SmythOS Product Updates
Update 6-24-2024
Claude 3.5 Sonnet has arrived!
The world’s newest best model. Sonnet is outperforming GPT4o in many tasks, including context length, image recognition, writing and coding.
Also several bug fixes and QA Improvements!
One outstanding feature is that we added improved automated JSON post processing, when the LLM fails to follow your prompt. This improves robustness.👏
To use Sonnet, add your Claude API key to the vault!
Update 6-19-2024
Oh boy – new integrations for you to make money with! ❤️
- Ahrefs
- Webflow
- Google Chat
Do you know what you’re going to build next? 
Update 6-14-2024
Overall, we have some big things cooking that will take some time, and the wait will be worth it. We have spent a lot of our energy on quality improvements to existing systems as well.
Models
- Removed non-working Together.ai models
- Updated Together.ai models to latest versions
Discord
Lots of changes are visible, as we are creating new channels and updating our existing ones. Thank you @SmythOS | Franklin 😊
Fixes
- Chatbot embodiment not following the system instructions properly
- Input variables not working on APICall url
- ForEach Loop not handling empty arrays
- Fixed Vault key usage in APICall component
- Fixed an issue where Chat debugging got stuck
Sentinel updates
Sentinel is our AI agent that helps us build SmythOS. Sentinel already created several features by itself, and keeps getting smarter. This will help us bring more improvements to you faster. Sentinel can now autonomously:
- Read project management tickets
- Analyze SmythOS code
- Suggest fixes
New agent template
Transcript Analyzer is 💰💰💰 – Use this for your business. We are!!
Transcript Analyzer in action 🚀🚀🚀
Update 6-10-2024
Microsoft has landed! Also new tools for SEO, news and beyond!
New integrations
- Microsoft Calendar
- Microsoft Teams
- PdfCrowd
- UUID Tools
- Keywords Everywhere
- Inflow Inventory
- Newsapi .org
Update 6-5-2024
Welcoming two new team members!
@Sebastian Schieke joins SmythOS to build out our course material and certification program!
He runs a podcast and several mentorship & coaching programs, and as such has many years experience with creating educational content.
Our immediate goal is to create a beginner-friendly course that takes only ~50 minutes to complete and gets you to expertise and certification.
@SmythOS | Franklin joins SmythOS as a new Discord administrator. He will lead our moderators, help drive engagement, improve our channels, automations, support and more. Get ready for an even more vibrant.
Daily Live Training!
We keep investing in our users and community! We have an entire month worth of daily (M-F) live training sessions with Q&A ready for you! All that training is completely free and open to everyone, because we believe the AI agent revolution should not be limited to the few, but to all.
Check the “events” section in discord. All livestreams are recorded and are stored on YouTube for on-demand viewing. Every day we cover new topics, so it will stay interesting, even if you are a pro!
Personal favorite
Agents now have avatars! I’ve been waiting for this one for a while
Give them new skills, tools, put them to work with schedules, review their work log. Move over, upwork!
Update 6-3-2024
Prod deploy without registering a subdomain 
MAJOR: You can now deploy agents to prod without needing a subdomain! This makes it easy to use scheduler & bulk runs in agent settings!
Agent Settings Update
Why you should care: you can now inspect and download logs.
- Logs widget added to the agent settings screen – MAJOR
- Logs call tree view (alpha) – MAJOR
- Agent avatars upload in agent settings page
Gemini pro and Gemini Flash
Why you should care: Gemini Flash is the new king of affordable smart models, with multi-modal capabilities and huge context lengths of 1 Million tokens!
Bug fixes
- Agent variable fix for file drop
- bug fix and multiple function call support for toolUseLLMStreamRequest
- Improve the title generation function
- Hugging Face multiple fixes due to API changes
- Modify Agent Capabilities widget to use latest prod version
- Job scheduler widget UI fixes
- Fix empty array causing ForEach component to fail
New Together.ai models
- Llama-3-8b-chat-hf
- Llama-3-70b-chat-hf
- Mistral-7B-Instruct-v0.3
- Mixtral-8x22B-Instruct-v0.1
- Nous-Hermes-2-Mistral-7B-DPO
- Nous-Hermes-2-Mixtral-8x7B-DPO
- Nous-Hermes-2-Mixtral-8x7B-SFT
- Nous-Hermes-2-Yi-34B
- OLMo-7B-Instruct
- OLMo-7B-Twin-2T
- Qwen1.5-0.5B-Chat
- Qwen1.5-1.8B-Chat
- Qwen1.5-4B-Chat
- Qwen1.5-7B-Chat
- Qwen1.5-14B-Chat
- Qwen1.5-32B-Chat
- Qwen1.5-72B-Chat
- Qwen1.5-110B-Chat
- dolphin-2.5-mixtral-8x7b
- dbrx-instruct
- deepseek-coder-33b-instruct
- deepseek-llm-67b-chat
- ggemma-2b-it
- gemma-7b-it
- Snorkel-Mistral-PairRM-DPO
- Ssnowflake-arctic-instruct
- UToppy-M-7B
- WizardLM-13B-V1.2
Update 5-30-2024
You know how models like Claude AI can sometimes be annoying and say “here is your json: { … }” instead of just returning JSON? Annoying right? Well, we now added a post-processing step to clean that up, increasing your successful outcomes.
- Optimized Claude JSON output : now Claude models generate much more accurate JSON responses
- Claude is now available for Chatbot settings
We also made several updates to sub-agent usage, for swarm intelligence.
- SubAgents : you can now select to run the sub agent in dev mode, prod mode or same as parent.
- SubAgents can now use claude
And released several bugfixes!
