Fonsalve
ESMap a workflow

Fonsalve Systems / Applied AI & Automation

We build the system between the work and the outcome.

Automation, software and applied AI that connect requests, documents, decisions and tools without replacing everything that already works.

Map a workflowSee the system

We start with one concrete process. We understand it, prove it and take it to production.

One process, step by step

Pick a process and run it. You will see the whole path with synthetic data: where it enters, what is understood, what is checked, who decides and what gets recorded.

Synthetic data demonstration
Pick a process

entradaEmail with drawing attached

  1. IntakeEmail received with drawing (rev. C) and a stated deadline
  2. UnderstandingCustomer, material and deadline identified in message and drawing
  3. VerificationRequired fields present · file revision checked
  4. Human reviewThe bid owner confirms scope and margin
  5. ActionOpportunity created and task assigned to engineering
  6. RecordCustomer confirmation and a full trace of the decision

Synthetic data. Nothing here connects to a live system.

The problem is rarely one tool. It is what happens between them.

Fonsalve builds that layer: it receives the work, applies rules, brings in AI where it adds value, asks for a decision when one is needed and records every step.

  1. Email
  2. Document
  3. Spreadsheet
  4. Person
  5. ERP
  6. SharePoint
  7. Customer

Four outcomes, one system

These are not separate services: the same system receives the work, understands it, puts it in front of whoever decides and records it in the tools you already use.

Workflows that execute

Receive, classify, validate, assign, notify and record without every step depending on someone copying information.

  • Requests
  • Quotations
  • Issues
  • Documents
  • Approvals
  • Follow-ups

Software that fits the operation

Internal applications, portals and control surfaces built around how the company actually works.

  • Customer portals
  • Supplier portals
  • Internal tools
  • Operational views
  • Mobile apps
  • Review flows

AI with evidence and control

Systems that search, extract, propose and assist while showing sources, permissions and human review points.

  • Document assistants
  • Classification
  • Extraction
  • Enterprise search
  • Verifiable summaries
  • Decision support

Systems that connect

We integrate the new layer with the existing environment so modernisation does not become another silo.

  • Microsoft 365
  • SharePoint
  • Email
  • SQL
  • ERP
  • APIs
  • Files
  • Legacy systems

Reference builds

Reference constructions with synthetic data showing how we solve each family of problem. They are neither clients nor finished products.

Quotation Intake

Reference build

Requests arrive by email with drawings attached and nobody knows how many are open.

Flow
Intake → customer, material and deadline extracted → revision check → review → opportunity
Human decision
The bid owner confirms scope and margin before anything is created.
Observable result
A single queue with status, owner and deadline visible.

Knowledge & Evidence

Reference build

The answer exists, but it lives across SharePoint, inboxes and three people's memory.

Flow
Question → permission-aware search → answer with citations → expert confirmation
Human decision
Nothing ships without a linked source and someone validating it.
Observable result
Answers with checkable provenance and reusable questions.

Service Operations

Reference build

Issues arrive through three channels and priority depends on who shouts loudest.

Flow
Intake → classification → proposed priority → assignment → follow-up
Human decision
A person confirms priority; the system only proposes it.
Observable result
First response time measurable by channel and type.

Operational Control

Reference build

Nobody sees the real state until someone builds a spreadsheet on Friday.

Flow
Process events → consolidated state → exceptions → owner
Human decision
Exceptions escalate to a named person, not to a shared inbox.
Observable result
One view of what is stuck, since when and who owns it.

How we work

We do not start with an abstract transformation. We start with one process that deserves to work better.

  1. 01

    Map

    We pick a process with volume, cost, an owner and an observable outcome.

  2. 02

    Prove

    We build a bounded demonstration and test data, rules, integration and risk.

  3. 03

    Build

    We take it to a real environment with identity, security, traceability and acceptance criteria.

  4. 04

    Operate

    We measure how it runs, handle exceptions and evolve the system.

Who this is for

We do not work by sector but by operational pattern. These are the three environments where that pattern shows up most.

Industrial companies

Orders, drawings, delivery notes and deadlines travelling by email between engineering, purchasing and the shop floor.

B2B operations

Requests, quotes and issues from customers and suppliers spread across inboxes and spreadsheets.

Document-heavy or regulated processes

Files, evidence and approvals that must be provable long after the fact.

The firm

Fonsalve is an independent technology firm.

We design and build systems for companies that need to move forward without replacing everything that already works. We combine business judgment, software engineering and applied automation.

Let's start with one process.

Tell us which one is costing you time. We will come back with the questions needed to know whether it deserves to become a system.

Map a workflow

Or send a short description

Rather tell us in detail? Use the Process Mapper