Sandhya

6.6K posts

Sandhya banner
Sandhya

Sandhya

@sandhya

Co-founder @Calibre_Labs | Applied AI research & consulting | Agents, AI Evals | prev EVP @amplitude_HQ | VC @khoslaventures @sequoia | @stanford @iitbombay

가입일 Mayıs 2009
343 팔로잉8.9K 팔로워
Sandhya
Sandhya@sandhya·
@spenserskates So thrilled to work on this with @Amplitude_HQ. Combining llm traces with user behavior is incredible context for coding agents, being able to tell which harness update improved user retention will be 🤯🤯
English
1
1
8
355
Sandhya
Sandhya@sandhya·
Refactored an "old" codebase from Jan today and lord, the improvement in performance with Opus 4.6 felt like a fresh breeze. wtf.
English
0
0
2
321
Sandhya
Sandhya@sandhya·
Super specialists sharing battle-tested Claude Skills is my favorite new software trend. It's the end of the "talk to my ai" era (finally). Random influencers learning how hard it is to maintain open source projects is a related and entertaining accident :D
English
1
0
2
323
Guido Appenzeller
Sorry to see Granola @meetgranola going closed. They encrypted their local db, no local and no cloud API. In a world where notes are managed by agents, the app now has zero value. Any recommendations for good alternatives? What are you switching to?
English
178
11
633
335.3K
Sandhya
Sandhya@sandhya·
@thesamparr You can literally just get Claude to create an interactive custom onboarding plan for your own needs and give you exercises to do to learn new workflows on Cowork and Code. Wasting time watching podcasts and hiring expensive consultants to run trainings is very 2024.
English
0
0
0
31
Sam Parr
Sam Parr@thesamparr·
How is everyone getting team adoption for Claude? I spent a lot of time on Twitter, as do you. We see all this AI stuff popping up. We're on top of it, or at least sorta. I know what's going on and are testing all these fringe ideas. But how are all you people getting your team to actually use it effectively without spending all their time on Twitter and learning, which we know they won't and probably shouldn't be?
English
274
21
513
282.8K
Sandhya
Sandhya@sandhya·
It makes little sense to be an indie software developer selling subscriptions to point SaaS solutions anymore. Completely out of vogue and not fun to maintain. Too much SaaS tool bloat in the ecosystem and personal software is the future. But not everyone will build their own. Open source + bring-your-own-key is the way for most use cases that are on-demand tasks. If it’s a background agent/always on task, maybe use stripe’s new LLM billing method to offer a fully hosted solution with a good open source API.
English
0
1
5
419
Sandhya
Sandhya@sandhya·
@HamelHusain Show me the incentive as Charlie said! … If you are attention farming for a living, volume is all that matters and the right slopbait gets more clicks
English
0
0
0
116
Hamel Husain
Hamel Husain@HamelHusain·
I don't want read things the author feels isn't worth the effort of writing/editing in many cases Many people authoring slop are simply swapping one kind of audience for another. I suspect in many cases trading for lower IQ audience
English
7
0
37
2.5K
Liz Wessel
Liz Wessel@lizwessel·
Wow. Gumloop has gone from a side project out of a Vancouver bedroom to an AI platform that now automates daily workflows at companies like Shopify, Ramp & Instacart, and a new $50M Series B led by @benchmark (all in ~2+ years!!). It’s been a wild journey for the team, and I feel incredibly fortunate that @firstround led their 2024 seed round. If you tried it a while ago and still think of @Gumloop as just drag-and-drop workflows, I’d *strongly* recommend giving it another shot, as the product has evolved in massive ways. They’ve now added their agent builder, Gumloop Agents (lets anyone at a company build/deploy AI agents across workspaces in minutes) and Gumstack (a separate security product that lets IT teams monitor/control how agents use company data across the org). As a user of theirs said to me recently, ever since Gumstack launched, “Gumloop *IS* Gumloop for Enterprise.” As they’ve built out their product, the Gumloop team has stayed super focused on making it maximally useful to everyone – not just technical folks. IMO, this is a big part of why teams are getting hooked, and usage spreads wall-to-wall, instead of getting stuck in one department. @MaxBrodeurUrbas, @rbehal1729, and their entire team (who are all amazing btw) have been obsessed with this from day 1 and truly stay embedded with their customers, flying to their offices, running hundreds of workshops, shipping features same-day, and personally answering thousands of questions in customer Slack channels. They’re hiring across the board right now, more info below!
Max Brodeur-Urbas@MaxBrodeurUrbas

gumloop raised a $50m series b led by benchmark here's a video we had fun making about the journey back to work.

