Realtime, or across hours of recordings.
Request access below.
Ace retweeted
This is a new decision model for audio input:
• You provide the audio
• You provide a list of questions
• Model outputs answers + confidence scores
The model analyzes the audio directly (no transcript needed).
Something really cool is the model's ability to answer questions based on tone, hesitation, laughter, and other audio clues that wouldn't be in a transcript.
Introducing the Hume Audio Decision API.
Ask anything: How many speakers? Is anyone laughing? Is the caller frustrated?
What used to take 10+ models is now one call.
Trained on 10M+ human judgments, so it answers the way human listeners would.
Realtime, or across hours of recordings.
Request access below.
Ace retweeted
روٹی غریب کی پہنچ سےدور۔۔۔حکومت کا گندم کی قیمت 4500 روپے فی 40 کلوگرام مقرر کرنے کا اعلان
کراچی میں لوٹ مار کابازارگرم۔۔۔تمام بڑے موبائل فون برینڈز کے ہینڈ سیٹس 15 ہزار سے 30 ہزار روپے تک مہنگے ہوگئے
گورنرراج کےبعد سےابتک وزیراعلی سہیل آفریدی لاپتہ،سی ایم آخر ہیں کہاجانئےمقابل پروگرام میں 👇👇👇
youtube.com/live/A6ZQOpoh5mc…
@AmirMateen2 @Rabiya_Ahsan @Rana_Atif101 @MehmoodJan1 #SohailAfridi #pti #pmln #PPP #kp #Emergency
Ace retweeted
President Asif Ali Zardari inspects the Guard of Honour along with German President Frank-Walter Steinmeier during the Official Welcome Ceremony at Germany.
Ace retweeted
President @AAliZardari stresses on the imperative of realising available potential for expanding existing political, economic and defence cooperation between Pakistan and Germany, in a meeting with Federal President of Germany Frank-Walter Steinmeier in Berlin today - with greater focus on economy and trade, emerging technologies, climate resilience, skills development, people-to-people and cultural exchanges.
Ace retweeted
President of Pakistan Asif Ali Zardari receives a Guard of Honour in Berlin, Germany, during his official visit at the invitation of German President Frank-Walter Steinmeier. Sindh Chief Minister Syed Murad Ali Shah, Senior Minister Sharjeel Inam Memon, and Provincial Minister Jam Khan Shoro are also accompanying the President. The visit aims to strengthen Pakistan-Germany relations and enhance cooperation in trade, investment, and other key sectors.
Ace retweeted
مادرِ جمہوریت بیگم نصرت بھٹو آمریت کے خلاف سیسہ پلائی دیوار ثابت ہوئیں۔ ان کی بے مثال قربانیاں، جمہوری جدوجہد اور لازوال استقامت پاکستان کی جمہوری تاریخ کا قیمتی اثاثہ ہیں، جنہیں ہمیشہ یاد رکھا جائے گا۔ ان کی برسی کے موقع پر پوری قوم مادرِ جمہوریت کو سلامِ عقیدت پیش کرتی ہے۔
@BBhuttoZardari
Ace retweeted
President Asif Ali Zardari pays tribute to Madar-e-Jamhooriat Begum Nusrat Bhutto, remembering her courage, immense personal sacrifices and unwavering commitment to democracy, women’s empowerment and the people of Pakistan on her 15th death anniversary.
Ace retweeted
Chairman @BBhuttoZardari pays tribute to Madar-e-Jamhooriat Begum Nusrat Bhutto on her 15th death anniversary, remembering her unwavering struggle against dictatorship, her sacrifices for democracy and her enduring commitment to constitutional supremacy and the rights of the people of Pakistan.
Ace retweeted
Honouring Madar-e-Jamhooriat Begum Nusrat Bhutto, whose courage, sacrifices and unwavering commitment to democracy continue to inspire generations of women and uphold the spirit of democratic resistance.
First Lady of Pakistan Bibi Asifa Bhutto Zardari on 15th Death Anniversary of Begum Nusrat Bhutto
@AseefaBZ
#MadarEJamhuriyat #BegumNusratBhutto
Ace retweeted
قومی اسمبلی کے اجلاس میں وقفۂ سوالات کے دوران خاتونِ اول و رکنِ قومی اسمبلی بی بی آصفہ بھٹو زرداری کا نوابشاہ سمیت ملک بھر میں 4G اور 5G انٹرنیٹ سروسز کی سست رفتاری کا معاملہ اٹھاتے ہوئے انٹرنیٹ سروسز کی بہتری اور مسائل کے حل کے لیے مشترکہ لائحہ عمل سے متعلق سوال
@AseefaBZ
Ace retweeted
مادرِ جمہوریت، بیگم نصرت بھٹو، پاکستان کی سیاسی تاریخ کی وہ باوقار شخصیت ہیں جنہوں نے جمہوریت کی بحالی، عوام کے بنیادی حقوق اور آئین کی سربلندی کے لیے بے مثال قربانیوں اور غیرمتزلزل عزم کے ساتھ جدوجہد کی۔
ذاتی صدمات اور کٹھن ترین حالات بھی آپ کے حوصلے کو متزلزل نہ کر سکے۔ آپ کی جرأت، استقامت اور جمہوریت کے لیے لازوال جدوجہد تاریخِ پاکستان کا ایک ناقابلِ فراموش باب ہے۔
@BBhuttoZardari
#MadarEJamhuriyat #BegumNusratBhutto
Ace retweeted
President @AAliZardari visited the Saudi Pakistan Real Estate Expo (SPREX) at Lake City, Lahore, underscoring the importance of strengthening economic cooperation and promoting investment between Pakistan and Saudi Arabia.
Ace retweeted
نیشنل انسٹیٹیوٹ آف چائلڈ ہیلتھ (NICH) میں سندھ حکومت کی جانب سے بچوں کو معیاری طبی سہولیات، جدید علاج اور بہترین طبی نگہداشت مکمل طور پر مفت فراہم کی جا رہی ہے، تاکہ کوئی بھی بچہ مالی مشکلات کے باعث علاج سے محروم نہ رہے۔
کیونکہ بچوں کی صحت پر سرمایہ کاری، ایک مضبوط قوم اور روشن مستقبل کی بنیاد ہے۔ ہر بچے کی زندگی قیمتی ہے، اور ہر بچے کو صحت مند زندگی جینے کا حق حاصل ہے۔
@BBhuttoZardari