Release version 1.2.0

Model Support added

• Google Gemini

Operations improved

• Read Image now supports image files as input and not only URLs
• Sentiment Analyze has been extended and provides detailed feedback on the rating of the prompt. Kudos to @probiswas for the contribution

Operations added

• Use tools ai native: The ability to leverage native LLM tooling capabilities and receive tool execution requests.
• Toxicity detection: Now you can detect toxicity based on moderation models, currently only support OpenAI and Mistral

What's Changed

• W-16235578: Embedding Operations Failure Scenarios by @arpitg-1 in #47
• W-16295319: Updated demo name to remove sonar issue by @arpitg-1 in #53
• Upgrade/34 by @amirkhan-ak-sf in #62
• Create Jenkinsfile by @garveshmehrotra-sfdc in #66
• Develop by @amirkhan-ak-sf in #67
• Support for Huggingface interference API by @amirkhan-ak-sf in #69
• Fix/dev sync from master by @dipeshkumardutta-sfemu in #77
• apply/changes by @amirkhan-ak-sf in #78
• Ignore failed test cases by @dipeshkumardutta-sfemu in #80
• Fix mulechain artifact by @dipeshkumardutta-sfemu in #82
• Update README.md by @dipeshkumardutta-sfemu in #83
• Enhancements by @amirkhan-ak-sf in #85
• Image read by @amirkhan-ak-sf in #86
• Gemini ChatModel Added by @amirkhan-ak-sf in #88
• mistral tool-native issue fixed by @amirkhan-ak-sf in #89
• Upgraded LangChain4J to 0.35.0 (latest) by @amirkhan-ak-sf in #90
• Exavalu: enhanced sentiment analyze operation, updated readme file. by @probiswas in #93
• Moderation Model added for Mistral AI by @amirkhan-ak-sf in #96
• Changes to support maven central release by @rhoegg in #97
• Update README.md by @amirkhan-ak-sf in #98
• Better README for multiple installation methods by @rhoegg in #100
• Update pom.xml by @amirkhan-ak-sf in #101
• V1.2.0 revisit by @amirkhan-ak-sf in #103

New Contributors

• @garveshmehrotra-sfdc made their first contribution in #66
• @probiswas made their first contribution in #93
• @rhoegg made their first contribution in #97

Full Changelog: v1.0.0...v1.2.0

Release version 1.0.0

Streamlined version of the officially supported & certified MuleSoft AI Chain connector

• Added Hugging Face support
• Improvedd Ollama Integration
• **Upgraded LangChain4J version to 0.34.0
• Improved Performance
• Added Support for MuleSoft Select Category

What's Changed

• Develop by @dipeshkumardutta-sfemu in #73
• Fix icon and set the version 1.0.0 by @dipeshkumardutta-sfemu in #74
• Fix/icon and version by @dipeshkumardutta-sfemu in #75
• Ignore failed test cases by @dipeshkumardutta-sfemu in #79
• Fix mulechain artifact by @dipeshkumardutta-sfemu in #81

Full Changelog: v0.3.0...v1.0.0

Release version 0.3.0

We’ve added some new features and enhancement:

1. New operations / attributes:

• Scanned documents via Image Models
• TokenUsage for Tools Operations
• TokenUsage and other meta data is now available in attributes and not anymore in payload.

2. Other enhancements

• Input & output type is now of type Inputstream, no need to transform String into JSON as post processing after each operation (@arpitg-1 @dipeshkumardutta-sfemu thanks)
• New logo (@AhmJibDev thanks)
• Demo project is added to the repository (@arpitg-1 @dipeshkumardutta-sfemu thanks)
• Performance improvements (@arpitg-1 @dipeshkumardutta-sfemu thanks)
• Security scan fixes (@arpitg-1 @dipeshkumardutta-sfemu thanks)

3. Bug fixes

• Configuration not loading - in some instances the configuration of the LLM provider and its model won’t load. This issue is fixed now. (@arpitg-1 thanks)

4. Impact on existing projects

• Existing projects created with 0.2.0 must be migrated.
• The migration will be:
• replacing “old” operations with new ones
• or recreate the mule app

Release version 0.2.0

📤 𝐔𝐩𝐝𝐚𝐭𝐞𝐬 𝐟𝐫𝐨𝐦 𝐭𝐡𝐞 𝐌𝐮𝐥𝐞𝐂𝐡𝐚𝐢𝐧 𝐏𝐫𝐨𝐣𝐞𝐜𝐭
We are releasing 𝐯0.2.0 of the MuleChain AI Connector

📎 𝐖𝐡𝐚𝐭 𝐢𝐬 𝐬𝐨 𝐬𝐩𝐞𝐜𝐢𝐚𝐥 𝐚𝐛𝐨𝐮𝐭 𝐭𝐡𝐢𝐬 𝐫𝐞𝐥𝐞𝐚𝐬𝐞?
It is all about feedback, we received tons of feedback from the community users, via Github/LinkedIn. The overall experience has been great.