English
13
16
130
31.7K
Sandhya
Sandhya@sandhya·
@ZacharyDeWitt Claude would frown on the lack of NPV analysis in this approach .. i mean where's the terminal discount rate for the singularity
English
0
0
2
136
Zach DeWitt
Zach DeWitt@ZacharyDeWitt·
how growth rounds are priced... 100–200× the net new arr added last quarter for example, if a startup added $5M in net new arr last quarter, the valuation could be $500m–$1b. the exact multiple depends on moats, margins, logo quality, roadmap, etc
English
7
3
109
19K
Sandhya
Sandhya@sandhya·
@ttunguz Are you testing open source models for browser use workflows? I'd love a deep dive!
English
0
0
0
50
Tomasz Tunguz
Tomasz Tunguz@ttunguz·
A $5,000 laptop — a MacBook Pro with enough memory to run Qwen locally — pays for itself after 556 million tokens. At my usage rate, that’s about a month. At 20 million tokens per day, it’s four weeks. After payback, the marginal cost drops to electricity. It isn’t an intelligence compromise. Reasoning, coding, agentic workflows, document processing, instruction following : the 9B model matches December’s frontier across the board.
Tomasz Tunguz tweet media
English
3
4
19
4.8K
Tomasz Tunguz
Tomasz Tunguz@ttunguz·
I burned 84 million tokens on February 28th. Researching companies, drafting memos, running agents. That’s running Kimi K2.5, a serverless model via API. At Claude or OpenAI rates — roughly $9 per million tokens blended — equivalent usage would cost $756 for a single day’s work. My peak days hit 80 million tokens. My average days run 20 million. Cloud inference at frontier-model pricing adds up fast.
Tomasz Tunguz tweet media
English
17
8
119
39.1K
Sandhya
Sandhya@sandhya·
@swyx Opencode with vercel’s agent-browser
English
1
0
9
1.6K
swyx
swyx@swyx·
ok are there any open source Claude Cowork clones because I can no longer function without a cowork pls recommend or i will build
swyx tweet media
English
59
3
116
64.9K
Sandhya
Sandhya@sandhya·
Dear influencers, Please have your AI tools write shorter articles, your thought leadership is getting too wordy and 10% of it is just it’s not X, it’s Y statements. Nobody needs this.
English
0
0
3
219
Gabe from Kodus
Gabe from Kodus@gamalinosqui·
@ericclemmons did you make this? It's absolutely gorgeous! I'm curious, was it vibe-coded?
English
3
0
4
8.2K
Sandhya
Sandhya@sandhya·
@avemii Yes but much harder to continue in that mode now, no excuses left
English
1
0
0
15
Aditya 🙏👋
Aditya 🙏👋@avemii·
@sandhya This isn’t all ai…mis management is def a big part of this
English
1
0
0
16
Sandhya
Sandhya@sandhya·
First of many many more to come
jack@jack

we're making @blocks smaller today. here's my note to the company. #### today we're making one of the hardest decisions in the history of our company: we're reducing our organization by nearly half, from over 10,000 people to just under 6,000. that means over 4,000 of you are being asked to leave or entering into consultation. i'll be straight about what's happening, why, and what it means for everyone. first off, if you're one of the people affected, you'll receive your salary for 20 weeks + 1 week per year of tenure, equity vested through the end of may, 6 months of health care, your corporate devices, and $5,000 to put toward whatever you need to help you in this transition (if you’re outside the U.S. you’ll receive similar support but exact details are going to vary based on local requirements). i want you to know that before anything else. everyone will be notified today, whether you're being asked to leave, entering consultation, or asked to stay. we're not making this decision because we're in trouble. our business is strong. gross profit continues to grow, we continue to serve more and more customers, and profitability is improving. but something has changed. we're already seeing that the intelligence tools we’re creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company. and that's accelerating rapidly. i had two options: cut gradually over months or years as this shift plays out, or be honest about where we are and act on it now. i chose the latter. repeated rounds of cuts are destructive to morale, to focus, and to the trust that customers and shareholders place in our ability to lead. i'd rather take a hard, clear action now and build from a position we believe in than manage a slow reduction of people toward the same outcome. a smaller company also gives us the space to grow our business the right way, on our own terms, instead of constantly reacting to market pressures. a decision at this scale carries risk. but so does standing still. we've done a full review to determine the roles and people we require to reliably grow the business from here, and we've pressure-tested those decisions from multiple angles. i accept that we may have gotten some of them wrong, and we've built in flexibility to account for that, and do the right thing for our customers. we're not going to just disappear people from slack and email and pretend they were never here. communication channels will stay open through thursday evening (pacific) so everyone can say goodbye properly, and share whatever you wish. i'll also be hosting a live video session to thank everyone at 3:35pm pacific. i know doing it this way might feel awkward. i'd rather it feel awkward and human than efficient and cold. to those of you leaving…i’m grateful for you, and i’m sorry to put you through this. you built what this company is today. that's a fact that i'll honor forever. this decision is not a reflection of what you contributed. you will be a great contributor to any organization going forward. to those staying…i made this decision, and i'll own it. what i'm asking of you is to build with me. we're going to build this company with intelligence at the core of everything we do. how we work, how we create, how we serve our customers. our customers will feel this shift too, and we're going to help them navigate it: towards a future where they can build their own features directly, composed of our capabilities and served through our interfaces. that's what i'm focused on now. expect a note from me tomorrow. jack

English
1
0
1
718
Sandhya
Sandhya@sandhya·
Going from synchronous agents (prompt & response) to swarms/cloud agents requires way more understanding of the domain. You have to - define the problem more clearly - anticipate what issues might come up - know the best way to test results in advance - avoid traps where LLMs might waste your token budget Every step of progress makes AI even more useful to people who are already experts…. And the outlook gets worse for people who can’t learn new things. AI is leverage and amplification beyond anything we have ever seen. The value of your knowledge is higher not obsolete.
Michael Truell@mntruell

x.com/i/article/2026…

English
0
0
2
636
Sandhya
Sandhya@sandhya·
The real version of “agents running all night” where you babysit them till 3 am and wake up excited at 6 ;) loved this thread. This is what the jagged frontier feels like. Pushing on it takes real sweat and tokens. None of us are working shorter hours cause it’s more fricking fun to try new things than just drink margaritas on the beach.
Jess Martin@jessmartin

today's attempt at raising my ambition: give Claude an entire release (~20-30 user stories) and have it ralph loop all night with codex as implementor. Can I hand off ~8 hours of work and have quality work come out?

English
1
0
3
665