MPLOI.AI

A solution built from the ground up for enterprises. MPLOI.AI is an on-premise Generative AI solution that provides a NO-CODE UI allowing administrators to add sophisticated AI Agents that can perform advanced enterprise level automation tasks and connect to local data.

What problems does MPLOI.AI solve?

• DATA PRIVACY: Many businesses are excited about the potential of GenAI solutions, but worry about data privacy. Sending data to public GenAI solutions can introduce serious privacy concerns and create security risks.
  
  Think of MPLOI.AI as a secure version of ChatGPT that lives entirely within your own data center. You have complete control over the system, its security and your data never leaves your premises - unless you choose to let it.
  
  MPLOI.AI connects seamlessly with your existing data sources, whether it's stored in databases (SQL), accessed through APIs, pulled from websites, or resides in documents. This allows you to leverage the power of GenAI without compromising data security.
  
  MPLOI.AI empowers you to explore the benefits of GenAI technology while keeping your sensitive data safe and secure. You get all the advantages of automation and intelligence without the security risks.
• SIMPLICITY: GenAI solutions are easy to use. You just load them into your browser and start chatting. The complexity comes when you want to do more than get a standard response to a standard prompt. It takes quite a bit of work, custom development, and effort to get a GenAI solution that can interact with your own data and take actions beyond standard web searches and text/image generation. MPLOI.AI simplifies this entire process into point and click actions. Every field is easiliy understood and nothing is further than a few clicks from start to finish.
• FLEXIBILITY: Most GenAI solutions are focused on a specific market. Like a tuned model for healthcare. Or a tuned model for Legal work. MPLOI.AI is a solution that can be used and reused for any market. Whether its technology, healthcare, finance, legal, etc..., you have the flexibility to create model-specific agents that have a specific expertise and can be further trained on custom data. GenAI sprawl can get out of control. Go for a solution that can fit all use cases. MPLOI.AI can be that solution.
• SCALE: Enterprise On-Premise GenAI solutions require horsepower. Figuring out what scale limits the solution you need takes a considerable amount of testing and development. MPLOI.AI provides scale tested turn-key solutions designed to meet your current and future demands.
• ENTERPRISE READINESS: Authentication, Authorization, Accounting, Data Retention, and Standards Compliant are just a few of the words that every enterprise expects out of an enterprise product. MPLOI.AI is built for the enterprise and commands a vast enterprise feature set developed specificallhy to adhere to the requirements of large secure enterprises such as U.S. Federal agencies allowing it to meet the requirements of almost any business.
• RISK MITIGATION: New AI solutions introduce new risk. There are cases of AI agents being reverse engineered out of defined system instructions by hackers to wreak havok. Poorly developed agents are in some cases carrying out tasks incorrectly which can cause extreme issues. MPLOI.AI has built-in AI safeguards that allow you to have total control of what an agent can or cannot do without depending on proper prompt engineering to do it. MPLOI.AI safeguards are built into the code that the agents depend on to perform their actions.

STEP 1: Deploy

MPLOI.AI can be installed as software on a server you already own such as a virtual machine (as long as the server has the correct GPU resources) or via a turn-key hardware appliance with all the software and hardware resources setup and ready to go. Work with our team to pick the right solution for your organization.

STEP 2: Add Datasets

With MPLOI.AI, you can add local or cloud based datassources. These can be SQL databases such as Postgres, MySQL, MSSQL, MongoDB, Oracle, etc... They can also be documents, Web URLs or Documents.

There are many layers of instructions you can provide the agent when adding a datasource. For example, if you have a SQL database table named inventory, and the inventory table had 5 columns, each column will have a toggle selector allowing control of which columns your agent has access to as well as a field to instruct your agent on what that column is and what to do with it. Based on this information alone, your AI Agents that have access to this dataset can put together very complex SQL queries (within the boundaries of the permissions you have assigned the agent within MPLOI.AI) on his own and either present the data back to you in any format you request it (ie. Tables, charts, graphs, etc...) or use the data to further complete the task you have assigned it.

Note: Datassets can be assigned on a per agent basis, allowing total flexibility as to what data an agent does or doesnt have access to.

