Inngest helps developers build their back-end workflows, raises $3M
Reading Time: 3 minutesInngest, an open source startup that helps developers build and manage serverless queues, background jobs and workflows, today announced that it has raised a $3 million seed round led by GGV.
Co-founded in 2021 by former Buffer CTO Dan Farrelly and Tony Holdstock-Brown, a former Docker engineer and the former head of engineering at healthcare company Uniform Teeth, Inngest sits on top of the company’s innovation in the queuing space.
‘Realistically, there’s been almost no innovation in the queueing space,’ Inngest CEO Holdstock-Brown said. ‘Yet queues and events are foundationally very similar: You want to send a message and run something in the future. So for some reason, there’s been a lack of innovation around the queueing space.’ Yet, he noted, there has been a lot of work around events and event-driven architecture in recent years, especially around NATS, Kafka and real-time databases like ClickHouse. ‘We’re able to take all of these technologies and merge them into a nice way to create something that handles events, queues, function state and serverless — which is new, novel and relatively rare,’ he noted.
With Inngest, developers can, for example, write serverless queues for TypeScript to manage a checkout workflow. Developers don’t have to configure anything. They simply send a ping to Inngest when an event happens and then the service will kick off the function you’ve defined for this event. Inngest will automatically call this function and handle retries if anything goes awry.
Developers can also take their existing TypeScript or JavaScript codebase and then have Inngest invoke functions to handle background tasks. With this, for example, developers can offload sending a welcome email for a new user from their API endpoint and have Inngest handle it in the background.
All of this, the team noted, also means that developers can now easily build stateful applications with serverless functions.
‘If you take a look at how people were building workflows, or tying together products, or webhooks, or building with queues — even if you’re doing things on servers — you still have to mess with queueing configurations, retries, concurrency, you have to have observability, so that you can manage your queues. And all of that disappears within Inngest. You just care about the code, the functions that you have to write and everything else is handled for you automatically,’ explained Holdstock-Brown.
In a way, applications written for platforms like Vercel are the ideal use case for Inngest. ‘We’ve actually seen a number of people — whether it’s really small companies or actually really big companies that you probably use multiple times every week, switching to places like Vercel and writing serverless functions and workflows that are now reliable on infrastructure that’s new.’
Inngest CTO Farrelly also noted that developers have been looking for an experience like this for a long time. ‘That developer workflow that these modern platforms promise is one of the things that allows developers to say: there should be a better way for x. And that’s where we come in. What we’ve been able to do so far is find a lot of developers and turn them into customers, turn them into internal evangelists and then spread throughout their companies to more use cases.’
One of those use cases, which the team didn’t even consider when it first built Inngest, is around large language models. These models need the tooling to handle chaining, retries and persisting state — and those are exactly the kinds of workflows that Inngest promises to handle for them. ‘We give you function state out of the box. We automatically give you history. So whether it’s chaining things or managing LLM function calls or the plugins that they have, we handle all of that,’ explained Holdstock-Brown. Tools like Inngest, he argues, make it easier for these companies to move their models into production.
‘Developers have adopted event-driven programming models and workflow orchestration systems to tackle this problem,’ explained GGV’s Glenn Solomon and Dan Cahana. ‘But doing so requires setting up and maintaining a bunch of new infrastructure, even for use cases as simple as background jobs or ensuring the reliability of third-party APIs. The ultimate goal of separating business logic from individual services often requires the attention of full teams to build, debug and maintain an event queue, various serverless functions and another database.’
Ref: techcrunch
MediaDownloader.net -> Free Online Video Downloader, Download Any Video From YouTube, VK, Vimeo, Twitter, Twitch, Tumblr, Tiktok, Telegram, TED, Streamable, Soundcloud, Snapchat, Share, Rumble, Reddit, PuhuTV, Pinterest, Periscope, Ok.ru, MxTakatak, Mixcloud, Mashable, LinkedIn, Likee, Kwai, Izlesene, Instagram, Imgur, IMDB, Ifunny, Gaana, Flickr, Febspot, Facebook, ESPN, Douyin, Dailymotion, Buzzfeed, BluTV, Blogger, Bitchute, Bilibili, Bandcamp, Akıllı, 9GAG