- Fix for builder UI freeze in debug mode when scraping some websites using scrapingbee
- Multiple fixes for strong data typing
- Multiple fixes for chatbot : responsiveness, crash when OpenAI returns LaTeX content, …
Update 5-29-2024
Another heavy hitter joins the SmythOS team full-time
We are pleased to welcome the world’s likely best Chief Marketing Officer, Michael Walker II as the CMO of SmythOS. Let’s welcome him! https://www.linkedin.com/feed/update/urn:li:activity:7201568717976047616/
Agents now gain Work Schedules
To try this out, make sure you’re on a paid plan and have an AI agent in production. See screenshots below. You guys, this is an incredible productivity boost – AI agents can now act like colleagues with work schedules! Mind .. blown!
- Imagine an agent that reads Reddit every morning and sends you an email summary.
- Imagine an agent that checks Asana tickets daily and tells you what needs attention.
- Imagine an agent that can build an SEO report every month.
Bulk Work Upgrade
So our previous release of bulk work upgrade ran bulk work in the browser. We’ve now rolled out server-background work. You can start a job, leave SmythOS and come back later to get all your results. Access via settings gear above your agent. Ensure you’re in a paid plan for bulk work.
- Imagine uploading a list of 100 keywords for an SEO article writer.
- Imagine uploading a list of 5000 prospects to create personalized marketing content for & web research.
Update 5-21-2024
Improved debug! Now with search & copy button, saving your even more time!
Update 5-20-2024
New Google integrations
We are excited to announce that we have integrated a comprehensive suite of Google services directly into our platform. This means you can now orchestrate your workflows seamlessly without the need for a Zapier paid subscription. 
Gmail
- Create Draft
- Delete Draft
- Get Draft
- List Drafts
- Update Draft
- Get Message
- List Messages
- Delete Message
- Move to Trash
- Send Draft
- Send Message
Google Analytics
- Run Report
- Run Real Time Report
- Batch Run Reports
Google Calendar
- Create Event
- Get Event
- List Events
- Patch Update Event
Google Docs
- Create Document
- Get Document
- Batch Update
Google Drive
- Create Drive
- Get Drive
- List Drives
- Delete Drive
- List Files
- Create File
- Get File
- Generate IDs
Google Sheets
- Create Spreadsheet
- Get Spreadsheet
- Spreadsheet Values
- Append Values
- Get Values
- Update Values
- Batch Update
- Batch Get
Google Translate
- Translate Text
- Language Code Validator
Impact and Value
With these new integrations, you can streamline and automate your business processes more effectively. The direct integration of Google services enables you to:
- Enhance Productivity: Automate repetitive tasks across Gmail, Google Calendar, Google Docs, Google Drive, and Google Sheets.
- Improve Data Insights: Run detailed reports in Google Analytics to gain real-time insights into your business performance.
- Simplify Document Management: Seamlessly create, update, and manage documents and files in Google Drive and Google Docs.
- Boost Collaboration: Easily schedule and manage events in Google Calendar, ensuring that your team stays on the same page.
- Facilitate Multilingual Communication: Use Google Translate to overcome language barriers in your communications.
Post-Script
These are a lot of new integrations, we’ll be adding documentation & tutorials soon!(edited)
Update 5-14-2024
We now support GPT 4-o! Time to build the future!
- Vision Model Component
- LLM Component
- AI Assistant Component
Enjoy 50% lower costs and 2x speed compared to GPT 4 Turbo!
Speaking on a stage previously shared by E. Roosevelt and Obama at the Harvard Club in Boston is a sincere honor. We’re partnering with AWS and the AI/R group to accelerate human potential!
Tune in at 1:30pm Boston time today! (1.5 hr from now)
https://lnkd.in/gKxPXPXq
Update 5-13-2024
The future is even closer. What an exciting time to be alive!!! 😎 OpenAI delivered. Certified amazing. This is the biggest update since the original ChatGPT. I’ll give you my hot take! Here’s what to expect:
- GPT Colleague joining your voice meetings
- GPT Making phone calls
- GPT Working alongside you on desktop and phone
- Meta Style glasses formfactor is coming. This is now a KILLER formfactor. I now 100% believe that this is what Apple and OpenAI is working on. Wild!
- Next Gen Brand Agents. This representing my company with clients.
- People getting emotionally attached to their AI
- Multimodal is so 2023. Omnimodal is the next open source frontier.
- Kids now have a personal tutor.
We’ll of course bring this new API to SmythOS as soon as possible.
Build the future with GPT-4o and SmythOS.com🙌
Update 5-6-2024
New embodiment: Postman!
This one is a big one!!! Makes building and integrating SmythOS into your existing projects super easy!
- Drag & drop your postman collection into SmythOS – we’ll auto-create your API components!
- Export your SmythOS API Endpoints to a postman collection for easy integration into your project!
Improving the User Experience
SmythOS just keeps getting better! We have obsessive compulsive disorder to make our project easier and better to use! There are a host of amazing new tweaks that will add up to an even more JOYFUL agent building experience! Advanced Session Management: Going forward, when we push updates to our production server, it will no longer require our users to re-authenticate. Since we push a lot of updates, we thought that was annoying to our users, and well .. we fixed it 🙂 Often forget to disable debug mode? We gotchu covered! We now auto disable debug mode if you close the agent builder so you can return to a fresh session. Need for speed! Have you noticed how much faster SmythOS is to use? The updates all stack up! A brand new debug log design and functionality! You will all 💕 it! See screenshots attached. It is entirely focused on usability. Magic LLM Component Labeling! Now, your LLM components will rename themselves with magic. 🪄 Just bring your prompt!
