Francis Thomas

@AlisonbobEth @p_millerd but it also short circuits information flow, juniors embedded in firms have different incentives. Consulting bypasses human org constraints. It is actually just presenting on slides, what people in the firm know and connecting enough dots.

I remember I worked for a company that was using McKinsey and they just had some junior person interviewing us about our jobs. I suggested to management that they could save $20m by hiring like 3 people and doing the same thing. Alas, the C-suite just wanted the rubber stamp, in case anything went wrong.

consulting is a great and valuable industry. always has been every company that gets big enough eventually realizes the constraint is implementation / adoption of their products or services. underneath that, human resistance to change every company from GE to IBM to you name it has eventually developed their own internal consulting teams in addition to relying on external consultants who quickly understand new market opportunities I think we are seeing these rapid and large scale moves into consulting (or call it FDE whatever you feel like) because of a few things: 1. downside risk on the capital deployed is enormous. these bets, while large are tiny % compared to total capex across the biggest players 2. The speed of adoption of the models is rapid, but the speed of turning tokens into dollars for businesses will likely not be as rapid. The success of the model companies relies on turning tokens into dollars (Either in opex savings or increased profits) as fast as possible. Consulting helps activate more efficient capital/labor/token allocation across enterprises consulting has been underrated by tech for years. it is a highly market efficient industry and the best firms adapt much much faster than traditional companies I happened to be working at McKinsey in 2008 during the global financial crisis. from september 2008 to december 2008, the firm had pivoted almost every global practice to responding to the practive. endless new service lines were created, new capabilities were developed. many also flopped or didnt go anywhere but large-scale change efforts became a much bigger business over the next few years, the big consulting firms will continue to thrive - mck, bain, bcg but also accenture, and other IT oriented firms. They have the labor at scale who understand change and implementation. Most people's model of consulting is based on a caricature of a 1980s strategy case, with a report at the end of the project handed to the board. that is no longer the case. McKinsey is a full service firm with a digital arm, an implementation arm, analytics and IT solutions, etc. the scale of these firms continues to astound me. They will only get bigger. The core "service" they offer is a 3-5 person team and since each 3-5 person team can work on different things and iterate on each project, these firms are highly adaptable. But in a growing market there is room for more, and who benefits? well, look at the investors here: "Investors also include leading consulting and systems integration firms, including Bain & Company, Capgemini, and McKinsey & Company." long consulting

@NarasimhaKan its not the count but the quantum. but yeah investments from india dont drive the momentum.

people routinely eulogize such rhetoric rooted in falsehood. for all this talk about momentum, <1mn indians have ever invested abroad. US is >60% of listed equity m-cap globally. this is a dumb yardstick to measure investing in a market with incredibly high earnings growth + frontier tech advancements also, Naren is not even an authority on momentum L take

@petergyang @eshita Should have compared with @antigravity . It breaks often. ,but flash 3.0 is a great workhouse.

I've spent way too long testing OpenClaw, Hermes, Claude Code, Codex, and Gemini as my personal agent. The truth is, nobody has won this race yet. Here's my new deep dive with my honest take on where each product stands, plus the personal agent stack I use right now. 📌 Read now: creatoreconomy.so/p/the-race-to-…

@PositiveGamma It’s the GMV of this era.

Indian NBFC stating token usage as proxy for AI adoption shows how AI-pilled the market is. a. All tokens are not created equal from a utility standpoint b. Only thing that matters is long term Return on Equity being substantially higher than cost of equity.

@kunalvg But it is also slop. Why bring more slop into the world.

AI + Astrology will be a billion-dollar category. And the rationalist builders who laugh at this will miss all of it.

@Patrick_Bush_VE @jessepollak Local stablecoins have little use in countries which are dollar hungry. Ability to spend USD stablecoins with ease digitally/irl will reduce costs than fees on off-ramps.

@jessepollak @pitdesi I appreciate the optimism. I guess it depends upon one's definition of time. None of the non-USD currencies listed in the report are major remittance sinks. Local currency stables appear far away because of onerous government restrictions. It will take a long time.

