Head of Product, @comet & Computer • I build great teams and products • Previously: @convex, @googlephotos, @bumptech, @zintinapp
Mountain View, CA
Joined October 2007
- Tweets3K
- Following894
- Followers1K
- Likes1.8K
Indy Khare retweeted
You have heard of the @convex LLM Leaderboard but did you know we also have a Decision Leaderboard?
Interesting datapoint is @perplexity_ai's Decider currently tops the leaderboard and costs less than half the price of Jev if you are willing to sacrifice a little response time!
One hypothesis I have is that at some point building and sharing custom websites will become a new norm, beyond just sharing static docs or presentations.
Some of the most powerful communicators use data, graphs, custom animations, diagrams and interactions to explain their ideas. Websites are already a universally available and infinitely flexible format for this.
AI is already making this easier, but more to be done.
Indy Khare retweeted
No one is going to care about the underlying AI model soon, especially for consumers. The models are already good enough for most of what people need. What consumers want is for AI to be useful, easy to access, and FREE or bundled into something they already pay for.
Clayton Christensen and Michael Raynor described this shift in the figure below which is in their book The Innovator’s Solution. When a product isn’t good enough, performance wins. Once it is good enough, convenience and cost start to matter more.
The winning question may soon be less Whose model is best? and more Why would I pay separately for this?
One of the things about software was that it made non-deterministic processes in the real world more deterministic. Something you could predict and build on top of. That can't be going away, that very stability let's us raise the level of abstraction in our work. And do more things with less energy.
The big win of modern AI is that it can help us interact further and deeper with the non-deterministic world. And further bridge it to the digital world.
There is a place for everything.
User interfaces are not going away.
Yes, humans will use computers alongside agents. When working with agents, we will need the ability to audit what agents do in a human-readable way.
Yes, some interfaces (ex: dashboards, in-chat UI for ephemeral tasks, etc) will be generated. Most important interfaces will be intentionally designed. We will want both deterministic and non-deterministic interfaces in different situations. We're thinking a lot about the best ways to help designers shape non-deterministic systems at Figma.
Yes, new input modalities like voice, VR/AR and BCI will change the way we use computers and introduce lots of new interaction paradigms. This is all very exciting :)
Overall, the surface area of design will continue to grow much broader. Design is the new code and the key differentiator for software moving forward. Expect more and more energy to go into design / craft in the years ahead... it will be how you win or lose.
Eric is right about plenty of things, but the market will prove him wrong on this one...
I remember a few years ago someone said to me that they only wanted to drink soylent.
You get the outcome, but you miss out on the joy of cooking, connecting with a culture of that cuisine, trying new and interesting things with friends.
I think Silicon Valley is making a fundamental mistake with personal agents… optimizing entirely for the outcome and forgetting that, for a lot of things, the process is a very core part of experience.
Travel is the obvious example. People will almost always never want “book me a trip.” They want to browse, compare, daydream, change their mind, send options to friends, and eventually book.
And the average person probably has far fewer recurring tasks worth delegating than the AI industry seems to think.
That’s why personal agents feel more like a feature that gets absorbed into existing products than a standalone category.
This is the power of a general purpose agent that has access to all the models, your memories, your projects. No context switching, or the human or the agent.
Yep that's right. The harness UI is where more and more work now begins for folks who live on the frontier. The browser as we know will become a feature. (At work)
Indy Khare retweeted
One of the things I loved most about academia was being surrounded by people rewarded for questioning assumptions.
We’ve tried to preserve that at Traversal, while combining it with something academia rarely offers: seeing new ideas put to work in the real world at scale.
Researchers, distributed systems engineers, infrastructure experts, and SREs work alongside GTM talent from across observability and infrastructure. We get to pursue hard technical questions without established answers, then turn those ideas into something enterprises can actually adopt and use.
Proud to see @traversal_ai recognized by @FastCompany as a Best Workplace for Innovators. Even prouder of the team that made it one.
We’re hiring across the company and adding new roles every week: traversal.com/careers?utm_so…
Lot's of talk of the beginning of the AGI era right now, and society of minds that are living above us.
It reminds of watching birds fly around the sky or squirrels jumping over our fences.
No matter how we've created our property lines, these animals don't care. They just go about living their lives.
Yes we've made it hard for them, but what us "higher order beings" care about and what they care about are so vastly different that it seems like a folly to try to relate.
Whether this is a thought exercise of doom or hope, I don't really know. But at least it points to just how different different beings can be.
Indy Khare retweeted
Connectors are now available in Agent API.
Connect your agents to GitHub, Slack, Google Drive, and Datadog without passing a server URL or token on every request.
API Group admins can connect each service once for all API Group members.
pplx.ai/api-console
Indy Khare retweeted
at perplexity we are really excited about multi-agent collaboration! we recently explored a simple instantiation of this with advisor escalation from a local model to a remote frontier model, where we showed it can significantly boost the local model's performance.
an alternative way i like to think about this: the local model becomes a **gateway** to remote models. its main job is to preprocess all the raw tokens, package them into fewer, more information-dense tokens, and send only those to the server. this preprocessing has several advantages: it reduces cost and latency, enables personalization, and allows privacy controls. already such approaches can save 50% of tokens, but imagine a future where 90%+ of tokens are processed locally and only 10% are sent to the server for the last-mile hardest reasoning. that is a 10x cost reduction and a much more pleasant UX due to low latency.
training such a multi-agent system is not an easy problem and still requires a lot of research in multi-agent RL, but we are making good progress and actively hiring for this. if this sounds interesting please DM me
Today we’re launching Portable Computer on @NVIDIA DGX Spark.
Portable Computer is a fully local version of Perplexity Computer, where the entire runtime: orchestrator LLM, subagent LLM, agent harness all run on your local hardware. No cloud dependency.
Having previously done the job, don't let this detract you. These folks are absolute gems to work with.
We've been hiring for a couple of product and product-adjacent roles, and it's been a really interesting revelation.
We're not the slickest salespeople. Neither James nor I have an eye-popping amount of Twitter followers. But we've tried to make sure the bucket isn't leaky. If people hit the product, they love it and stick around.
Makes us pretty paranoid about giving up ownership of something we've obsessed over so much.
Brain is an incredible feat. If you are building agents and self improving loops this is a great read.
Really proud of the team! An exciting new world of possibilities is now open.
Today we’re launching Portable Computer on @NVIDIA DGX Spark.
Portable Computer is a fully local version of Perplexity Computer, where the entire runtime: orchestrator LLM, subagent LLM, agent harness all run on your local hardware. No cloud dependency.