Agent Settings Page
Have you tried the new agent settings page yet? Access it from the settings gear on the agents list. There is a new bulk run tool. We made some updates today. This is in alpha pre-release, so we’re actively looking for feedback!
- Form Call
- Bulk Call
Try it!
Update 5-3-2024
New Templates
- Bidirectional Messenger Agent
- LinkedIn Leads Scraper
23 New Component Integrations
- WordPress.com
- WordPress.org
- OpenRouter
- Perplexity AI
- AITable
Updated Templates
- News Researcher
- Content Keyphrase Analyzer
- GitHub Tutorial
Updated Components
- Airtable – List Tables
- PlayHT 1.0 – TTS
- PlayHT – Get Output
- Leia – Generate Animation
- Leia – Generate disparity map
- Leia – Get Pre-signed Url
- Firestore – Update Field Value
- Firestore – Get Document
- Firestore – Delete Document
- Firestore – Create Document
UI Tweaks
- Proper Error Messaging for subdomains management
- Fix blurred 403 page
- update icons for Logic Components
- Components Sidebar in the builder
- For Long Names of Component Integrations we have eclipes for overflow text and you can see full text with a tooltip
- Changed the request button color in the builder
- Only show offer banner for free users
- Updated descriptions for new pricing plans
- Fixed error messages for the roles pages
- Adjust position of component gear (Settings) button [top-right corner of component card]
- Relocation of Help and Delete Icons [top-right corner of component card]
- Change the delete button color to red
- Show the Gear (Settings) icon only for those component, that has configurable settings field
- Text Size Increase for Component Type Label [top-left corner of the component card]
- Show tooltip on hover for component’s short description, as some text are cut for some components
- Fix errors in /account-deleted page
Other Dev updates
- Enhanced Strong data typing and support for new data types.
- APIEndpoint Support for binary data (file upload) APIEndpoint
- Auto-generated titles for LLM Components
- Enhanced RSA data signing component
- Enhanced Components log viewer
Infrastructure updates
- Adding HTTP/3 + QUIC protocols support
- Implementing Node App clusters
- Updating deployment CI/CD pipelines
- Sessions stickiness : Avoids nav interruption on server update + better load balancing
- Increasing servers load capacity
Update 5-1-2024
SmythOS Announces Former OpenAI Trust & Safety Specialist Joining Leadership Team.
Houston, TX – [5/1/2024] – SmythOS, a GenAI app orchestration framework for rapid, secure and visual AI agent development, is thrilled to announce the appointment of Ishant Singh Chauhan, a former key member of OpenAI’s trust and safety team, as its new Director of Customer Success.
From 2021 to 2023, Ishant served as the AI Safety Specialist – Trust & Safety at OpenAI where he played a pivotal role in developing relationships with the first 2000+ clients.
During his tenure at OpenAI, Ishant played a pivotal role in advancing the safety and security frameworks of Applied AI technologies. His experience in automated safety solutions, product policies, risk evaluations is an invaluable asset to our organization.
Since exiting OpenAI, Ishant Singh gained further recognition as the CEO of an AI startup, an AI agent company he sold in March 2024. Business Insider ran a story on him and 18 other ex-OpenAI employees who became successful AI entrepreneurs.
Ishant had many options for his next career move. He chose SmythOS – in his words, because “SmythOS understands how to safely deliver on the promise of AI agents better than any other product in the market, with a vision that aligns with everything I know to be true, and a team that I have tracked since my days at OpenAI”.
At SmythOS, Ishant will lead the customer success team, focusing on delivering exceptional service and fostering long-term, sustainable partnerships.
His deep understanding of AI applications and unwavering commitment to customer safety will be invaluable as SmythOS continues to expand its reach and push the boundaries of AI innovation.
Welcome Ishant 
Update 4-30-2024
Asynchronous Components
The Async component is instrumental in orchestrating asynchronous operations, acting as a foundational block for managing processes that are not immediately completed.
Example use case: ChatGPT doesn’t want to wait longer than 15 seconds for your agent to complete its’ task, but the task requires a lot of operations. Now you can send a response while the Agent keeps working asynchronously.
This is one of our most requested features, and we’re happy to deliver!
https://app.smythos.com/doc/#/components/Async
Agent Settings Page (Alpha Pre-Release)
Access via the settings gear on your agent card.
Bulk Run Agents
Via the agent page, you can now run API endpoints in bulk, you can even bring a CSV.
Example use case: let an agent write 20 SEO articles in a sequence.
Update 4-25-2024
New integrations
We are excited to announce that we have integrated a comprehensive suite of Google services directly into our platform. This means you can now orchestrate your workflows seamlessly without the need for a Zapier paid subscription.
