Launch HN: Vela (YC W26) – AI for complex scheduling
Scheduling is a constraint satisfaction problem disguised as email! It’s easy when it’s two people, one timezone, one channel. But it becomes a constraint satisfaction problem when inputs are unstructured natural language across multiple communication channels, constraints change mid-solve, and the objective function includes social dynamics that don't exist formally anywhere.
What if scheduling just happened? For example: a recruiter sends one message, and every interview across five candidates, three hiring managers, and two time zones gets booked, confirmed, and updated automatically. No links, no back-and-forth, no one spending hours with 20 emails. Everyone just gets the right invite at the right time, on whatever channel they actually use. That's what we built Vela to do.
You loop in Vela into your emails, SMS, WhatsApp, Slack, phone or integrate into an ATS etc and it takes over: reads context, checks calendars, proposes times, follows up when people ghost, and rebooks when things shift.
One of our first customers is a staffing firm that searched for a scheduling solution for almost eight years. Their coordinators manage hundreds of candidate-client interviews where each side needs separate email threads, separate Zoom accounts to avoid double-booking links, and calendar invites connecting parties who never directly communicate. A client reschedules one interview and it cascades into four others. A candidate responds on SMS to a thread that started on email. Vela solved this in just 10 minutes of onboarding.
The hardest part has been the data problem. Scheduling behavior varies enormously across populations. C-suite folks respond to email within hours and expect formal 3-option proposals. Truck drivers applying for logistics roles respond to SMS at odd hours from shared devices with "y tm wrks." The failure mode isn't parsing -- it's applying the wrong interaction pattern for the wrong segment and watching the conversation die. We've been building behavioral datasets from thousands of real interactions: response latency by role, channel preference by demographic, follow-up timing curves, how many options to propose before you hit decision paralysis. This data doesn't exist anywhere.
The core agent challenge is state across channels. When someone responds on SMS to a thread that started in email, Vela needs to unify identity, merge context, and continue without losing information. Phone numbers don't map cleanly to emails, people use nicknames on text, shared devices mean the responder might not be who you reached out to. Temporal NLU is its own problem -- "next Friday" means different things on Monday versus Thursday. We extract structured constraints from natural language and resolve against calendar state. When ambiguity can't be resolved, Vela asks -- but deciding when to ask versus infer depends on the stakes of getting it wrong.
We're live with paying enterprise customers and every client still surfaces edge cases that surprise us. Case studies on our site (https://tryvela.ai/case-studies/). You can check out a demo here: https://www.youtube.com/watch?v=MzUOjSG5Uvw.
We'd love feedback from anyone who's worked on multi-agent coordination, conversational AI across channels, or constraint satisfaction in messy real-world domains. Looking forward to your comments!
23 comments
[ 3.1 ms ] story [ 46.6 ms ] threadMy very strong advice would be to pick one of these use cases and niche hard. Multi channel, multi party scheduling isnt a problem anyone thinks they have (even if they actually do). They wake up thinking they have 40 truck driver shifts to fill tomorrow.
Deputy cleaned up by going after rota scheduling for independent coffee shops. Logistics sounds like a great shout. Each have messy edge cases which you can develop a strong solution around but you'll get crushed trying to go horizontal in this space. Best of luck!
Created a problem statement and then solved it with Gurobi, repo here: (https://github.com/aleda145/interview-scheduling-kontaktsamt...)
Agents feel like the perfect fit for the whole rescheduling loop that happens in the real world!
Have you had to use an optimization solver yet? If so, which one?
edit: after reading a bit more of description looks like yall are taking a similar approach, kudos!
[0] https://github.com/r33drichards/minizinc-mcp
[1] https://github.com/r33drichards/bocce-scheduler
generally when i give someone my calendar link, i'm pretty happy for them to just choose whatever time within those constraints. i like the future where everyone opts in ("i will meet as long as my preferences are considered") & there doesn't need to be any manual clicking/coordination whatsoever.
as a tidbit of feedback: are you explicitly targeting b2b? i would like to just sign up, but i'll book a demo if that's the only option :)
How is this better than spending 2-5 mins making a poll and letting people vote?
https://doodle.com has been around forever and doesn’t cost anything.
A lot of similar solutions came up in the early chatbot era, when Facebook published Ducking and it became trivial to parse dates from natural language. I also looked into building such a product in the time, but ultimately found it hard to find an entry to the market: Most people that actually need something like this do have secretaries (who will also schedule a lot of other things in regards to the meeting) and most other people that have a less severe form of that problem rarely want to actually pay for such a product.
[0]: https://claralabs.com
E.g. I'm friends with so-and-so and I don't want to be a jerk and schedule a 4:30PM Friday meeting (regardless of whether this is sensible, it's the reality). Or, I see continuous blocks of meetings on someone's calendar with only one open slot, presumably where they'll eat lunch; I shouldn't take that slot. Except for that one guy who I know doesn't eat lunch. Alternatively, I'm getting on a flight at 3PM and working the last 2 hours from a plane; I haven't actually blocked by calendar (people are lazy), so I can't actually do meetings then. Or, I know there's a conflict but someone told me to book over it.
You can go on with the "hidden context". Perhaps this works in some industries where calendars can be trusted, but I've always found the "hidden preferences" to make scheduling optimization essentially impossible. How do you know, for example, when it's okay to reschedule a meeting? How do you say, well, if X person can't be there but Y can, it's okay, but ideally they'd both be there, but not if we have to move it further than a week out from today, then it's fine; but I'll check with X and Y anyway on that?
And each industry differs with this respect. So we collect the data and generalize it across our customers if they fit the group: all while giving them enough flexibility to schedule their own individual way.
This is why we believe it is important to have a self-learning and adapting system as well.
Vella" (वेल्ला) is a North Indian/Delhi slang term for someone who is lazy, jobless, or has nothing productive to do (an idler or couch potato). Often used in Hinglish, it describes a person wasting time, similar to the word "loser" or someone with no work.
But it also allows our customers to be lazy :).
The staffing use case makes a lot of sense as a wedge. 1000+ interviews a week across SMS and email is exactly the kind of workflow where coordinators are drowning and no one's built the right tool yet. Good luck with it.