MONOLITH

Enterprise knowledge infrastructure

One foundation. Infinite intelligence.

Your enterprise knowledge is everywhere. MONOLITH connects it into one intelligent layer, helping people and AI agents access the information they need.

One interface. Every source. Any question.

Explore the platform Discover the architecture
A faceted glass monolith at the centre, fed by glowing data streams from SQL databases, documents, code and apps, cloud storage, business apps and Slack and tools.

Answer · people

January revenue: $250,000

142 customers · 2 sources cited

JSON · AI agents

{ "status": "success",
"request_id": "MON-1001" }

  1. 01

    Multiple enterprise sources

    Databases, documents, cloud storage, apps

  2. 02

    One intelligence layer

    Governed, permission-aware, grounded

  3. 03

    Answers for people and AI agents

    Natural language or structured JSON

Positioning

Not another chatbot. An intelligence layer for your enterprise.

MONOLITH doesn't just generate answers. It connects enterprise systems, understands requests, orchestrates retrieval across structured and unstructured data, and delivers governed, evidence-backed intelligence to people and AI agents.

A conventional chatbot

Question → Answer

One model, one conversation, no direct connection to your systems or their permissions.

MONOLITH

Connected, governed, orchestrated

Every request is authenticated and permission-checked before any source is queried.

  1. Request
  2. Understand
  3. Govern
  4. Retrieve
  5. Orchestrate
  6. Respond

  • Understand

    Interpret natural-language questions and structured JSON requests.

  • Connect

    Access information across enterprise databases, documents and business applications.

  • Govern

    Authenticate users, validate requests and enforce permissions before retrieving information.

  • Orchestrate

    Determine whether structured, unstructured or hybrid retrieval is required.

  • Deliver

    Combine relevant information into unified, evidence-backed responses.

  • Enable AI agents

    Expose enterprise intelligence through APIs for applications and external AI agents.

The problem

Your knowledge is everywhere. Your intelligence shouldn't be.

The facts a business needs are real and current, but they live in separate systems that were never designed to answer questions together. The same customer appears as a row in one database, a clause in a contract and a ticket in a support tool.

Finding one answer across them means knowing where each fact lives, which copy is current, and whether you are allowed to see it. Names do not match, copies go stale and no single search reads everything. A simple question becomes a chain of messages.

AI agents face the same wall. Without a governed way in, each one needs its own connection to every system, and none of them knows what it is allowed to see.

  • 01

    Databases

    Tables, warehouses, CRM records

  • 02

    Documents

    Contracts, policies, PDFs

  • 03

    Business applications

    Billing, HR, support tickets

  • 04

    Cloud storage

    Drives and shared buckets

  • 05

    Internal tools

    Wikis, dashboards, scripts

  • 06

    Communication platforms

    Chat threads, email, meeting notes

One question. Multiple sources. One answer.

Which customers renewed late, and what did the contract say?

Three teams, four systems, one question. MONOLITH connects each authorised source once and returns one governed answer, with sources.

Why one answer takes so long

01

Different formats

A contract is a PDF, a customer is a row, and a renewal date sits in a spreadsheet. Each system stores the same fact in its own shape.

02

Different permissions

Finance may see revenue and support may not. Every source has its own access rules, so an answer has to respect all of them at once.

03

Different owners

Each team maintains its own system and naming. No shared search connects them, so people ask three teams and wait for replies.

The solution

One intelligence layer. Every source.

MONOLITH sits between your sources and everyone who needs an answer, so each source is connected and governed once.

  • Databases
  • Documents
  • Business apps
  • Cloud storage
  • Internal tools

MONOLITH

  • Employees
  • Developers
  • Business applications
  • AI agents

01

Connect once

Authorised databases, documents and apps link to MONOLITH a single time, with their existing permissions carried across.

02

Govern once

Access rules, query checks and audit logging are enforced in one place, not rebuilt for every tool.

03

Answer everywhere

Employees, developers and AI agents get grounded answers from the same governed layer, wherever they work.

How it works

From question to intelligence.

Every request, whether it comes from a person or an application, passes through the same four stages in the same order.

  1. 01

    Connect

    Connect authorised enterprise sources.