Stability AI
- Upscale Image ===> Image-to-Image Upscale
- Image Generator ===> Text-to-Image SD 1.6
- Image Masking ===> Inpaint
- Outpaint
- SD3 Text To Image
- SD3 Image To Image
- Stable Image Core
- SD3 Turbo Text To Image
- SD3 Turbo Image To Image
- Creative Upscale
- Get Creative Upscale Result
- Search and Replace
- Remove Background
- Image-to-Video
- Get Image-to-Video Result
Shopify
- Get Blogs
- Create Blog
- List Articles
- Get Article
- Create Article
- Create Article With Image
- Create Draft Article
ScrapingBee
- Basic: Scrape Web Page
- Advanced: Scrape Web Page
- Advanced: Google Search
- Basic: Google Search
HubSpot
- Create Contact
- List Contacts
- Get Contact
- List Deals
- Create Deal
- Get Deal
- List Tickets
- Create Ticket
- Get Ticket
- Next Contacts Page
- Next Deals Page
- Next Tickects Page
DataForSEO
- Live YouTube SERP Advanced
- Live Google Images
- Live Google SERP Advanced
- Live Google SERP Regular
- Live Google News Advanced
AvesAPI
- Google SERP
Asana
- Get Project
- Create Project
- Create Team Project
- Get Workspace projects
- Get Section Tasks
- Get Project Tasks
- Get Task
- Create Task
- Get Multiple Tasks
- Get Goal
- Get Multiple Goals
- Update Goal
- Get Section
- Get Project Sections
- Add Task To Section
- Update Section
Ayrshare
- Send Post
- Send Post With Media
- Get a Post
- Send Youtube Post
- Delete a Post
GitHub
- Get Commit ===> Get Branch Info
- Get Repo File Content
- Get Repo Directory Content
- Get All Repo Content
- Search Github
Update 4-24-2024
Models Update
New GPT 4 turbo model ‘gpt-4-turbo-2024-04-09’ under ‘gpt-4-turbo’ and ‘gpt-4-turbo-latest’. As it has vision capabilities, so we’ve added it to VisionLLM. OpenAI recommends now to use ‘gpt-4-turbo’ INSTEAD OF ‘gpt-4-vision-preview’.
We’ve released the new GPT-4 Turbo model ‘gpt-4-turbo-2024-04-09’, which is now available under both ‘gpt-4-turbo’ and ‘gpt-4-turbo-latest’. Since this model includes vision capabilities, we’ve integrated it into VisionLLM. OpenAI suggests using ‘gpt-4-turbo-2024-04-09’ instead of ‘gpt-4-vision-preview’ from now.
New GPT 3.5 turbo model ‘gpt-3.5-turbo-0125’ is now available under gpt-3.5-turbo and gpt-3.5-turbo-latest. ‘gpt-3.5-turbo-0125’ has higher accuracy at responding in requested formats and a fix for a bug which caused a text encoding issue for non-English language function calls.
We have a small distinction between ‘GPT-<version>-turbo-latest’ and ‘GPT-<version>-turbo.’ Essentially, ‘GPT-<version>-turbo-latest’ always offers the most recently released model, while ‘GPT-<version>-turbo’ provides the stable version of that model family. For instance, ‘gpt-4-turbo-latest’ will have the latest model, even if it’s a preview version. However, you will always receive the stable version with ‘gpt-4-turbo’, INDICATED BY the stable tag.
Added 3 new models from groq.com. Llama 3 is amazing 🤩
- Llama 3 8B
Speed – ~870 tokens/s
Context Length – 8K
Price per 1M Tokens (Input/Output) – $0.05/$0.10
- LLaMA 3 70b
Speed – ~280 tokens/s
Context Length – 8K
Price per 1M Tokens (Input/Output) – $0.59/$0.79
- LLaMA 2 70b
Speed – ~300 tokens/s
Context Length – 4K
Price per 1M Tokens (Input/Output) – $0.64/$0.80
Improvements
- Improved error messages team and role management
- Replaced error toasts with in-page errors for better UX
- Replaced text field with textarea for vault key and default value in the builder to preserve new lines
- Permission and Error Log fix in Data Pools Page
Template update
News researcher Template was updated with new features. Documentation and video update to follow.
Update 4-19-2024
We’re squeezing in one more mega update before the weekend! Enjoy!
Enhancements
- Support array/object input type
- When the ‘String’ input type is selected and passed binary file we will get the base64 Data URL
- When the ‘String’ input type is selected and passed date like ‘April 20, 2024’ we will get ‘2024-04-19T18:00:00.000Z’ (ISO 8601 standard)
- Introduce array and object data type for input
- Now we have two types of numbers!
- Number (can include floating point)
- Integer (without a fraction part)
UI/UX Improvements
- Proper Error Messaging for subdomains management
- Fix blurred 403 page
- Update icons for Logic Components
- Components Sidebar in the builder
- For Long Names of Component Integrations we have ellipses for overflow text and you can see full text with a tooltip
- Changed the request button color in the builder
- Only show offer banner for free users
- Updated descriptions for new pricing plans
- Fixed error messages for the roles pages
- Adjust position of component gear (Settings) button [top-right corner of component card]
- Relocation of Help and Delete Icons [top-right corner of component card]
- Change the delete button color to red
- Show the Gear (Settings) icon only for those component, that has configurable settings field
- Text Size Increase for Component Type Label [top-left corner of the component card]
- Show tooltip on hover for component’s short description, as some text are cut for some components
Bug fixes
- Fix errors in /account-deleted page
- Signing RSA Signature with private key that saved in the vault
- Read base64 data for binary input, which includes new line
- [API Endpoint] file upload with multipart/form-data content type
- [API Call] unexpected ‘Sign Out’ button in OAuth section
Update 4-18-2024
New Template: Mid Journey v6
Use Mid Journey version 6 in your agents!
https://smythos.com/templates/midjourney-agent/
New Offers
Our new offers are live: https://app.smythos.com/partners
- Lifetime accounts for video creators
- Support for students, women, veterans people from poor countries & non-profits that serve them.
4 ways you can make millions with SmythOS:
- Build AI solutions for others, charge consulting, managed services fees
- Build your own AI solutions and use internally, take over a market
- Build a SaaS and win a market by orchestrating the latest AI
- Resell SmythOS or become affiliate and make money promoting SmythOS
Update 4-17-2024
Major: Affiliates
We have a major update for our affiliates, and it’s raining financial opportunity! This week all the hard technical stuff is out of the way, so next week we can start building up assets for you to share SmythOS with the world.
