Abhishek Sisodia

🇨🇦 Building AI products & workflows | Founder @ ⚡️
| Author of Abhi’s AI Playbook 📬 Practical AI for business & builders

[Abhishek Sisodia @abhirb]

🧵 Everyone’s building RAG prototypes…
But what does it really take to go from demo to production-ready AI systems?

After scaling GenAI apps across business use cases, I’ve learned some hard truths.

Here are 10 lessons for building RAG systems that actually deliver ROI:👇

[Wise @trikcode]

You basically need to be unemployed to keep up with all this AI stuff.

[Abhishek Sisodia @abhirb]

@yfcherries Cheapest is in thousands not hundreds

[Abhishek Sisodia @abhirb]

Let's go $HOOD time for some disruption in Canada!

[Vlad Tenev @vladtenev]

Robinhood has officially arrived in Canada. 🇨🇦

We’ve closed our acquisition of WonderFi, marking our entry into Canada through one of the most well-respected crypto platforms in the country.

With 1 million international funded customers, Robinhood’s mission is going global.

[Abhishek Sisodia @abhirb]

@vladtenev Congrats and welcome! Canadian market looks easy but will be tough to challenge the banks (protected by govt and regulations) and existing Wealthsimple offerings.

[Abhishek Sisodia @abhirb]

You only need @SpotHero

[Abhishek Sisodia @abhirb]

@gregisenberg the harness around the models is becoming more important

[Abhishek Sisodia @abhirb]

@internetvin tesla should add a vibe mode profile in fsd that does exactly this and over time it learns your favorite roads, views, neighborhoods, and drive lengths.

[Abhishek Sisodia @abhirb]

@suekhim Great concept! How will it enforce kids attention the way a real human tutor would?

[Abhishek Sisodia @abhirb]

@Tesla We need another one with wet weather conditions

[Abhishek Sisodia @abhirb]

@DriveTeslaca @Tesla @DavidMoss @DevinOlsenn @scotsrule08 Amazing ! Would love to do it myself sometime. Did you charge overnight stays or mainly supercharger network?

[Abhishek Sisodia @abhirb]

@lucyhargreaves4 Preach.
We should be making it easier to invest, build, hire, and compete.
Canada has all the ingredients. We just need to stop putting roadblocks in front of the people trying to create value.

[Akshay 🚀 @akshay_pachaar]

Hermes agent masterclass.

In this video, I cover everything you need to understand and customize Hermes Agent. Self-evolving skills, three-tier memory, GEPA optimization, and going from 1 to 10 agents that work for you 24/7.

Enjoy!

Chapters:

00:00 - Intro
02:03 - How to get the most out of this video
02:32 - What we're building (and why it's wild)
07:11 - How the whole thing works under the hood
09:27 - The SOUL.md: your agent's personality file
11:15 - The 3-tier memory system that keeps it all together
14:16 - Skills: what your agent can actually do
16:49 - The self-evolving loop (agents that improve themselves)
19:58 - The curator: Hermes' built-in garbage collector
22:56 - GEPA optimization: making your agent sharper
25:08 - Installation and setup
27:38 - Connecting your agent to Telegram
30:36 - Configuring programmer with Claude Code
31:53 - Adding new skills (from a hub of ready-made skills)
34:59 - Going from 1 to 10 agent profiles
36:49 - Building a custom designer from scratch
40:42 - Anatomy of the .hermes folder (where everything lives)
45:05 - Skill taps: sharing skills via a GitHub repo
45:59 - Skill bundles: stacking skills for workflows
47:19 - Hermes Kanban (coming soon)
48:05 Outro

Cheers! :)

[Akshay 🚀 @akshay_pachaar]

the anatomy of the perfect 𝗦𝗢𝗨𝗟.𝗺𝗱 file for AI agents.

