Travers (@travers00) / X

Travers

551 posts

Travers

@travers00

// building a parallel web for AIs // cofounder @p0 //

San Francisco, CA

Joined August 2021

Pinned
Travers
@travers00
Jul 13
turbo mode is web search that's super fast (~200ms), super affordable ($1 per 1,000 searches), and high quality, opening up a whole new frontier of what agents can do with the web it's built for the scale and shape of demand the agentic web requires
Parallel Web Systems
@p0
Jul 13
Introducing Parallel Search Turbo: the fastest and most affordable way to bring the web to your AI.
00:00
Travers
@travers00
Nov 13, 2025
post-fundraise, back to shipping we launched our Monitor API today - google alerts for the AI web for 30 years, every web interaction started with a request: you navigate to a URL, you search, you refresh one of our core beliefs starting @p0 is that the web's information
Parallel Web Systems
@p0
Nov 13, 2025
Today, we're launching the Parallel Monitor API. Define a query once and receive ongoing updates whenever new related information appears on the web. It's like a web search that's always on, or a webhook for the entire web. parallel.ai/blog/monitor-a…
00:00
Travers
@travers00
Nov 13, 2025
our mission at @p0 is to build an open, transparent, and competitive web that can be used by all AIs, on behalf of humans the most fun part of building: working on one of the longest-standing challenges in computing (IR/search), reimagining business models for the AI web, and
Parag Agrawal
@paraga
Nov 12, 2025
Travers
@travers00
Aug 14, 2025
Parallel already powers millions of web research tasks daily for leading AI companies. We're the first to outperform humans and GPT-5 on deep research benchmarks. Today, our API is available for everyone to build with.
Parallel Web Systems
@p0
Aug 14, 2025
Replying to @p0
Parallel outperforms on independent benchmarks: • BrowseComp: 58% accuracy vs. GPT-5 (41%) vs. humans (25%) • DeepResearch Bench: 82% win rate over reference vs GPT-5 (66%). 74% win rate over GPT-5. For $0.10, we exceed what takes humans 2 hours on BrowseComp.
GIF
Travers
@travers00
Oct 8, 2025
attention on the web is shifting from humans to AIs search rankings used to optimize for pages for humans to click and browse ranking for AIs optimize for context windows, retrieving the tokens agents need to answer or make the next best decision in a loop
Travers
@travers00
Nov 6, 2025
we shipped the highest accuracy search API and the only truly built from the ground up for AI agents to use result is higher accuracy, lower cost, lower latency on end-to-end agent workflows we often talk about the AIs as our primary user, but what does that mean in practice?
Parallel Web Systems
@p0
Nov 6, 2025
Today, we’re launching the Parallel Search API, the most accurate web search for AI agents, built using our proprietary web index and retrieval infrastructure. Traditional search ranks URLs for humans to click. AI search needs something different: the right tokens in their
00:00
Travers
@travers00
Aug 21, 2025
Replying to @lucyhargreaves4
here's an interesting research report from @p0 with a deep analysis of Canadian-citizen founders with US-based companies: platform.parallel.ai/shared/deep-re… lots of companies I didn't realize had Canadian founders! @lucyhargreaves4
Travers
@travers00
Oct 22, 2025
on one side, you have to appreciate the raw product velocity of OAI as they've scaled their org immensely - it's like nothing we've seen before on the other side, it feels like at some point a year ago they had a consultant draft up a slide in a boardroom of every category AI
Travers
@travers00
Nov 12, 2025
.@nikitabier there is 2FA yubikey bug on X @paraga and a bunch of others are locked out of their accounts. Know when the fix is getting rolled out?
Travers
@travers00
Aug 14, 2025
Replying to @paraga
the web is about to get very, very fun
Travers
@travers00
Aug 25, 2025
Replying to @PalmerLuckey
we ran this on @p0's find all to generate dataset from the web: platform.parallel.ai/shared/find-al… some cool results of companies building robotics products with six or eight legs
Travers
@travers00
Oct 19, 2025
the line that stuck with me from @karpathy's @dwarkesh_sp interview: "most of that compression is memory work instead of cognitive work. but what we really want is the cognitive part, delete the memory" he's describing what he calls the "cognitive core" - the idea of stripping
This post is unavailable.
Travers
@travers00
Aug 16, 2025
every AI app will need access to two things: intelligence + the web
Will Schreiber
@breakfastbybill
Aug 16, 2025
This is what peak performance looks like @paraga @travers00
Travers
@travers00
Sep 26, 2025
Replying to @paraga
need more space on the y-axis