    Technically: Read-only connectors with scoped credentials link databases, documents and apps; existing permissions are carried across.

    In plain terms: Tell MONOLITH which systems it may read. Nothing is added without approval.

  2. 02

    Understand

    Interpret the request and identify relevant information.

    Technically: Intent classification, entity detection and source selection produce a retrieval plan.

    In plain terms: MONOLITH works out what you are really asking and which systems hold the answer.

  3. 03

    Retrieve

    Retrieve structured data, documents or both.

    Technically: The router runs a validated SQL query, a semantic document search, or both in parallel, with permissions checked at each step.

    In plain terms: It fetches the exact figures, the relevant paragraphs, or both.

  4. 04

    Respond

    Return grounded answers with relevant sources.

    Technically: A response engine merges the evidence and returns the answer with citations and a request ID.

    In plain terms: You get the answer plus proof of where it came from.

Example question, illustrative trace

“Which customers have not purchased in six months, and what does our inactivity policy define?”

  1. 01 · Connect

    3 authorised sources

    Authorised

  2. 02 · Understand

    Customer data and policy

    Interpreted

  3. 03 · Retrieve

    Path C, hybrid retrieval

    Retrieved

  4. 04 · Respond

    2 sources cited

    Grounded

Retrieval paths

One engine. Three ways to find answers.

Some questions need exact numbers, some need the wording of a document, and many need both. Select a path to see how MONOLITH handles an example question.

MONOLITH · Path C · Hybrid retrieval

Illustrative simulation

Example question

“Find customers who have not purchased in six months and check whether they qualify as inactive under our policy.”

What MONOLITH does

  1. 1Database path: queries orders to list customers with no purchase in six months.
  2. 2Document path: retrieves the written definition of an inactive customer from the policy.
  3. 3Convergence: applies the policy to the list and returns one grounded answer citing both sources.

In simple terms

A question that needs both numbers and wording. MONOLITH finds the customers in the database, reads the policy that defines what counts as inactive, and merges the two into one answer.

Result preview

Data

38 customers have no purchase in the last six months.

Orders database · orders

Policy

Under the policy, 31 of the 38 qualify as inactive. 7 hold an active subscription and are excluded.

“A customer is inactive after six months without a purchase, unless they hold an active subscription.”
Customer-Policy-2025.pdf, p.3

All three paths share the same security controls: authentication, permission checks, read-only query safety and audit logging. A path never returns data the requester is not authorised to see.

Technical architecture

One engine. One gateway. Three retrieval paths.

Natural-language questions and JSON requests enter the same security gateway and the same intelligence engine. Select any block to see its role.

01 · Input layer

02 · Security gateway

03 · Intelligence engine

04 · Retrieval router

05 · Retrieval paths

06 · Unified response engine

07 · Outputs

Two interfaces, one engine

Ask in plain language. Call it as an API. Same engine either way.

People ask questions in natural language and systems send structured JSON requests. Both paths pass through the same governed retrieval layer, so an answer means the same thing wherever it is produced.

INTERFACE A · NATURAL LANGUAGE

For employees, analysts, managers

MONOLITH Studio · Ask

Answer

Total revenue for January was $250,000.

Summed across closed orders in the sales ledger for 1 to 31 January.

  • sources: sales_database.orders, sales_database.invoices
  • permission: finance role, read access confirmed
  • evidence: 2 tables, 1,284 rows matched

INTERFACE B · JSON API

For developers, applications, AI agents

Illustrative simulation

POSThttps://api.monolith.example/v1/query(example domain)

application/json
{
"query": "What was our revenue last month?",
"sources": [
"sales_database",
"finance_documents"
],
"include_citations": true
}

MONOLITH INTELLIGENCE ENGINE

  • Same authentication
  • Same permission checks
  • Same query safety

Returns

  • Structured data
  • Unstructured knowledge
  • Hybrid retrieval
See the full architecture

Six product components

The building blocks of enterprise intelligence.

MONOLITH is built from six components that each own one part of the job, from connecting sources to giving developers and agents controlled access. Select a layer to see what it does.

Platform stack

01

MONOLITH CONNECT

Purpose

Connects enterprise information sources.