We implemented some nice features to make our play more compelling:
✅ 𝐀𝐝𝐝𝐞𝐝 𝐭𝐨𝐤𝐞𝐧 𝐜𝐨𝐧𝐭𝐫𝐨𝐥 𝐚𝐬 𝐚 𝐩𝐚𝐫𝐚𝐦𝐞𝐭𝐞𝐫 in the configuration with 𝒎𝒂𝒙𝑻𝒐𝒌𝒆𝒏
✅ 𝐀𝐝𝐝𝐞𝐝 𝐭𝐞𝐦𝐩𝐞𝐫𝐚𝐭𝐮𝐫𝐞 𝐟𝐨𝐫 𝐭𝐡𝐞 𝐋𝐋𝐌 to come up with a response
✅ Moved the 𝐦𝐨𝐝𝐞𝐥 𝐜𝐨𝐧𝐟𝐢𝐠𝐮𝐫𝐚𝐭𝐢𝐨𝐧 𝐭𝐨 𝐭𝐡𝐞 𝐋𝐋𝐌 𝐩𝐫𝐨𝐯𝐢𝐝𝐞𝐫 𝐜𝐨𝐧𝐟𝐢𝐠𝐮𝐫𝐚𝐭𝐢𝐨𝐧 (thanks to Arpit Gupta and Dipesh Dutta). This has 𝐈𝐦𝐩𝐫𝐨𝐯𝐞𝐝 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐟𝐨𝐫 𝐋𝐋𝐌 𝐜𝐚𝐥𝐥𝐬.
✅ Added 𝐬𝐮𝐩𝐩𝐨𝐫𝐭 𝐟𝐨𝐫 𝐉𝐚𝐯𝐚 11, 𝐉𝐚𝐯𝐚 17, next to Java 8
✅ 𝐈𝐦𝐩𝐫𝐨𝐯𝐞𝐝 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 for In-memory Vector DB
✅ 𝐈𝐦𝐩𝐫𝐨𝐯𝐞𝐝 𝐭𝐨𝐨𝐥𝐬
✅ New operation added: 𝐄𝐦𝐛𝐞𝐝𝐝𝐢𝐧𝐠 𝐚𝐝𝐝 𝐟𝐨𝐥𝐝𝐞𝐫
✅ New operation added: 𝐄𝐦𝐛𝐞𝐝𝐝𝐢𝐧𝐠 𝐪𝐮𝐞𝐫𝐲 𝐬𝐭𝐨𝐫𝐞
✅ New operation added: 𝐄𝐦𝐛𝐞𝐝𝐝𝐢𝐧𝐠 𝐦𝐞𝐭𝐚𝐝𝐚𝐭𝐚 𝐚𝐝𝐝𝐞𝐝 𝐨𝐧 𝐭𝐡𝐞 𝐪𝐮𝐞𝐫𝐲 𝐚𝐧𝐝 𝐠𝐞𝐭𝐭𝐢𝐧𝐠 𝐢𝐧𝐟𝐨
✅ 𝐓𝐨𝐤𝐞𝐧𝐔𝐬𝐚𝐠𝐞 has explicitly shown in the output
✅ 𝐀𝐝𝐝𝐞𝐝 𝐧𝐞𝐰 𝐝𝐨𝐜𝐮𝐦𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧 next to the v0.1.0 release

🔔 𝐆𝐞𝐭𝐭𝐢𝐧𝐠 𝐬𝐭𝐚𝐫𝐭𝐞𝐝 𝐰𝐢𝐭𝐡 𝐭𝐡𝐞 𝐥𝐚𝐭𝐞𝐬𝐭 𝐚𝐧𝐝 𝐠𝐫𝐞𝐚𝐭𝐞𝐬𝐭 𝐯0.2.0 𝐫𝐞𝐥𝐞𝐚𝐬𝐞
✅ 𝐃𝐨𝐜𝐬: https://lnkd.in/eTStcNT6
✅ 𝐎𝐩𝐞𝐫𝐚𝐭𝐢𝐨𝐧𝐬: https://lnkd.in/eNqXmCAv

Release version 0.1.0

Currently supported using JDK8
All operations tested with OpenAI

The initial release of the MuleChain AI connector supports the following operations, categorized into different topics for ease of use:

• [AGENT]:
  • Define Prompt Template
• [CHAT]:
  • Answer Prompt
  • Answer Prompt with Memory
• [Embedding]:
  • Add Document to Store
  • Get Info from Store
  • New Store
• [IMAGE GENERATION]:
  • Generate
  • Read
• [RAG]:
  • Load Document
• [SENTIMENT]:
  • Analyze
• [Tools]:
  • Use AI Service