𝗦𝗢𝗨𝗟.𝗺𝗱 is the one file you write yourself for an AI agent.

it sits at the top of the system prompt, before memory, before skills, before tools. it defines who the agent is when it shows up.

an hour spent on it changes every conversation that follows. most other layers update themselves. this one is yours.

i just broke down what a 𝗦𝗢𝗨𝗟.𝗺𝗱 file that actually works looks like.

here are the 8 sections that matter:

→ identity (a one-line statement of who, not what)
→ core truths (imperative principles, each with a one-line unpacking)
→ worldview (opinionated takes by domain, sharp enough to predict)
→ voice (concrete rules for how the agent talks, not adjectives)
→ expertise (primary domain, fluent tools, where it defers)
→ boundaries (explicit "won't" lines, no soft language)
→ memory policy (what persists, what stays private)
→ pet peeves (phrases and tones the agent never produces)

generally people write "be helpful and professional" and call it done.

that changes nothing. every model already tries to be helpful and professional by default.

the agents that compound have 𝗦𝗢𝗨𝗟.𝗺𝗱 files with real opinions, hard limits, and a voice you can predict before you read the response.

a strong 𝗦𝗢𝗨𝗟.𝗺𝗱 is 30 to 80 lines. specificity beats coverage.

bookmark this. the first agent you build will need it.

i wrote a full masterclass on Hermes Agent that walks through the 𝗦𝗢𝗨𝗟.𝗺𝗱 layer, the three-tier memory system, the self-evolving skills loop, and how to run three specialized agents on your machine 24/7.

the article is quoted below.

[JR Farr @jrfarr]

Distribution is no longer optional.

Everyone wants a growth hack or a viral loop, but most of the growth at Lemon Squeezy came from doing a lot of small things for years.

Some practical things that worked for us:

1. Shipping constantly

We created “Lemon Drops” and every Friday, we shipped something. Sometimes big, sometimes tiny. But every single week we had:

>tweets
>a blog post
>a product release
>a changelog update
>screenshots/videos
>customer conversations

This mattered way more than trying to engineer one giant launch every 6 months.

2. Turning product work into distribution

Every feature became content. Every integration became content. Every customer problem became content.

We stopped thinking: “How do we market this?”

And started thinking: “How do we package the work we’re already doing into something discoverable?”

3. Building evergreen content loops

We built “Wedges” and gave it away for free (open source). It wasn’t directly monetized, but it gave us something useful to share constantly.

And designers and developers loved it. People tweeted it. It ranked on Google, and it introduced people to the brand. Over time, it became a flywheel.

A lot of good distribution is just creating assets that keep working long after you publish them.

4. Obsessing over onboarding

I think founders massively underestimate this. Reducing friction is distribution. Every extra step in onboarding kills word of mouth.

We spent a huge amount of time improving signup flow, activation, dashboards, copywriting, error states, emails, and all the boring stuff.

Growth gets easier when people actually make it through the front door.

5. Making docs part of the product

Our docs drove an insane amount of traffic. Not because we “did SEO” but because we answered real questions developers were searching for.

Most company docs sound vanilla. We tried to make ours actually helpful.

Distribution increasingly comes from being useful at scale.

6. Integrations everywhere

Every integration unlocked another ecosystem. Another search surface. Another community.

Integrations are underrated forms of distribution because they borrow trust from existing platforms.

7. Founder-led content

I think I had <1,000 followers when I started Lemon Squeezy, but people trust people more than logos. Especially now.

I posted constantly.

>lessons
>launches
>podcasts
>screenshots
>customer stories
>product thoughts

Founders underestimate how much simply showing up every day matters.

8. Customer support as marketing

Early on, support was one of our biggest growth channels. Answering fast and being human matter. People remember how you make them feel when something breaks.

If you follow me, you know I still live by this, and I've carried this mentality into my role(s) at Stripe.

Support builds trust faster than ads ever will.

9. Screenshots matter more than people think

It sounds silly, but it’s true. Products that look good spread easier. People tweet screenshots, and good design is distribution.

10. Launching over and over again

We never really stopped launching.

>every feature = launch
>every milestone = launch
>every integration = launch
>every partnership = launch

Not in an annoying way. We just consistently stayed in motion. The internet rewards momentum.

11. Building in public before it was cool

We talked openly about numbers, growth, problems, product decisions, and lessons learned.

Transparency created trust, and that trust created distribution.

12. Creating systems instead of random bursts of marketing

This is probably the biggest thing. Most startups market in bursts. They build towards one big launch and then disappear for 3 months.