Technical responsibilities

  • Authenticates each source with scoped, read-only credentials by default
  • Maps database schemas and document libraries into a governed catalogue
  • Keeps the permissions of the source system attached to every item
  • Shows sync status so stale sources are visible

In plain terms

This is the set of connections that lets MONOLITH read from your databases, documents and business apps, using only the access you approve.

Example

Connect the CRM accounts table and the 'Sales Playbooks' folder, each with read-only access for the sales team group.

Illustrative configuration.

Product overview. These components describe intended capabilities of the platform. No performance figures or usage statistics are shown.

Security

Intelligence without compromising control.

Every request, whatever its origin, passes through the same ordered checks before any information is retrieved.

Lane A

Natural-language request

Lane B

JSON API request / AI agent

Single security flow

  1. Stage 1

    Authentication

    Confirms who or what is asking, whether a person, an application or an AI agent, before anything else happens.

  2. Stage 2

    Request schema validation

    Checks that the request has the expected structure and data types, and rejects anything malformed.

  3. Stage 3

    Permission verification

    Checks that this person or agent is allowed to see this source and this information, every time.

  4. Stage 4

    Query safety

    Confirms every generated or submitted query is validated and permitted before it is allowed to run.

  5. Stage 5

    Controlled execution

    Runs only the approved operation, read-only by default, within defined limits on scope and time.

  6. Stage 6

    Audit logging

    Records who asked, what was checked and what was returned, so every request can be reviewed later.

Natural-language and JSON requests follow the same security controls. JSON requests and AI agents cannot bypass permissions.

  • 01

    Authenticated is not the same as authorised

  • 02

    Generated queries are validated before they run

  • 03

    Every request leaves an audit record

Developer experience

Give your agents access to enterprise knowledge.

External applications and AI agents call the MONOLITH API instead of connecting to every database and document store themselves. Each caller uses a scoped credential, and every request passes the same gateway that governs natural-language questions.

  1. 1

    Agent or app

    A support copilot, workflow or internal tool needs knowledge it is allowed to use.

  2. 2

    POST /v1/query with scoped token

    The request carries a question, a retrieval mode and a credential limited to specific sources.

  3. 3

    MONOLITH security gateway and engine

    Authentication, schema validation and permission checks run before any source is queried.

  4. 4

    Grounded JSON response with sources

    The answer, data rows, citations and applied permissions return together in one typed payload.

Illustrative simulation — not a live endpoint

POST https://api.monolith.example/v1/query
Authorization: Bearer <AGENT_TOKEN>
Content-Type: application/json
{
"question": "Which enterprise accounts renew this quarter?",
"mode": "hybrid",
"sources": ["crm.accounts", "docs.renewal-policy"],
"agent_id": "support-copilot",
"response_format": "json",
"include_sources": true,
"citations": "inline"
}
Source selection
The sources field narrows a request to named sources. It can only narrow: a source outside the token’s scopes is refused.
Citation support
Each answer returns the databases and document sections it drew on, so an agent can show its evidence.

Agents receive only what their scoped credentials permit.

Enterprise use cases

One intelligence layer. Endless possibilities.

Each team works from questions that span systems. Select a team to see the question, how MONOLITH finds the answer and which sources it draws on.

Sales intelligence

“Which open deals over $50k mention a security review in recent call notes?”

How MONOLITH finds it

MONOLITH filters open deals above $50k in the CRM, then searches the matching call notes for security-review language. Deal value and stage come from the database. The evidence comes from the notes, each linked back to its call.

Retrieval path

  • Hybrid

Sources

  • CRM deals
  • Call notes
  • Calendar events

Plans

Choose the intelligence layer that fits your business.

Start with the essentials. Scale your knowledge infrastructure as your organisation grows.

Contact for pricing

MONOLITH Starter

For teams beginning to unify enterprise knowledge.

  • Core enterprise source connections
  • Natural-language querying
  • Basic structured-data retrieval
  • Basic document retrieval
  • Essential API access
  • Standard security controls
  • Basic usage visibility
Get started
Recommended

Contact for pricing

MONOLITH Professional

For teams building advanced enterprise intelligence workflows.

