Abhishek Sisodia (@abhirb) header
Abhishek Sisodia (@abhirb) profile picture

🇨🇦 Principal AI Engineer | Founder @ Circuit Studio 🚀 | Building AI agents, shipping products & sharing what actually works beyond the hype

Toronto, Ontario
Joined January 2010
Photo by Abhishek Sisodia (@abhirb)Photo by Abhishek Sisodia (@abhirb)Photo by Abhishek Sisodia (@abhirb)Photo by Abhishek Sisodia (@abhirb)Photo by Abhishek Sisodia (@abhirb)Photo by Abhishek Sisodia (@abhirb)Photo by Abhishek Sisodia (@abhirb)Photo by Abhishek Sisodia (@abhirb)
Replying to @MicroLaunchHQ
Every business is rushing to adopt AI. Almost nobody knows what's actually happening under the hood. We're building something to change that. Circuit Signal 👀 circuitstudio.ca/command-cen…
1
83
AI pricing has a trust problem. If an agent gets stuck, retries 17 times, and finally completes a simple task… should the customer pay for all that effort?
Replying to @DogecoinNorway
The cost of serving Grok Bot was unsustainably high relative to revenue
1
1
53
Abhishek Sisodia retweeted
I encourage anyone who is leading an engineering org to read this in its entirety. It is short, extremely well written, and very much directionally correct. The job and the organization needs to entirely change to take full advantage of these tools. That path is largely a human alignment challenge at this point. If you are a leader, your job is to get people on the new ship, and get them sailing together. The old ship doesn’t exist anymore, although many people wish it did and are incentivized to continue to believe it exists. Software Engineering is an entirely different thing now, and the faster you can get your org to understand this and align on how to work now, the less painful it will be for you long term. If you don’t do it, I simply believe your org won’t exist because it will get disrupted by an org that does
Replying to @josephfcox
Here is Proton CTO's AI manifesto, that I obtained multiple copies of 404media.co/manifestos-fear-…
43
202
15
2,344
368,547
How many voice AI startups just got put on notice? 👀 Jack Dorsey's Block just unveiled Dessa. And this is bigger than another realistic AI voice. The old voice-agent stack: → Speech-to-text → LLM for reasoning → Text-to-speech → Extra logic for interruptions, timing and emotion Now we're moving toward a single model that can listen, speak and react in real time. ElevenLabs launched Eleven Realtime just 4 days ago. OpenAI has GPT-Live. Google has Gemini Live. And now Dessa takes a fascinating approach: trained entirely on human conversations, without a text LLM. Apparently, it can even run locally on a MacBook Pro. So what happens to startups whose entire product is stitching 3–4 voice APIs together? I don't think voice AI is dying. I think voice AI infrastructure is becoming a feature. The real value moves to workflows, integrations, distribution and trust. Still a research preview, but wow. This space is moving FAST.
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence. block.xyz/dessa
1
3
146
Companies spent the last 2 years encouraging employees to use MORE AI. Now they're realizing they don't know: - Which AI tools are being used - What company data is being shared - Who approved access - Who's accountable when things go wrong And AI agents are about to make this 10x more complicated. The next big enterprise AI challenge isn't adoption. It's knowing what AI is actually running inside your organization. You can't govern what you can't see.
‼️ Sweden's Government Offices will block ChatGPT and Claude on staff computers from Monday, Sveriges Radio reports. Officials are to use approved tools like Microsoft Copilot.
1
2
44
Most businesses don't even have a clear picture of which AI tools they're using, who's responsible for them, or what they're spending. I am building a lightweight way to bring that visibility together. Because you can't manage what you can't see. Take a look 👇 circuitstudio.ca/command-cen…
1
15
Railway's security incident should be a wake-up call. Users are being told to rotate ALL their secrets. But here's what nobody is talking about, Why are we still paying a premium to avoid AWS/GCP complexity when AI agents can now do much of the heavy lifting? You can literally give Claude Code, Codex or Cursor official cloud skills to help: → Set up infrastructure → Configure deployments → Manage IAM and secrets → Set up monitoring and logging → Build CI/CD pipelines Two GitHub repos worth bookmarking: AWS Agent Toolkit github.com/aws/agent-toolkit… Google Cloud Skills github.com/google/skills Both are official. Of course, you still need to review permissions, security and costs. AI doesn't magically make infrastructure secure. Railway made deployment easy. AI agents are making cloud infrastructure easier. And that convenience premium is getting harder to justify. Maybe it's time to reconsider where you deploy.
Spent multiple hours now evaluating Railway alternatives to move my shit to. Conclusion is that Railway's product is uniquely good and nothing comes close to a drop in replacement 🙃 Minimum cost to replicate my setup today is 3x higher than I am paying them currently
1
94
Whats this glitch showing up on my @X timeline?
