Kelsey Petrich

@hark_labs

[Kelsey Petrich @krpetrich]

@heyderekj @MostlyTechPod Thanks!! On Iowa!

[Kelsey Petrich @krpetrich]

For fun, I tried to get close to a generalized browser agent for this challenge.

So far it’s working with planning → reasoning → primitive browser actions, without any hardcoded knowledge of the specific pop-ups/traps in the challenge.

Video below is sped up for brevity. If I keep going, the next improvements are speed + better training + robustness.

cc: @adcock_brett

[Brett Adcock @adcock_brett]

Solve this in under 5 minutes and I’ll offer you $500k/year in cash plus several million in equity

I'm building a Computer-Use team, goal is to use computers better than humans

No experience or PhD needed

Instructions:
1. Solve all 30 challenges on this website in under 5 minutes: https://t.co/7DmmS5OQNl

2. Feel free to use any tools or vibe code it. Provide us a zip folder with instructions on how to run the agent and reproduce your results, as well your run statistics

3. The agent should be able to solve all the challenges, use browser, and provide overall metrics around time taken, token usage and token cost. Your agent must solve this challenge in under 5 minutes

Email your response: [email protected]

If you have any questions about this challenge, feel free to email us

[Brett Adcock @adcock_brett]

Late night sesh with the Hark agent team; I'm so excited for the potential here

p.s. we have some exciting tomorrow morning @hark_labs

[Brett Adcock @adcock_brett]

Today I'm excited to share that Hark has raised $700M at a $6B valuation

When I use these AI models today, they feel basic. They should be able to listen and talk naturally, understand vision, retain persistent memory, and become deeply personalized over time. They should be able to see the world, interact with it, and take action

To build that future, the capital we raised today will be used to:
→ scale our GPU infrastructure
→ accelerate future AI model development
→ grow the Hark team from ~70 to 200 engineers
→ design and build the next generation of AI hardware

The Series A round was led by Parkway Venture Capital with participation from NVIDIA, Align Ventures, AMD Ventures, ARK Invest, Brookfield, Greycroft, Intel Capital, Prime Movers Lab, Qualcomm Ventures, Salesforce Ventures, and Tamarack Global

At Hark, we are building the most advanced personal intelligence in the world. Intelligence that begins to think like you and sometimes, ahead of you to offload your mental workload

[Kelsey Petrich @krpetrich]

@r00k @the_transit_guy https://t.co/Jbima7FOzv - not sure if this will help you at all but they have an enormous dataset of permits.

[Kelsey Petrich @krpetrich]

@aarondfrancis Happy birthday 🎈

[Kelsey Petrich @krpetrich]

@aarondfrancis 🙏

[Kelsey Petrich @krpetrich]

@jasonlbeggs @aarondfrancis @n_haberkamp 😂

[Kelsey Petrich @krpetrich]

It did run a close modal action that forced closed it after so many attempts of just trying one by one. I've changed the modal handling since and a more pure solution for handling the survey modal. Here's some of the output from the current version where it's trying the different survey options based on what answers make sense. I'll try to post another video later this weekend after I make some more improvements.

[page] https://t.co/gr4xEbQ7uR | 40 elements | 2 modals
[planner] Turn 10
[observation] Two stacked, nearly identical survey modals exist. Scrolling the top modal reveals its own enabled Submit button (e_39) after selecting the correct option; earlier disabled Submit (e_28) was a decoy or from the upper section.
[action] click({"eid":"e_39"})
[reasoning] 1) Objective: dismiss the topmost survey modal first. 2) After scrolling, the option 'The right choice' (e_33) is now checked and a Submit button (e_39) appears enabled for this modal. 3) Submitting s
[result] ✓ click

[Kelsey Petrich @krpetrich]

Agreed! It's way too slow right now. I'm trying to see if I can get the agent to just work through the challenges at a slower rate first and then add in some performance improvements.

Right now the agent does one action at a time, so I'm hoping I can start doing multiple agents at once to really speed things up especially on modals.

[Kelsey Petrich @krpetrich]

Tried my agent on another website. And it reasoned about what to do!!

[Kelsey Petrich @krpetrich]

Before Solo, I’d constantly stop one project, cd into another folder, and start it up again. Not anymore with Solo!

[Aaron Francis @aarondfrancis]

I've been building Solo for several months, and it's out today!

It's a desktop app that manages your entire dev stack.

Add a project, Solo detects your processes, and you can start everything with one click.

It's free! Try it today: https://t.co/804mFNwGV4

[Kelsey Petrich @krpetrich]

@adcock_brett https://t.co/E2UFTskziH

[Kelsey Petrich @krpetrich]

For fun, I tried to get close to a generalized browser agent for this challenge.

So far it’s working with planning → reasoning → primitive browser actions, without any hardcoded knowledge of the specific pop-ups/traps in the challenge.

Video below is sped up for brevity. If I keep going, the next improvements are speed + better training + robustness.

cc: @adcock_brett

[Kelsey Petrich @krpetrich]

@aarondfrancis @claudeai @saradietschy @allenwalton @RoxanaCodes So fun!!

[Kelsey Petrich @krpetrich]

Respectfully, as a daughter, the highest calling is your dad texting “my freezer door bracket broke” and you immediately start measuring things like a NASA engineer.

And now he has a replacement bracket.

[Josh Pigford @Shpigford]

there is no higher calling in life than when your wife asks you if you can 3D print something for her

[Kelsey Petrich @krpetrich]

@adamwathan https://t.co/tvk0RCQsLu

[Kelsey Petrich @krpetrich]

@slaFFik @aarondfrancis Hey - We just enabled iDEAL as a payment method. Hope that helps!

[Aaron Francis @aarondfrancis]

They say AI is the end for developers.
They say the job market is gone.
They say learning to code is pointless.

I don’t buy it.

I think we're just getting started.

[Kelsey Petrich @krpetrich]

@TylerCLaprade @aarondfrancis That's a great idea! Are you creating any Sentry issues manually or just responding to issues Sentry automatically creates?

[Kelsey Petrich @krpetrich]

After this quick back-and-forth with @aarondfrancis at 10am this morning from a bug I DM’d him, I had an “aha”. I need to hook my bug reports and change requests into his agent loop, not his inbox.