Shubham Bajaj @bajajcodes

solutions engineer @Vapi_AI Joined July 2021

Tweets

331
Followers

130
Following

151
Likes

507

Amanda @hey_amandam

2 weeks ago

Thanks to @FarzaTV for the build inspo today. Here's my version using @Vapi_AI , a native swift app, leveraging browser automation and macos controls. This was alot of fun!

Farza 🇵🇰🇺🇸 @FarzaTV

2 weeks ago

Watch me control my computer with just my voice. This is the future of operating systems. No hands. GPT-Realtime 2.0 is very, very underrated. Demo:

932 842 14K 3.7M 8K

19 11 225 27K 109

View Details

Shubham Bajaj @bajajcodes

5 months ago

basically context hotness > instruction completeness

0 0 0 102 0

View Details

Shubham Bajaj @bajajcodes

5 months ago

the counterintuitive thing is more files actually works better than fewer monolithic instructions degrade over the conversation but modular ones stay fresh when you read them at the right time

2 0 0 117 0

View Details

Shubham Bajaj @bajajcodes

5 months ago

so i've been building these claude code skills and finally figured out why long instruction files break down

1 0 1 145 1

View Details

🚀 Hiring: Customer Support Engineer at Vapi Debug production issues, help customers succeed with voice AI tech. Need: 1+ year dev experience Strong debugging skills Clear communication LLM experience (bonus) DM your resume + projects! #hiring #VoiceAI

3 1 12 6K 5

View Details

@

57 years ago

0 0 0 0 0

View Details

Vapi @Vapi_AI

10 months ago

At Vapi, we're obsessed with the quality of conversation. But the most human-like AI is useless if the line drops. In the latest blogpost we cover how we achieved 99.99% uptime while servicing millions of calls.

8 8 43 450K 10

View Details

Shubham Bajaj @bajajcodes

11 months ago

@MarketersMenu didn't understood. can you share what your trying to build and steps to reproduce?

1 0 1 26 0

View Details

Shubham Bajaj @bajajcodes

2 years ago

I know how to use Vocode action agents (actions) since yesterday, however as of this morning, I feel otherwise. Need to go deeply into the internal workings of Vocode's action agents. However, I must admit that these action agents are really strong.

1 0 2 174 0

View Details

Shubham Bajaj @bajajcodes

11 months ago

@MarketersMenu can you share more context?

2 0 1 31 0

View Details

Vapi @Vapi_AI

12 months ago

We’re hosting an AMA on X Spaces this Monday at 5PM PST. Join us for a live Q&A on building with Vapi, scaling voice infrastructure, and what’s next in voice-powered applications. If you’re building agents, tools, or just curious - tune in.

2 6 22 4K 5

View Details

Vapi @Vapi_AI

12 months ago

📞 Launched today: Vapi Outbound Campaigns Outbound calling has been one of the top Vapi use cases since launch. Teams were building their own workflows with CSVs, custom queues, and retry logic. So we productized it.

5 4 53 4K 26

View Details

Vapi @Vapi_AI

12 months ago

Introducing Vapi integration with @Inworld_AI. TTS‑1 is Inworld's first text-to-speech model, built for real-time latency, high speaker similarity, and multilingual support. Now available as a voice provider in Vapi. Try it today in Vapi dashboard.

2 9 32 2K 9

View Details

Vapi @Vapi_AI

a year ago

Last night's YC AI Startup School afterparty was absolutely incredible. 150+ AI founders and builders packed into Vapi HQ in San Francisco for an evening that perfectly captured why this community is so special. Full Jordan interview recording in the thread 👇

2 10 30 3K 5

View Details

Nikhil Gupta @nikhilro_

2 years ago

news to share friends- we've raised 20m from bessemer. it's been an incredible year. voice ai has become real. last year, when we started, it seemed it would take at least a couple years to feel like you're talking to a person but like everything in AI, things have accelerated. it feels really beautiful and exciting to imagine in a couple years we'll manifest our dreams with our words. i'm so curious how kids born post-voice AI will feel. what bright futures they will create? as for us, there's lots to be done. this revolution needs a better platform. models are magic, in the same sense that ibm mainframes and punched cards were magic. last 50 years have shown, reducing the schelp to the magic, that's the unlock to changing the world. our developers want: * better uptime * reliable agents * faster development cycles uptime: there's only a couple hundred people in the world who know how to scale real-time distributed systems. it's the hardest freakin systems problem. reliable agents: love o1 but that's paradigm for seconds. we operate in ms domain, real-time reasoning. distillation gets you far but not enough, lot more juice to extract. faster development cycles: telephony was invented before the internet. integration into these systems is archaic. that's even before you think about development process of agents. so much to build and everything feels unknown and risky, but someone wise once said we do these things not because these are easy, but because they are hard. there's a 3T customer service industry waiting for us. let's fucking go! ps- we're HIRING. DMs open. we serve 100,000+ developers making millions of calls every week.