1
33
Abhishek Sisodia retweeted
Replying to @MicroLaunchHQ
Every business is rushing to adopt AI. Almost nobody knows what's actually happening under the hood. We're building something to change that. Circuit Signal 👀 circuitstudio.ca/command-cen…
1
83
Abhishek Sisodia retweeted
A plumber gave his AI agent access to his business accounts. It saved him 5 hours of admin work. Then it cost him $3,200. Here's a scenario I worry we're going to see a LOT more of. He runs a small plumbing business. Hates paperwork. So he sets up an AI agent to handle emails, invoices, and customer follow-ups. Connects his Gmail. Accounting software. Payment accounts. Tells it to take care of things. Life is good. Until a fake supplier invoice lands in his inbox. The agent reads it, follows the instructions to update payment details, and pays $3,200 to the wrong account. No sophisticated hack. Just a very helpful AI assistant with way too much access. With everyone rushing to try Grok Bot, Muse, OpenClaw, etc., here's a basic security checklist for your agents: 1. Create a SEPARATE email for your agent. Don't hand over your personal Gmail. 2. Give it its OWN browser profile. No saved personal passwords or banking sessions. 3. Give it access to ONLY what it needs. Read-only wherever possible. 4. NEVER let it move money without your approval. 5. Turn on alerts. Check what it's doing. Revoke access you no longer need. You wouldn't hire someone on Monday and hand them your bank passwords, house keys, and entire customer database. Yet that's exactly how some people are setting up their agents. I'm a huge believer in autonomous agents. But autonomy without boundaries is a disaster waiting to happen. Give your agent a job. Not your identity.
1
2
29
Abhishek Sisodia retweeted
Everyone says AI code reviews are the new bottleneck. Brother, you're a solo founder. AI wrote the feature. AI reviewed the feature. Another AI reviewed the review. You've built an engineering approval committee for an app with 6 users. Maybe the bottleneck is you. 💀?
2
1
5
36
Maybe people are just tired 😂 Every other week there's a new smartest model and new reasons to switch. But at some point, good enough and works reliably starts beating another 10 points of intelligence. Model fatigue is real.
OpenAI, Anthropic and other leading AI labs face a growing challenge: Better models aren't necessarily attracting more usage. A 10-point intelligence advantage now earns just 20% more developer traffic, down from nearly 69% in late 2025. Full story: thein.fo/4hGJ9jY
1
35
Abhishek Sisodia retweeted
SAVE THIS: Anthropic, OpenAI, Cloudflare and Google have all released AI tools that hunt security bugs in your code. Here's each one and how to run it. 1. Claude Security by Anthropic A team of agents maps your app, builds a threat model, hunts through every component and checks each finding independently. It can turn confirmed bugs into patch files. /plugin install claude-security@claude-plugins-official Then type /claude-security 2. Codex Security by OpenAI (11.1K stars) Scans a whole repo, one folder or just your latest changes, validates what it finds, writes patches and drafts a SECURITY.md. npm install --global @openai/codex-security cs login cs scan . 3. Security Audit Skill by Cloudflare (31K stars) Six phases, and every finding is rechecked by a fresh agent that didn't find it. Works in most coding agents. npx skills add github.com/cloudflare/securi… --skill security-audit Then ask: "security audit this codebase" 4. Mantis by Google (2.5K stars) Builds a threat model, tries to reproduce each bug, then patches it and attacks the fix to prove it holds. Google says to run it only in an isolated environment. github.com/google/mantis Run at least one before you ship anything that handles user data.