- Affiliates now make 30% recurring commission on revenue, lifetime, no cap
- Your referred users will now get DOUBLE tasks on any plan they buy, up to 10,000 free tasks!
- Free lifetime review copies for everyone have now discontinued, which means users now have a reason to upgrade.
- Conversion optimization: new pricing page to boost sales https://smythos.com/pricing/
- We updated the affiliate management portal.
- If you want to promote SmythOS to your audience, sign up at affiliate.smythos.com
Web Updates
We have a new site menu on SmythOS.com and added dozens of new pages this week to the site and blog. Check out all the new content!
- Templates: https://smythos.com/templates/
- Integrations: https://smythos.com/integrations/
- Deploy anywhere: https://smythos.com/deploy-agents-anywhere/
- Customer wall of love: https://smythos.com/customer-reviews/
- Public documentation: https://smythos.com/documentation/#/
All the templates also have matching videos now and updated documentation entries!
Update 4-16-2024
Early Lifetime Review Copies – Last Day
Tomorrow, our partner page app.smythos.com/partners will end the free lifetime review copies.
New offers will be introduced for influencers and those in need, such as students & veterans.
If you or a friend still want access to a free lifetime review copy, today is the day you should contact your friend, or get your account upgraded. Talk to Janice, Chelle, or myself if you have questions.
We reserve the right to still make exceptions, but the general availability of free forever plans is ending.
Can you think of 5 friends that would love to have this offer? Contact them today, before it’s gone!
Update: we’ll honor everyone who has reviews in progress, and everyone we’ve talked to so far. Our intention is not to cut anyone off. Just DM our team and we’ll take care of you!
Update 4-11-2024
We’re preparing some larger updates for later this month, so this update is a maintenance one.
We paused on releasing new integrations this week in favor of revisiting all the ones we already created, and making them better.
Every integration is re-tested and updated for easier use. In particular, we created custom outputs.
We are also updating documentation for all the integrations, which is now 80% complete.
Updated Integrations
- Airtable – Create Records
- Airtable – Get Record
- Airtable – List Records
- Airtable – Update Record
- Algolia – Full Text Search
- PlayHT2.0 – Text to Speech (More customization in component settings)
- Dev.to – Publish Blog
- Discord – Send Message
- Elevenlabs – Text to Speech
- Fal.ai – Generate Animation (More customization in component settings)
- Fal.ai – Stable Diffusion XL Image Generation
- Fal.ai – Stable Video Diffusion (More customization in component settings)
- Fal.ai – Text to Image Generator
- Fal.ai – Whisper Speech to Text
- Github – Add Labels to Issues
- Github – Commit File
- Github – Count Open Issues
- Github – Create Branch
- Github – Create Pull Request
- Github – Fetch Commits Since Last Release
- Github – Fetch GitHub User
- Github – Get Issue
- Github – Get Issue Labels
- Google – Maps Geocode Address
- Google – Maps Get Nearby Places
- Google – Translate Language Code Validator
- Google – Translate Translate Text
- Linear – Create Linear Issue
- Linear – Create Team info
- MailChimp – Add Email to Mailing List
- Medium – Publish Blog
- Meilisearch – Add a Document
- Meilisearch – Add a single document
- Meilisearch – Add Documents
- Meilisearch – Delete a Document
- Meilisearch – Full Text Search
- Meilisearch – Update a Document
- Meilisearch – Update Synonyms
- Replicate – Answer Image Questions
- Replicate – Background Remover
- Replicate – Face Restoration
- Replicate – Face Swap
- Replicate – Image Super-Resolution and Restoration
- Replicate – Llama 2 Chat
- Replicate – Music Generator
- Replicate – Llama Text Generator
- Replicate – Stability AI Image Generation
- Replicate – Stability AI Text Generation
- Replicate – Upscale Image
- Resend – Send Email via Resend
- Signnow – Send E-Docs for Signature
- Slack – Send Message
- Stability AI – Image Generator
- Stability AI – Image Masking
- Stripe – Create Stripe Customer
- Stripe – Get Stripe Customer
- Stability AI – Upscale Image
- Supabase – Insert Row
- Supabase – Delete Row
- Supabase – Get Columns
- Supabase – Get Row
- Supabase – List Supabase Projects
- Supabase – Update Row
- Telegram – Send Audio with Telegram
- Telegram – Send Telegram Message
- Trello – Get an Action
- Trello – Update an Action
- Trello – Get the Board for an Action
- Trello – Get the Card for an Action
- Trello – Get the List for an Action
- Trello – Get a Board
- Trello – Create a Board
- Trello – Update a Board
- Trello – Create a List on a Board
- Trello – Create a Label on a Board
- Trello – Create a new Card
- Trello – Get a Card
- Trello – Update a Card
- Trello – Add a new comment to a Card
- Trello – Get a List
- Trello – Create a new List
- Trello – Update a List
- Twilio – Send SMS
- Twilio – Send WhatsApp Message
- Twilio – Create Webhook
- Upstash – Create Record
Template videos
We added a lot of template videos this week, as you may have noticed. This is because we added a section to our website for templates.
Check out: https://smythos.com/templates/
Fun fact, all the SEO content on these pages was written by an AI agent!
Fal.ai upgrade
Fal.ai just announced powerful v3 voice to text. SmythOS already supports it!
🌟 Big news from our labs! Our Speech to Text service now boasts a 105x speed factor at just $1.15/(1000 min), powered by Whisper v3 with an 8% WER. Fast, affordable, and efficient transcription is here!