We built systems:

>weekly emails
>weekly content
>weekly launches
>weekly improvements
>weekly customer conversations

Consistency compounds harder than intensity.

Product still matters deeply, but a good product alone is rarely enough anymore.

Looking back, almost none of this was one giant breakthrough moment.

It was thousands of small reps stacked on top of each other for years.

That’s what compounds.

[Abhishek Sisodia @abhirb]

@jai_mansukhanii Amazing opportunity. Would be even better if you also offer hybrid/remote flexibility

[Abhishek Sisodia @abhirb]

@AutismCapital It's indirectly prepping them for nothing post school

[Abhishek Sisodia @abhirb]

Time to retry @Replit this weekend! May 2, 8pm EST onwards unlimited 24hr usage. Let's go!!

[Amjad Masad @amasad]

Set your calendars like @aroogle did

[Aaron Levie @levie]

Great read. AI lets you get tremendous leverage that wasn’t available before in almost any domain.

That means we’re at a unique moment in history where anyone with a high level of ambition and core skills in any area can overcome a lot of historical experience requirements for their role.

This can apply to anyone who’s junior or senior, but it’s pretty sweet that you can do far more than you could have accomplished as a newer employee than even a couple years ago. The people that take advantage of this will get ahead massively.

And the companies that find this talent within or outside should put them in key positions to get as much out of them as possible. These people will seem strange and from the future, but they will help you figure out where things are going. Everyone company should be doing whatever they can to find them.

[Mind of Saka @WFAFOOTBALL14]

“Pressure is a Privilege. And if you’re feeling any pressure or weight of expectation, you are breathing rare air, that very few of us get to live inside”.

Unbelievable piece of poetry .

[wincy.eth @gusik4ever]

the fastest growing GitHub repos in finance this week:

1. shiyu-coder/Kronos (+6.5K ★)

first open-source foundation model for financial candlesticks. trained on 45+ global exchanges. predicts OHLCV candles as tokens — literally GPT for price charts. accepted at AAAI 2026.

2. virattt/ai-hedge-fund (+4.9K ★)

a team of AI agents simulating Buffett, Munger, Ackman, Cathie Wood and others. each agent runs its own strategy, a Portfolio Manager makes the final call. one of the most viral finance repos right now.

3. TauricResearch/TradingAgents (+~3K ★)

multi-agent LLM trading framework. fundamental analyst, sentiment analyst, technicals, risk manager — all working together. supports GPT-5.x, Gemini 3.x, Claude 4.x, Grok. built by UCLA/MIT researchers.

4. ZhuLinsen/daily_stock_analysis (+~2K ★)

LLM stock analyzer for US, A-share and H-share markets. auto-builds a daily decision dashboard with exact entry/exit levels. pushes to WeChat/Telegram/Discord/Email via GitHub Actions. zero cost, zero server.

5. hsliuping/TradingAgents-CN (+~1.5K ★)

Chinese fork of TradingAgents. fully localized for A-share markets (Shanghai/Shenzhen), Chinese data sources, and domestic LLMs. 5.1K forks — very active community.

6. OpenBB-finance/OpenBB (+~1K ★)

open-source Bloomberg alternative. stocks, crypto, options, derivatives, fixed income — one platform. integrates with AI agents via MCP. 66K total stars and still climbing.

7. freqtrade/freqtrade (+~700 ★)

free, open-source crypto trading bot in Python. supports all major exchanges, full backtesting, strategy optimization, Telegram control. release 2026.3 just dropped.

8. AI4Finance-Foundation/FinGPT (+~500 ★)

open-source financial LLMs trained on real market data — news, filings, earnings. built for sentiment analysis and robo-advisors. models on HuggingFace, ready to deploy.

9. juspay/hyperswitch (+~400 ★)

open-source payments router in Rust. one API to connect Stripe, Adyen, PayPal and 50+ providers. smart routing, high performance, built for fintech scale.

10. microsoft/qlib (+~350 ★)

Microsoft's AI quant investment platform. covers the full pipeline: alpha seeking, backtesting, model training, live trading. supports ML/DL, RL, and auto-quant.

bookmark this and start today.