How can stablecoins reduce remittance fees to zero? makes no sense to me whatsoever. The bulk of remittances $ is migrant workers sending money to family, from US->Mexico, US->India, Gulf->India Stablecoins can make the money-movement leg much cheaper/near-zero but that isn't the bulk of the cost! You still need cash-out, FX, compliance/KYC, fraud controls, customer support, and distribution. When I send money to my cousin in a village in India, there is nothing he can do with USDC. He needs to get the money out into INR and that last mile is where most of the cost lies. Stablecoins can make remittances much cheaper but nowhere near zero.

@joeljohn Also to let people buy tokens, open router let’s you buy with stables but then Coinbase commerce takes 5%.

if i were working at a stablecoin company - i'd beg the team to let me just spend 2 weeks in any south-east asian market haggling with $100 all day long to understand the psyche of the end user.

stablecoins will find wicked pmf in any tourism fuelled south-east asian country with ±10% fx cost from foreign currency to local ones (say INR) currently this works when you use a stablecoin-native card. think its a function of time before the segment is disrupted entirely

@championswimmer This was really nice.

Added a lot more runs of elevator bench to the website. Few conclusions on model quality (this is true for 25 Apr 2026) 1. Opus 4.6, 4.7 are very good models, but lead is not that big. 2. Opus is both extremely costly and token inefficient 3. Many models clearly better than Sonnet (GLM 5.1, Kimi K2.6, GPT 5.3, 5.4, 5.5) and yet costs much much lower than Sonnet. I would highly recommend using Opus only for extremely complex tasks. It is costly, slow and generates way too many tokens for simpler tasks. I would also recommend not to use Sonnet and Haiku at all. Those are the worse cost-to-intelligence models available on earth right now. Also, locally runnable models (that can run on Macbook with enough RAM) have crossed the Haiku 4.5 threshold clearly. Qwen3.6-35B is hands down better than Haiku 4.5 for any coding task. championswimmer.in/elevator-bench/

@robotevm @withfabricxyz is another source.

huge shoutout to @EnsoBuild, great data, love to bring more transparency in favor of getting traders the best execution on their trades!

1/ new aggregator performance dashboard just dropped from @EnsoBuild people accuse me of being biased when posting our own data so love to finally see another team open up the black box deep dive + how it’s different from our traders data

Meta got to launch personal agents connected to one’s Instagram now.

@evisdrenova Atlas was great till chrome got the Gemini button.

crazy how the AI browser battle completely fizzled out Arc Dia Atlas Comet pretty much all dead

@julien_c @huggingface Isn’t antigravity half of cursor ? It is also free, but has way too many bugs . Codex also benefits from Go plan which is 4 dollars per month in India. Claude is min 20 dollars . Codex is very very good.

Currently: - Claude Code is 4x bigger than Codex - Codex is 2x bigger than Cursor - Antigravity is almost as big as Cursor (but probably just because all Googlers use it? 😅) We'll add tagging for other agents asap. (Source: one data point from the @huggingface Hub team. Your usage may vary)

What did you think about @aboutcircles , it was the gnosis team UBI thing with coordination based on trust getting money. Ideally should incentive virtue and people to move away from land traps and do positive sum interaction , encourage agglomeration and re order irl for effiency.

Obviously the libertarian caveats are irrelevant here but perhaps adding them will lead to a direct answer

Let’s try this again AI replaces human labor, and accrues some fraction of the $100T in annual wages; humans cannot earn an income in the market Do those companies, on a private, voluntary basis, then regularly airdrop tokens to humans as customers (to drive revenue and inform resource allocation?), or let everyone starve as end demand collapses to zero

@RonakTwts @StartaleGroup Congrats.

I've been promoted to Head of UAE at @StartaleGroup. I'll be leading Startale Group's entire expansion across the UAE: strategy, partnerships, ecosystem, everything. Startale Group is now officially building from the heart of the UAE’s most ambitious tech ecosystem. First, some context on what we're building at Startale Group: → @Soneium for Onchain Entertainment in partnership with @Sony Block Solutions Labs. → Strium for Tokenized Securities & RWA Trading (Onchain Finance) in partnership with SBI Holdings. → @JPYStableCoin (JPYSC): The world's first Trust Bank-backed JPY stablecoin in partnership with SBI Holdings. → @StartaleApp, an all-in-one Super App gateway to the Startale ecosystem. These are infrastructure being built with two of the largest corporations on the planet. And today, even more news, @StartaleGroup has been selected by @hub71ad, Abu Dhabi's global tech ecosystem, an initiative by @AbuDhabiDED, powered by @Mubadala, and based in @ADGlobalMarket. I’m especially grateful to @WatanabeSota, @w3_shunp, @henskensm, @0xShaunW, @cypher_nk, Jasmine Tan, @yujikumagai_, @chrischoitweets, @KNG_massu and others for the opportunity and the trust. Shoutout to my awesome team @adriagf, Alicia Lee, @arhit_das, @Juminstock, @gaius_sama, @businessj386, and @0xRamz. If you're based in the UAE (or have strong ties to the region) and are building for the globe I want to hear from you. 📩 Drop a reply, send a DM, or just follow along. The door is open. Let's build something the world talks about.