See independant benchmarks by Artificial Analysis https://twitter.com/ArtificialAnlys/status/177846501571842478
Update 4-5-2024
New features
- Input and output sorting! For all you OCD people! 🙂
- OAuth2 support strengthened
- New Sleep function: add a wait time before running next step of agent
- New Timestamp function: get the server timestamp, useful for signing keys with API calls
- New component features added – this will allow us to make better pre-built integrations for you!
Update 4-3-2024
New features
Important: Strong typed inputs have arrived at SmythOS!
We now have support for enforcing certain input types, starting with:
- String
- Number
- Binary
The binary Input checkbox will be migrated automatically into the Binary type.
Improvements
- Updated cards info for “Free account upgrades” – https://app.smythos.dev/partners
- There is a new free plan option available!
- Addressed some issues of OAuth implementation for the API Call component
- Focus name field when add/edit popup open for component input/output endpoint
Bug Fixes
- Prompt of Classifier component
- Chatbot sessions log
- Selectbox glitch in the right sidebar (disappeared when trying to search from the list)
- Template variables (tags) disappearing when typing (under input fields)
Update 4-1-2024
Breaking News: Groq models are now supported in SmythOS
Get your api at wow.groq.com and add it to the Vault, then use the model in the LLM Component.
Groq inference is ridiculously cheap and up to 800 tokens per second, perfect for improving speed & latency.
We previously supported open source models from together.ai – however with the addition of Groq you will save even more money, and have much faster inference!
Happy creating!!
UI Updates
We added visual indicators to show the context window of different models.
Update 3-29-2024
OAuth Integration News
- Google Analytics (see attached image)
- New Demo video for Google Docs https://youtu.be/mOGNxcG5ADc
New Component Integrations
- Twilio – Send Whatsapp Message
- LeiaPix – Get acces token
- LeiaPix – Get pre-signed URL
- LeiaPix – Generate disparity map
- LeiaPix – Generate animation (from image)
- Telegram – Send message
- Telegram – Send audio with message
Update 3-28-2024
OAuth API Calls
We have released beta access to Open Auth API components. Different providers are free to implement OAuth in many different ways, so building a no-code solution that works universally, is near-impossible. That said, we worked for weeks on end to get as close as possible. We release this today in Beta as a way for developers to get early access and help us discover its limits. (Can save you $80/mo vs Zapier)
Pricing Launch
We have launched your plan page, find it from the user icon dropdown (top right). Here you can manage your plan, see your task usage and upgrade. https://app.smythos.com/my-plan We have also publically launched our pricing page https://smythos.com/pricing/ We have also publically launched our about us page https://smythos.com/about-us/
Lifetime Review Copies
Our early access free lifetime offers are quickly going away. Especially now that pricing has been launched, grab it while it’s there. It’s not coming back when it’s gone. Tell your friends, get them in discord. Once it’s gone it’s gone. https://app.smythos.com/partners
Bug Fixes
- Fixed an error that occurred when opening component settings after running component(s) in Safari.
- Resolved a UI display bug in the left sidebar when searching components in the Builder.
- Fixed an issue with saving agent variables for integrations.
- Prevented toggle (opens and closes) vault key dropdown when clicked anywhere on the screen after opening for the first time in key-value edit popup.
Enchancements
- Implemented parsing of multiple image sources from the given input for Vision LLM.
- Updated to allow full URLs inside template variables for API Calls.
- Fixed Agent Variables that were breaking the APIEndpoint in debug mode.
- Resolved frequent crashes of the chatbot.
Other
Made UI improvements, including adjusting the height of the top announcement bar and fixing some UI issues in the data source and domain page.
Update 3-24-2024
It’s now possible to create bi-directional conversations with people over SMS.
Use the LLM Assistant combined with the Twilio Component!
Update 3-21-2024
- We rolled out server upgrades last night. Everything went smooth!
- Transparent image maker template is temporarily under maintenance, will be back soon.
21 New Integrations
- Stripe create customer
- Stripe get customer
- Trello get an action
- Trello get a board
- Trello get a card
- Trello get a list
- Trello create a board
- Trello create list
- Trello create board list
- Trello create board label
- Trello update an action
- Trello update a card
- Trello Get the board for an action
- Trello Get the card for an action
- Trello Get the list for an action
- Trello update a board
- Trello add new comment to a board
- Trello update a list
- Stability.ai image generator
- Signnow send e-docs for signature
- Sendgrid send dynamic email
- Sendgrid static email
OAuth Integrations
We have completed work on OAuth component integrations and will soon start releasing such integrations. OAuth is the way you authenticate access to services like gmail and google sheets. Until now, you needed a $80/mo Zapier account to easily orchestrate those kinds of actions. Very soon we’ll save our users that cost and natively support such integrations.
Partner Page
We launched a partner page at https://app.smythos.com/partners. This contains early access and launch promos. Take advantage while it’s there. We’ll be adding an affiliate referral program here as well soon, so you can make money by sharing the news of SmythOS. Stand by for updates.
New Homepage
We launched our new homepage.
https://smythos.com/ – We hope you like it!
Update 3-15-2024
Core components updated
- Vision Component now supports Claude Haiku
- AI Assistant now supports Claude Haiku
- Classifier now supports Claude Haiku
- LLM Component now supports Claude Haiku
Legacy Model Indicators
- Now indicates if models are legacy or new
- Everyone is recommended to replace legacy models with new models in their agents
Global Variables for Agents
Did you ever wish you could set variables upfront and then re-use them anywhere in any component? Now you can!
Default Values for Inputs
If you set default values, then the component can still run in case there is no input. Very handy!
UX improvements
There are several new UX improvements throughout the app that are designed to make the app easier for you to use. For example, when you deploy agents, we added a hyperlink for you to add a subdomain in case you don’t have any. It’s the little touches that matter, and we are adding those touches throughout the application.