  • Everything in Starter
  • Advanced hybrid retrieval
  • Expanded API capabilities
  • Multiple AI-agent integrations
  • Advanced source permissions
  • Usage analytics and monitoring
  • Priority support
Talk to sales

Custom pricing

MONOLITH Enterprise

For organisations requiring tailored enterprise deployments.

  • Everything in Professional
  • Custom enterprise integrations
  • Advanced governance controls
  • Custom deployment options
  • Enterprise-level access policies
  • Dedicated support arrangements
  • Custom implementation and onboarding
Contact sales

Sales enquiries use the contact form below.

Plans and features are indicative. Contact our team for current pricing and availability.

Our philosophy

A better foundation for enterprise intelligence.

Enterprise knowledge should not be fragmented. Intelligence should not be isolated.

We believe the next generation of enterprise software will be built around connected knowledge rather than isolated tools and disconnected information.

MONOLITH brings enterprise data, documents and business systems together through one intelligent, governed infrastructure, making knowledge accessible to people, applications and AI agents.

  1. Connected by design

    Enterprise knowledge should work together, regardless of where it lives.

  2. Governed by default

    Intelligence should operate within the right permissions, policies and controls.

  3. Built for people and agents

    The same knowledge foundation should serve human users, applications and AI agents.

Frequently asked

Questions, answered.

Straight answers about what MONOLITH does, what is still being confirmed, and how control is kept at every step.

Talk to sales
What is MONOLITH?
MONOLITH is an enterprise knowledge infrastructure layer. It connects authorised sources such as databases, documents and business applications, and lets people and AI agents ask questions across them through one governed engine.
How is it different from a chatbot?
A general chatbot answers from its training data or from text you paste into it. MONOLITH retrieves information from your own connected sources, shows the evidence behind each answer, and applies the permissions your systems already use. Answers are grounded in retrieved records and documents rather than general knowledge.
Which data sources can it connect to?
MONOLITH is designed to connect to databases, documents, cloud storage, business applications and internal tools. Specific connector availability is confirmed with our team, so we do not list every supported product yet.
What is structured data?
Structured data is organised into fields and tables, such as customers, orders or invoices held in a SQL database. It can be counted, filtered and totalled exactly, which is why MONOLITH answers these questions by querying the database.
What is unstructured data?
Unstructured data has no fixed schema: contracts, policies, reports, wiki pages and emails. MONOLITH reads authorised documents, splits them into passages and retrieves the ones closest in meaning to a question, with a citation back to the source.
What is hybrid retrieval?
Some questions need both kinds of data. For example, finding customers with no purchase in six months is a database query, while checking them against the written policy that defines an inactive customer is a document lookup. Hybrid retrieval combines both evidence streams into one answer, and each stream keeps its own permission checks.
Can AI agents use MONOLITH?
Yes. External AI agents can request knowledge through the MONOLITH API using their own scoped credentials. An agent receives only what it is authorised to see, and its requests are logged like any other request. The agent does not need to connect to each source system directly.
How does the JSON API work?
An application submits a structured JSON request to a defined endpoint, authenticates with scoped credentials, and receives a typed response containing the result, its sources and a request ID. Each request is checked against a defined schema before anything runs. Examples on this site are simulations, not live calls.
How is enterprise data protected?
Every request starts with authentication and passes permission checks before any data is retrieved. Database access is read-only by default, requests are validated, and actions are written to an audit log. Specific encryption and compliance details are shared on request.
Does MONOLITH execute SQL?
For structured questions, MONOLITH generates SQL, validates it, checks permissions, and then runs it under controlled, read-only execution by default. Generated SQL is never assumed to be safe. Write operations are not part of the default configuration and require an administrator to configure them explicitly.
How are permissions enforced?
Natural-language and JSON requests pass through the same security controls: authentication, schema validation, permission verification, query safety, controlled execution and audit logging. Nothing bypasses these checks, including API calls and AI agents. Permissions are verified at each retrieval step, not only once at the start.

Talk to sales

Your knowledge deserves a better foundation.

Talk to our team to see how MONOLITH connects your authorised enterprise sources into one governed intelligence layer for people and AI agents alike.

Explore the architecture

Product demos on this site are illustrative simulations.

Built with