The team will be composed of experienced MeTTa and Prolog programmers having wide ranging experience and core Mettalog knowledge. We need to add a multifaceted manager-visionary-evangelist.

Recruit manager/evangelist.

$1 USD

Arguably this is the most important step. We need a team ready to really push Mettalog into mainstream. Without a synergy here we shouldn't proceed. We must have a shared, clear vision of this goal.

Deliver new server functionality and clear documentation to enable MeTTa programs running on Mettalog to respond to network requests.

MeTTa programs will load once upon initial invocation persist in memory and serve multiple users concurrently. With Python as the main control language developers will be able to use standard web servers like Apache and nginx as well as cloud services such as AWS EC2. Integration with popular libraries such as LangChain will allow MeTTa to be used as part of larger pipelines that combine symbolic reasoning with language model output. The system will support communication through the Model Context Protocol (MCP) enabling external language models (e.g. Claude ChatGPT) to send queries and receive structured responses from MeTTa programs.

$20,000 USD

This step is successful if the described functionality has been thoroughly exercised with documentation. Mettalog needs to reliably handle these scenarios.

New built in command(s) to enable the MeTTa knowledge graph to be a system of record supporting concurrent users not just read-only analytics.

Presently there is a “get-atom” MeTTa method. This will be extended to a new “get-atom-lock” method to prevent potential asynchronous update problems. Other methods will be introduced if necessary. Thus the deliverable is source code enhancements for above as well as test cases documented.

$20,000 USD

This step is successful if tests are passed and documentation produced.

The MeTTa language is a unique language with a variety of potential uses but it is arguable that MeTTa lacks a determinate identity. Is it a knowledge graph? Is it LISP? Is it an alternative to a procedural language? Why does it insist on giving you all possible results? This makes it problematic to appeal to mainstream developers who in the age of LLMs do fundamentally need the stability of knowledge graphs for augmented generation and could potentially benefit. Some of the essential features of MeTTa are: * Reminds one of a network type database can be updated like a database (CRUD) * Capable of Turing completeness * Processes/returns all possible paths and results of a call by default while providing means of managing such an explosion (eg. the "collapse" statement) * Suited for general ontologies while also able to act as narrow AI * Poised to be anti-hallucinogenic stabilizer vis a vis generative AI * Could act in a RAG type role that is as a KNOWLEDGE GRAPH

This deliverable will concentrate on clear examples of MeTTa based applications in a symbolic Turing complete knowledge graph role and how this can support mainstream use (eg. LLMs). While examples are good it's not enough. The developer will have the means to input raw data and output a starting MeTTa knowledge graph. Rejuve.bio has developed a data import tool. This important no-code tool enables importing of a set of CSV or JSON files for the creation of a custom knowledge graph (the tool is also known as an "atomspace builder"). Once the KG is created MeTTa code in the form of a directed knowledge graph can be generated. This code can be then modified and extended as needed for customization and fine tuning of the domain application.

$30,000 USD

This is probably the most challenging step of the project. The step will be successful if a completely coherent message can be made to mainstream AI developers supporting MeTTa's role in neurosymbolic architectures taking the form of an LLM front end and MeTTa back end. The user must be able to clearly see -- with minimized complexity -- the progression from their input raw data files (eg., CSVs) into MeTTa code doubling as knowledge graphs.

Mettalog is a maturing fast MeTTa interpreter and transpiler based upon Prolog. Although largely complete some additional shoring up may be needed if not covered under other funding or otherwise incomplete.

1) Enable the easy installation of mettalog by apt docker and similar means without complexity or any suspense whatsoever regarding whether or not it works. It will work. 2) A highly reliable LSP server for programmer assistance (gives help while coding). 3) A largely functioning transpiler. If the transpiler is incomplete we must have a "fall through" for the interpreter. There is a working interpreter now still reasonably fast but it is slower than the transpiler which generates Prolog code directly. This is a crucial requirement as developers will really be running their MeTTa code as fast Prolog programs. As of this writing the transpiler is very close to being complete.

$4,999 USD

All of the above work reliably.