3 New Templates
- Transparent Image Maker
- PDF to Text Converter
- Text to Audio Translator
Fun fact: transparent image maker was created in 5 minutes after we saw this video: https://youtu.be/vyEJpeeannI?si=T1LSArGPdN4L3MDu&t=487 This underscores our belief that anything anyone builds WITH GenAI, can be easily built with SmythOS.
24 New Component Integrations
- Google Translate
- Google Maps
- Replicate
- Linear
- Mailchimp
- SignNow
- Sendgrid
- Discord
Update 3-14-2024
Announcement
Today, Anthropic made their haiku model available via API. It is ridiculously affordable and quite smart! After much testing we discovered that it is capable of image recognition as well. 🥰
Haiku will be available in SmythOS on our very next production update. Stay tuned.
Update 3-12-2024
UI updates
- UI color palette unification update creates more consistency throughout the app.
- Several small touches add to the enjoyment of use. For example, the file input dialog makes it easy to select a file from PC for testing while debugging a binary input field for the API endpoint component. 🥰
New: Cryptography Components
- Hash can be used to generate hash functions : md5, sha1, sha256, sha512 … etc
- Encode/Decode is for data encoding, supported formats are : base64, hex, base64url, latin1
- Sign: supports both RSA and HMAC signatures
You can use these together with our integrations!
Update 3-8-2024
New templates and new Integrations!!
Big ones like Notion and Medium – this is for all you content creators!❤️
Using Claude
We have done a lot of testing with Claude’s new models. They are excellent!
Sonnet beats GPT 3.5T at 1/3rd of the cost.
Opus is a credible alternative to GPT 4.
Here are prompt guidelines unique for Claude: https://docs.anthropic.com/claude/docs/use-xml-tags
Update 3-7-2024
New Integrations
- Meleisearch – Full text search
- Meleisearch – Delete a document
- Meleisearch – Update a document
- Meleisearch – Add single document
- Meleisearch – Create an index
Update 3-6-2024
Add your keys to the vault
- Breaking News: We now support Google AI Gemini models
- Breaking News: Support for Claude 3.0 Sonnet and Opus
Core components updated
- Vision Component now supports Gemini and Claude
- AI Assistant now supports Gemini and Claude
- Classifier now supports Gemini and Claude
- LLM Component now supports Gemini and Claude
Templates have landed!
- Brand Agent
- Content Extractor
- Content Keyphrase Analyzer
- Content Writer
- Dropbox Folder Creator
- Hubspot Contacts Manager
- InFlow Inventory Agent
- Lead Contacts Scraper
- Meme Maker
- Multimodal Thinker
- News Researcher
- Product Description Writer
- Sentiment & Emotion Analyzer
- Shopify Blog Manager
More updates coming later today!
Update 3-4-2024
Last week, we added the mini map feature. What we didn’t tell you, is that our internal Sentinel Agent wrote, debugged and documented that feature.
This is the first instance where SmythOS has been building itself. The machine is building the machine.
And this is why I’m so excited to announce that we now have Github integration:
- Get issue
- Get issue labels
- Create branch
- Fetch user
- Create pull request
- Count open issues
Coming soon:
- Add labels
- Commit file
This way, you are able to build your own sentinel agents.
Update 2-29-2024
New Templates Released
- Basic Folder Creator (Dropbox)
- Sentiment and Emotion Analysis Agent
Update 2-28-2024
New integrations
We’ve added support for:
- Fal.ai Whisper audio to text
- Fal.ai Text to Image
- Fal.ai Stable Diffusion XL
- Fal.ai Generate Animation
- Fal.ai Stable Video Diffusion
Update 2-27-2024
Huge announcement – Integrations have begun rolling out!
We’re kicking off with:
- Easily send emails with Resend
- Text to speech with PlayHT
- Airtable: list, create, get and update records!
- Algolia: search
With Airtable, SmythOS now offers robust support for structured data with AI.
Update 2-26-2024
UX Updates
- Several new UI updates aimed at making the application more consistent and enjoyable to use.
- Left & Right sidebars are updated. You will see the most changes to the right sidebar. The UI is more clean, consistent & enjoyable.
- Mini map added
Update 2-23-2024
Several new Templates were added to SmythOS.
Have you checked them out yet?
From inventory management to contact scraping and more!
You can now link to external ECMAscript in the code component and import your files this way.
Update 2-21-2024
We plan to support Groq
Groq is a new type of AI chip that makes AI requests to open source models like LLama 2 and Mixtral 20 times faster (and much cheaper).
SmythOS already supports these models today via together.ai API – which has open API access, however Groq promises much faster speeds.
Why would you care? If you plan to make agents that need to work really fast, such as voice agents or real-time response, you’ll want the fastest possible LLM Components.
We applied for our own API key already, so if you want to try this with SmythOS as soon as it’s available, you should go to https://wow.groq.com/ and click “Request Access” under “Apply”.
New security testing features
We made it easier to test your OAuth security settings for your agents!
UI and UX updates
We made tweaks to the UI for components, opting for a more ‘flat’ design language.
This is a continued work in progress. We are prioritizing consistency, clarity and reliability.
Feedback is welcome, we hope you like the updates.
Additional updates
- Deployed a hotfix for agents listing and sorting (Agent Page)
- New image previews for multi-modal debugging!
- “Infinite” Workspace Canvas
- Click settings gear to open sidebar to avoid accidental right sidebar opening
- Larger Agent Setting behavior control makes it easier to type
Update 2-20-2024
Revamp of Components Sidebar
We want to bring dozens, if not hundreds of new components to our users for everything from email to databases, software integrations. To achieve this goal, we needed to create total revamp of our component sidebar, and a system for creating and maintaining new integrations without taxing our development team directly. We are now releasing the results of that multi-week effort.
