Today micro1 is announcing a commitment to spend $1B over the next 12 months on the acquisition & licensing of enterprise data for AI training, with financial backing from Citi and Hercules.
https://t.co/2fubOrbW1D
100% agree, but I also don’t fault the vendors. It’s a natural side effect of the industry. Tons of vendors compete for the same labs, and labs play them against each other on price and volume.
It’s a rat race. Unfortunately, it all comes down to the SPLs to execute haha
biggest issue with data vendors: overpromising and underdelivering
this is obvious, but many ignore it. you'll pitch a lab a higher volume/capacity than you can actually do, and then you have to decide between shipping not much high quality data (disappointing) or a ton of shit unusable data (even worse)
have seen so many companies hardly make it through the pilot, and proceed to fail the scale-up when they do
Every enterprise will have benchmarks + simulations across EVERY possible workflow in the future. Companies who get there the quickest have the biggest right to win in their space purely on an efficiency and cost perspective
as an enterprise, i would turn every user into an eval suite. you essentially have to:
(1) retain every user interaction with your agent
(2) create a really good synth suite that's able to understand user behaviors and create verifiers for each task (workflow the user has with your agent).
(3) create isomorphic tasks for each workflow that covers all possible axis changes for the task (i.e. change the user's name to something harder to pronounce, make their mouse clicks slower if CUA, change wording of initial prompt, etc).
(4) do all of this for every single user so you build the ultimate regression suite / ability graph over time and mine every possible failure mode.
doing this is extremely difficult btw and takes weeks of thought, but we've managed to solve the loop for a few agents.
Wow, seems like Google is buying Spirit Airlines' enterprise data for $10m (outbidding Mercor at $7.5m).
Basically includes every internal document, email, workflow, and codebase for a once $6B company.
Honestly, $10m for 34 years of operational data really seems like a steal.
Six trillion dollars will be spent online this year.
By 2030, five trillion of it won’t be spent by a person.
Every store now has two customers: people and their agents. We built Kinect so brands can sell to both.
Introducing @trykinect
This is what enterprise AI will increasingly look like as the space matures.
Customers ultimately care about three things:
• Cost
• Security & IP
• Outcomes
Data from enterprises will become increasingly valuable (more than it is right now!). Companies who take a fundamental product first approach and are able to create a forward deployed feedback flywheel at scale will be winners.
Bridgewater used their unique financial knowledge and partnered with us on @tinkerapi to fine-tune a model that helps their analysts focus on what's important. Experts improving AI that empowers experts.
https://t.co/6RJITMG2BJ