calljmp

Build AI agents as code in your app.

[Vlad Lykhonis @vladlykhonis]

Develop prompts independently from AI agent code. Iterate, evaluate, and share. AI agent automatically picks up the version you select.

Prompt studio here is our first iteration, many more advance features are coming soon.

[Vlad Lykhonis @vladlykhonis]

Spin up an agent under 5 minutes with TypeScript. LLM, durable execution, observability, web browsing, datasets (RAG), secrets, OpenAI, Slack integrations, and much more.

[Vlad Lykhonis @vladlykhonis]

New landing page. Had some fun with shaders, connecting energy, water, and digital world.

[Vlad Lykhonis @vladlykhonis]

Just published and available on Calljmp AI agentic runtime. My favorite addition is REALTIME.

Your agent is now a cloud instance with own memory and bi-directional realtime communication with your web apps.

https://t.co/KQC2PXUW6u

1. Short term memory for LLM with automatic management of history - perfect for AI chats/copilots/etc.
2. Live events / realtime - call your agent's functions realtime, publish live events when handling tasks in agent
3. Web SDK - seamless integration into frontend web apps

More advance and essential features are in the pipeline already!

[Vlad Lykhonis @vladlykhonis]

Revamp of UI and UX to bring better usability and upcoming AI metrics.

Running RAG and Web scraping alongside OpenAI and out-of-box LLM tooling with workflow orchestration.

[Vlad Lykhonis @vladlykhonis]

Augmented retrieval is coming to agents AI runtime

[Vlad Lykhonis @vladlykhonis]

Build AI agents with AI runtime in TypeScript. Deploy and share a portal with multiple agents for your team or your customers.

People can easily launch and talk to agents in friendly user interface.

[Vlad Lykhonis @vladlykhonis]

Have been working on adding CLI support for agentic AI workflows.

Right now infrastructure is up to be able to define phases (read steps), retries, auto caching per phase, and parallel execution.

Realtime observability and 30 day retention of all traces and logs.

[Vlad Lykhonis @vladlykhonis]

Also write code your app and see traces right in @expo logs without leaving the app

[Vlad Lykhonis @vladlykhonis]

Realtime observability from AI agent stages, to user define traces right from your code in React Web and Native app.

[Vlad Lykhonis @vladlykhonis]

AI agents and workflows right in your NextJS and React Native app's code.

- Create type-safe multi-step agentic workflows right in your app
- On edge, managed for you on demand
- Use any model or data source you like with secrets stored securely
- Observability with logs and realtime traces
- Fully featured BaaS available to support your apps

Beta is coming out soon. @calljmpdev

[Vlad Lykhonis @vladlykhonis]

Adding observability to AI workflows and agents in @calljmpdev. From breakdown of each step to live visualization of running workflows.

Here we write TypeScript and implement our steps (operators) right inside React app (Web/Native). After invoking a workflow and running it, we observe logs appearing on a dashboard with the name of the workflow, and each step.

Few interesting points to pay attention:
- Operators gets assigned a name (see variable name), e.g. Merge results.
- Console log/warn/error work out of the box. Anything you print in operator's code, will appear in logs on a dashboard.
- Inputs, outputs of each step are observable on a dashboard.

Live preview is next big thing.

[Vlad Lykhonis @vladlykhonis]

Your react app is now AI agent!

- Write workflow as a code directly in the app (@reactjs and @reactnative)
- Your code is executed as you described on edge (@Cloudflare)

Parallel and branching flows are also supported. Realtime observability and polling via SSE are coming to @calljmpdev soon.

[Vlad Lykhonis @vladlykhonis]

AI sdk from @vercel on Cloudflare edge functions seems like a perfect fit.

With trends on AI and React Native synergies there is no better time or way to do this in few seconds.

With @calljmpdev you can easily enable AI capabilities in your mobile app completely safe with few dozens of code lines.

AI sdk also nicely integrates with @expo built mobile apps!

https://t.co/iSRea8MeE8

[calljmp @calljmpdev]

Calljmp - The Mobile-First Backend with Realtime, Auth, and Storage Built In

Start building for free → https://t.co/zMmPnx4TIL

[calljmp @calljmpdev]

@em We definitely should add passkeys as an option for accessing accounts for developers.

[calljmp @calljmpdev]

Realtime for arbitrarily defined topics and SQLite are coming!

[Vlad Lykhonis @vladlykhonis]

Realtime is coming to Calljmp!

Starting with pub/sub system:
- 3-part dot notation (e.g. https://t.co/kMojLE8S5S)
- Fields projections - receive only fields you need
- Filters - logic tree object filter format similar to MongoDB

Next is SQLite select queries converted into filters and applied to database related changes.

[Vlad Lykhonis @vladlykhonis]

New release for Calljmp - your mobile backend with realtime, SQLite, and edge functions.

Flutter SDK v0.0.6
• Realtime database & pub/sub with .observe()
• Events & field filters
• WebSocket support + auto-reconnect

Service Library v0.0.4
• Build secure TypeScript cloud functions
• No API keys
• Fully compatible with Cloudflare Workers
• SQLite support with type-safe access

CLI v0.1.2
• Fast local dev, deploy, and secrets handling

@flutterdev @CloudflareDev

[Vlad Lykhonis @vladlykhonis]

We just launched realtime for mobile apps - no WebSocket setup, no backend to manage.

@calljmpdev is a backend for @reactnative & @FlutterDev. Auth, SQL, file storage, and now: realtime pub/sub + database change streams.

Powered by @Cloudflare Durable Objects, built to scale.

https://t.co/bGcOjepA51