Initially you will see a different layout with more structure, better search, and a request option. In the coming week you will start seeing new integrations, and that will continue all the time! That way, you can focus more on building agents and less time on reading API docs from third-party products!
TIP: Use the request button to request new integrations from our team.
Enchancements
- Improved Binary data support for API Call and API Endpoint
- Proxy Support for APICalls.
- Code component now supports up to 0.5M lines of code
- Code component now supports loading external scripts (experimental)
- Agents can now run multiple instances of the same sub-agent in parallel
Bring your component requests!
Analytics
Use analytics view to understand AI LLM model token usage per Agent!
Update 2-9-2024
Update from OpenAI
GPT 3.5 input tokens are now 50% cheaper !!
Since we don’t add to the cost of your API use, this immediately benefits our users!😎
Update 1-19-2024
Production Enhancements
We’re investing heavily in making our production environment the best it can be. Happy to announce major updates that will contribute to your productivity!
- Continuous Health Checks
- Proactive Issue Detection & Fix
- Data replication monitoring
- Daily backups
- Performance monitoring
Update 1-16-2024
Configure Your Own Storage Vector DB
Introducing ability to bring your own Pinecone Vector Database API key via configure storageThis allows you to own your data and scale to any size & service level you prefer.
Components
New
- Vision LLM Component: Talk to images with AI. Useful for visual analysis and multimodal chains! Idea: generate multiple images, have Vision LLM pick the best one. Idea: take a website screenshot, use Vision LLM to give visual feedback or scrape.
- LLM Assistant: A language model chat with memory of a conversation in a session. Unlike the LLM Component it remembers the conversation history.
Updated
- LLM Component now supports Anthropic Claude 2.1 and Claude-Instant 1.2. This is recommended for processing prompts with very long context lengths!
- LLM Component now supports dozens of Open Source models hosted on Together.ai – 3X faster and 6X lower cost inference
- Key Vault now supports these new models.
- LLM Component supports Advanced Settings for controlling Temperature, Maximum Tokens, Stop Sequences, Top P, Frequence & Presence Penalty.
- Bring your own API key for OpenAI to use GenAI without restrictions. Default mode restricted to 4096 input tokens. Enter in the Vault.
User Experience
- Added feedback buttons in all documentation pages so you can give precise feedback if anything is missing or confusing.
- Added feedback to the agent builder so you can easily give feedback, report bugs, and more.
Debug Experience
- Massive improvements to debugger and logging.
- Visual checkmarks next to components that succeed / fail.
- Warning boxes if components miss essentials, for example an API endpoint with missing endpoint configuration.
Stability
- Many backend improvements, including more encryption and hardening and scaling.
- Many fixes for SubAgents, delivering on Multi Agent Systems.
- Loop component received several fixes.
Hotfixes
- Fix for a debugger UI glitch
- Update to the LLM tokens limit messages to make it less confusing (we’ll add information about context window length in a next update)
- We also applied a patch that may address situations where API call fails and does not return an error
- Fix for together.ai open source models
Update 1-5-2024
Coming Soon
- Vision & Generator Components
- New Debugging Features
Happy New Year Everyone!
Update 12-24-2023
- Fixed the view mode lock
- ForEach component cannot add input/output
- Code Component has _error by default and input size increased to 5 Mb
- Fixed a bug with error display
Update 12-15-2023
New features
- Introducing Vault system, a key manager to simplify key management accross your agents and avoid storing API keys directly in your components (e.g with APICall or huggingface)
Components
New
- Code component can execute Vanilla JS (external libraries and packages are not allowed, and only ECMA standard is implemented, HTML5 and nodejs specific functions are not available)
- ForEach : A component that can run a loop and return the result
Updates
- APIEndpoint supports form-data requests and binary inputs
- Note component : specify text and background colors
Debugger
- Auto-run feature (click run button if you don’t want to keep clicking steps)
- Attach external requests (e.g postman) : send a request to an agent in debug mode, capture it in smythOS and run the workflow step by step
- Pin intput/output : debug logs can now be pinned in order to allow resizing the output box
- Debugger inputs/outputs can be disabled with a toggle button in order to simulate “no data” (if you leave the component active, debugger sends empty string which smythOS interprets as data)
UI
- Smart connections placement
- Harmonized UI popups
- Optimized inputs/outputs edit popup
- multiple UI updates and fixes
Bug fixes
- SubAgent component calls fixed
- OAuth inconsistencies when used with new OpenAI custom GPTs
- Fixed Session expiration issues for Smyth Studio causing the builder to fail loading if a session expires and you hit refresh button.
Warning
- In order to prevent infinite loops in both SubAgent Calling and ForEach components, we’ve set a loop limit to 100 cycles. This value is hardcoded fro now, but we’ll allow you to control if from the Agent menu very soon. we’re in an experimental stage of both SubAgents and ForEach loops, and we need this measure to be able to analyze the behavior of agents properly without having to interrupt the server (if an infinite loop is detected)
The documentation is not up to date with new component, but we’ll post messages or videos on discord tomorrow in order to explain how to use the new components.
Update 12-8-2023
We’ve just re-launched our server for SmythOS.
The purpose of this channel is to support Agent Engineers build & create the most amazing AI agents on the SmythOS GenAI Application Orchestration Framework.
Here, you can connect with peers, get answers to your questions and showcase your achievements.
SmythOS is a product by INK Content, Inc. and the future of enterprise worldwide.