49
104
14
800
98,045
Abhishek Sisodia retweeted
Turns out the future does have flying cars
47,593
99,061
32,815
829,374
80,422,835
Abhishek Sisodia retweeted
try pstack! it’s my personal set of skills for rigorous and autonomous engineering with agents. it’s updated frequently as I continue to tune and simplify them 10 second tutorial: 1. run /setup-pstack and choose your budget 2. add /poteto-mode in front of any prompt you want done rigorously 3. ask /poteto-help whenever you’re unsure what pstack skills to use 4. that’s it! Grok Bot: x.ai/bot/plugin/9717366 Cursor: cursor.com/marketplace/curso… Source: github.com/cursor/plugins/tr…
pstack is the new solid
253
255
68
4,935
336,589
Abhishek Sisodia retweeted
i've been bashing on "software is dead" for a while now. it's probably time for me to start sharing what i think will happen instead as well there's so much to unpack, so i'm going to split it into a series of a few parts today i'll start with part 1 - jevons paradox story time - let's rewind to the industrial revolution, when steam engines powered all the factories, railways and effectively the economy but steam engines had to burn coal - a lot of them. there's a limited amount of coal available, so people started to optimize the steam engines to use less coal that's when the counter-intuitive thing happened - when steam engines used less coal, the industry actually ended up using more coal in total how?? what jevons found and documented in his book was - when steam engines used less coal, they became cheaper to deploy and run, and hence became economically viable in more places, so the total demand rose significantly as a result when the barrier of entry to create software is significantly reduced, the same will happen - software will become much cheaper and more accessible hesitant to pay $100/month for the adobe suite today? well, what if it becomes $10 lifetime purchase? sick of so many mobile apps having a paywall? well, what if the entire app store is $5/month all-you-can-eat? the price point for each individual piece of software will drop, but the software economy as a whole will become a lot larger because of the rising consumption demand so many more people and businesses will get access to software that's too costly for them to use today so many more people who aren't developing software today will be able to participate, create, help grow the economy and make a living out of it software is nowhere near "dead". it's just about to take off
88
48
15
717
63,516
Abhishek Sisodia retweeted
While sandboxes are a commodity, serving millions of them under production-grade traffic requires a lot of detail to scale agents for businesses and consumers. We're open-sourcing the sandbox platform SPACE that powers all of Perplexity Computer to support the ecosystem. github.com/perplexityai/spac…
118
123
32
1,319
90,614
Abhishek Sisodia retweeted
Yesterday I went through the replies under India's Olympic bid. Indians swimming in garbage. Cows on the running track. Men squatting on a race course. Most of it made by AI. I wanted to know how this got so easy. The story starts in October 2022. That month, Elon Musk bought Twitter. Within weeks, thousands of staff were gone. X later told Australia's online safety regulator it had cut 80% of its trust and safety engineers. The people whose job was to catch hate. Researchers at UC Berkeley tracked what came next. Hate speech on the platform ran about 50% higher in the months after the takeover. In January 2024, about 12,900 hate posts against South Asians in the US went up on X in a single month. That number comes from a report released two weeks ago by the Center for the Study of Organized Hate in Washington. By the end of that year, X's own AI, Grok, could make any image in seconds. The report says AI is now mass producing cartoons of South Asians as dirty, as scammers, as vermin. By June 2026, the monthly count had reached 43,800. Up 240%. The fastest growing kind called us dirty, up 171% in eighteen months. Almost half called us invaders or job thieves, with H1B at the centre. And it does not stay online. In July, Syed Sohail Uddin from Hyderabad was working at a mall kiosk in Utah. A man asked where he was from, then asked if he was Muslim. Then he stabbed him 15 times. In September, a Sikh truck driver was stabbed at a Wyoming rest stop. Police have not named a motive. Sikh leaders say hate against Sikh truckers had been building online for months. On October 2, someone threw bricks at a Swaminarayan temple in Ohio while priests and devotees were inside. It was the start of Hindu Heritage Month, days before Navratri. Muslim. Sikh. Hindu. The hate does not ask which. It only sees an Indian. Then came this month. Indians asked Musk simple questions about Starlink, and he called them Ambani's army of shills. The same week, the IOC confirmed India's 2036 bid. The replies filled up with AI cartoons. One had 198,000 likes. India has problems. We know them, we fix them, and we talk about them louder than anyone. America counted more than 7.7 lakh homeless people on a single night in 2024, the most ever. More than 1 lakh Americans died of drug overdoses in 2023. Indians don't fill American posts with AI cartoons of addicts in San Francisco. If we did, the world would call it racism. And it would be right. This is not casual racism. It is organised. The report found accounts doxxing South Asians over H1B, influencers filming confrontations at their homes and offices, and AI cartoons pumped out faster than any human could draw. And it runs on a platform that fired the people meant to stop it, built the tool that draws it, and carries it to millions. When the owner himself calls Indians shills, everyone below him gets the message. I can't tell you what is in Elon Musk's head. I can only tell you what his platform allows. He is fighting in public to sell Starlink to 140 crore Indians. He should look at what his own platform is selling about them.
120
318
19
1,551
63,957
Abhishek Sisodia retweeted
okay, honest opinions on this? Grok @bot recorded almost the entire product demo by itself. It handled the script, VO and pacing cleanup. It tried @ElevenLabs for the VO, but switched to @higgsfield as it decided that actually sounded better here. I’m actually pretty surprised by how little manual work this took. Be honest what still feels off? Does this feel good enough to ship as a real product demo?
What’s the secret to making a genuinely good AI product demo with VO? Asked @bot to record the product, then used @ElevenLabs for the voiceover. In theory sounds easy and it did provide a recorded one quickly. But in practice, the result just isn’t there. Timing, pacing, syncing the VO to what’s happening on screen… something still feels off. Anyone cracked a good workflow for this? How do you get this done entirely with AI agents?
1
1
34
how I imagine my Grok @bot after I give it 14 tasks and say just handle it
2
54
Anyone made real money yet purely by something setup by @bot @muse etc?
1
32