@drydenwtbrown So UHI should fund content graded by AI to be truthful, positive sum outcomes.

White collar workers unemployed by AI will not retrain as electricians They will become socialist pamphlet distributors or jihadists

@theakshaytyagi @anmolm_ It’s useful to create a base of operators for these companies. People who have hacked around the edges , have discovered some form of knowledge through action, time spent etc.

@anmolm_ I think our frames of reference are different. Power laws come into play after discovery has happened. But startups have to die and VCs have to lose money for that funnel to exist in the first place.

in the venture capital biz, only a handful of companies matter every year - expanding the base of entrepreneurship is fairly irrelevant

@neilshroff People losing investment on risky ventures is how S&P becomes risk adjusted for normies though.

WTF & NK have been at it nonstop for the last 3 years: OG podcast killed it, a grant fund that probably wouldn’t beat S&P500 if it was a fund, some other goodish media stuff, and now an interesting incubator for FMCG brands... the lesson from this is that even if you’re giga rich you still don’t get to win easily (unlike what most may believe), but you infinite lives to keep trying and trying until you win, die, or simply get bored.

@gfodor @Timcast Probably simulated work, post content graded by AI. People will end up at different tiers by motivation alone.

@Timcast what's the plan then when 80-90% of people are unemployable. so far the only plans I've heard are UBI or Communism.

Nails it UBI does not work

@madhavg @janwhyy yes. but walking 500m was still ardous and a chore, when chained to a laptop or taking a detour while driving back home.

@franciskthomas @janwhyy Bigger cities usually had natures basket adjacent grocery stores to get premium stuff ig

Quick commerce is a $5.4B market in India growing at 70-80% CAGR. Blinkit, Zepto, Instamart collectively do 4M+ orders a day. But the interesting story isn’t the business. It’s what it did to how Indians buy things. 7 shifts nobody saw coming. 1/ 73% of q-com orders happen outside traditional shopping hours. 10pm-1am is now peak for ice cream, condoms, skincare, snacking. Three years ago this buying window didn’t exist. An entirely new consumption slot was invented and nobody’s talking about it. 2/ ₹149 mini sunscreen outsells ₹599 full size on Blinkit. The full bottle is a commitment. The mini is a maybe. Consumers are treating q-com like a sample store. Brands without trial SKUs are invisible. 3/ Brand loyalty disappeared in grocery. Search “atta” on Zepto. 8 brands sorted by delivery time. The one in the nearest dark store wins. Not the one your mom used. For staples, proximity replaced preference. Terrifying if you’re a legacy FMCG brand. 4/ Kirana shops aren’t losing staples. They’re losing the ₹50-200 impulse buy. The chocolate, the chips, the random face mask. The small purchases that used to happen because you were already in the store. That foot traffic is gone and it’s not coming back. 5/ Men started buying skincare. The anonymity of tapping “face wash” on Blinkit vs asking for it at a medical store broke a psychological barrier nobody was talking about. Embarrassment was the barrier all along. Men’s grooming on q-com is growing faster than any other beauty subcategory. 6/ Blinkit and Zepto aren’t delivery companies anymore. They’re media businesses. Blinkit’s ad revenue grew 220% YoY. Both crossed ₹1,000 Cr in annual ad revenue by FY25. Ads are now 15% of Blinkit’s total revenue. If you’re thinking of q-com as just a listing channel, you’re missing the point. 7/ The delivery bar moved for everyone. If Zepto delivers in 10 minutes, why does your D2C site take 5 days? The consumer doesn’t separate “q-com speed” from “normal speed.” Every brand shipping in 3-5 days is now competing against a 10-minute standard they didn’t set and can’t match. Quick commerce didn’t just create a new delivery channel. It rewired how 50 million Indians think about buying things