STEP 3: Create Agents

Agents are very easy to create using MPLOI.AI. There is no complex coding requirement. Simply go to the agents section of the MPLOI.AI UI and click add agent. In the Agent definition screen, you can give your agent a name, some simple high level instructions, and even give it a personality. This is also where you can turn on or off whichever datasets an agent has access to. This step is also where you turn on/off which actions an agent has permissions to perform.

STEP 4: Empower Agents

You can now give each agent access to perform actions far beyond the agent actions available in most other GenAI products software (ie. web search, text/image generation) The following list are availabe actions that agents can be permitted to use on a per agent basis:

• SQL: With access to the SQL actions, Agents can interact directly with SQL databases. This allows an agent to read, update, and delete records as needed based on the assigned task. Per agent permissions can be defined that allow agents to have more or less priviledges as needed. For example, if you want an agent to only have the ability to use SQL select commands and nothing more, these priviledges are enforced at the code level that the agent uses to carry out the action (meaning an agent can never perform an action not permitted to perform, even in error).
• SSH: Agents that have access to SSH can directly interact with any SSH server such as Servers, routers, switches, etc... SSHv2 and SSHv3 are both supported and MPLOI.AI uses FIPS compliant encryption libraries. With SSH, you can enable agents to both update remote devices as well as help troubleshoot remote devices. Agents can be trained on specifc vendor equipment such as a Cisco Systems router, a Palo Alto Networks firewall, or a Red Hat Linux Server.
• REST API: Virtually every enterprise technology product produced today can be controlled through the use of ReST API. This means that agents can be trained on that API to automate tasks. APIs are very powerful and as such safeguards are built into MPLOI.AI that allow adminsitrators to grant agents permissions to use APIs based on the endpoint IP address and the ReST API action (ie. GET, PUT, POST, DELETE, etc...).
• LINUX COMMAND LINE: MPLOI.AI agents can be granted command line level processes to perform CLI tasks and present the output back to the user or use the output to make further decisions based on the task requirements and permissions granted.
• PYTHON: Although we are proud of building a product that requires no coding to take advantage of, some of our customers want the flexibility to match up our agents with their code. And for them, we built support to access the python scripting engine directly. This allow agents to execute administrator created-code within permitted boundaries as well as create code to execute.
  
  NOTE: Today, a global python virtual environment is provided with access to PIP manage packages through the MPLOI.AI UI. In a future release, per agent VENV's will be available.
• SCHEDULING SERVICE: Sometimes agents need to do more than ad hoc tasks. Sometimes we need them to have the ability to work on a schedule. Whether it be working through a change control process and executing actions during an approved window, or checking things on some pre-defined interval. By granting an agent access to the scheduling service, it can determine based on your prompt, how it needs to further schedule itself to run later.

NOTE: MPLOI.AI will be releasing tuned models that are specific to devops. These models will allow agents to be trained in advance on many vendors technologies such as Cisco, Vmware/Broadcom, Palo Alto Networks, HP/Juniper, AWS, Azure, Google Cloud, etc... This training will include API and command line training as well as technology training.

STEP 5: Secure Your Agents

Your organization is only as strong as its weakest link. There are many points to consider when designing agents who can, in many ways, act and think as a human if allowed to. Securing an agent is far more complex than securing the traffic to and from the GenAI solution. Agents should be secured in a more secure manner than we secure humans. They should be given granular role based access to whichever actions they perform. With MPLOI.AI, we can provide granular access to each function an agent has access to. For example, if an agent has access to SSH, we can define exactly which commands an agent can execute and to which devices it has access. We also keep very close logging of all actvity produced by the MPLOI.AI users and its agents. The conversation and the actions taken are all verbosely logged. Infinite data retention designs are fully supported as are required by many Government agencies.

STEP 6: (optional) Create Agent Workflows

Sometimes you may have very large complex tasks that just arent cut out for a single agent to handle. With MPLOI.AI, you can chain agents together into an agent workflow, allowing each agent to have more defined skillsets and portions of the task. Just like any large project with a large team, enterprises can leverage similar concepts with software agents to take down large projects.

STEP 7: Talk to Your Agents

This is where the fun begins. Now you can select an agent or agent workflow and start talking to it. Ask it questions about data, ask it to carry out a task at some specific time in the future, take on multi-step tasks and processes, etc.. Your limits are largely defined by your own imagination.