FOGOARAI · v1

AI employees that never leave your company

Fogoarai lets large companies put AI to work inside their own systems. The data never leaves their network, the company's IT team controls what each AI worker does, and every action gets recorded.

Inside your own systems Your IT controls what AI does Every action recorded No data to OpenAI, Google, or Anthropic
// built for regulated industries — insurance, mining, banking, healthcare, finance
fogoarai · control plane · single-tenant · illustrative
OPERATING
02 · THE GAP

Why large companies can't just use ChatGPT

Most AI today sits outside the company. It can't see internal systems, can't be governed, and can't safely handle the data that matters. For a regulated business, that means it stays at the edges instead of doing real work.

01 / COST

Frontier AI for every task is expensive

Most enterprise AI work doesn't need the most expensive model. Companies that route to smaller specialized models save 50–70% on cost.

02 / DATA

Sensitive data can't leave the company

Contracts, customer records, financials, tickets, internal documents — none of this can go to OpenAI, Google, or Anthropic without breaking compliance.

03 / SYSTEMS

AI is disconnected from internal tools

Real work needs the AI to read from CRM, ERP, databases, documents, and approval flows — not just chat.

04 / CONTROL

No way to govern what AI does

Without controls, audit logs, and approvals, AI can't get past a security review at a regulated company.

03 · ARCHITECTURE

Here's what's inside the box.

Seven layers turn a single AI request into a governed action the company can trust — from who asked, to which model answered, to which system was touched, to who approved it, to what got logged.

01
USER / WORKFLOW

Where the request starts

A person, a ticket, a schedule, or another system kicks off the work — with the user's role and the request's context attached.

02
POLICY LAYER

Who can do what

Before any work begins, the company's rules decide what this user, this worker, and this workflow are allowed to do.

03
RUNTIME

The AI worker

The worker plans the steps, remembers context across them, and uses the right tools to complete the task end to end.

04
ROUTER

Picks the right model for the job

Routine work goes to smaller, cheaper specialized models. Harder work escalates to a frontier model. The company sees what was chosen and why.

05
TOOLS & DATA

Connection to company systems

CRM, ERP, databases, documents, ticketing, email, internal APIs — the worker reads and writes only what policy allows.

06
APPROVAL

Human approval where it matters

Sensitive or irreversible actions stop and wait for the right person to approve before anything is committed.

07
AUDIT

Records of every action

Every input, model choice, tool call, approval, and output is recorded — searchable, exportable, and built for security and compliance review.

// example workflow stream
illustrative
04 · COST-OPTIMIZED ROUTING

Small models for routine work. Large models for the hard parts.

Most of the AI work inside a company is narrow and repeatable — classify a ticket, pull data out of a document, validate a field, summarize a report. That work belongs on smaller, cheaper, specialized models. Frontier models only step in when the task actually needs them.

One frontier model for everything TYPICAL

  • You pay frontier prices on every call, even simple ones
  • Slower responses under high volume
  • Locked to one model vendor for every task
  • No way to specialize for narrow, repeatable jobs
  • Hard to govern across teams at scale

Routed workers with Fogoarai FOGOARAI

  • Lower cost per completed task
  • Faster responses for routine, high-volume work
  • Runs inside the company's own network
  • Specialized behavior tuned to each workflow
  • Structured outputs the company can validate
  • Falls back to frontier models only when needed

Model Router ROUTING

▸ Inspecting task
▸ classify ticket → category
DEFAULT · SLM fog-classify-7b conf 0.94
FALLBACK · LLM frontier-l on-demand
policy: spend-cap · region-lock · pii-redact illustrative · roadmap
50–70%

The target cost reduction on high-volume workflows when routine work runs on smaller specialized models, with quality protected by validation and confidence-based escalation.

// design target · varies by workflow · not a universal guarantee
Status

In active design. Today, routing happens at the MCP client layer (Claude Desktop). The internal router shown above is the next major release.

05 · PRIVATE DEPLOYMENT

Your AI workers. Your data. Your infrastructure.

Fogoarai is built for companies that can't send sensitive work to an outside black box. It runs inside the company's own cloud, dedicated VPC, or on-prem environment — with the company in control of where data sits, which models are used, what gets logged, and what each worker is allowed to do.

Private cloud & on-prem deployment
vpc · dedicated · air-gapped
Data residency controls
region-locked execution
Local or private model serving
slm + frontier · private endpoints
PII redaction workflows
pre-prompt + post-output
Encrypted logs
at-rest + in-transit · key custody
Tenant isolation
per-org boundaries
Internal identity integration
sso · scim · idp
Security-review friendly
future-ready architecture
06 · CUSTOMERS

Today, Fogoarai runs inside two large companies in Chile.

Two real workflows, in production, in regulated industries. Both are companies that could not send their data to OpenAI, Google, or Anthropic — so the AI runs inside their own systems instead.

01 / LIFE INSURANCE ● LIVE · daily use

Competitive intelligence on rival policies

An AI worker that reads competitor disclosures and filings, builds side-by-side comparisons across coverage, pricing, and policy terms, and points out gaps for the company's analysts.

read filingsextract termscompareflag gapsbrief analyst
Industry · Insurance Country · Chile Data · Public filings + internal notes
02 / INDUSTRIAL MINING TECH ● LIVE · daily use

Knowledge synthesis across mining experiments

An AI worker that connects the engineers to their archive of mining experiments — surfacing prior findings and outcomes so teams can use relevant past work instead of repeating expensive studies.

search archivesummarizecompare runscite sourcerecommend
Industry · Industrial mining Country · Chile Data · Internal experiment archive
Next workflows in scope
Document review Competitive intelligence Knowledge synthesis Internal IT support Finance ops Compliance evidence
07 · CONTROL PLANE

A single place to see what every AI worker did, and why

The control plane is what the company's IT, security, and operations teams use to run AI workers like any other production system — track cost, latency, errors, approvals, and every action recorded against policy.

PRODUCT UI Single-tenant view shown below. Numbers reflect the scale of one live customer today, not a fleet — what the same UI looks like with one company's two workflows running.
fogoarai · control plane
Overview Workers Workflows Evals Approvals Audit Policies
range: last 24h LIVE
Active workersstable
2
2 workflows · 1 tenant
Cost / completed task−54%
$0.063
vs frontier-only baseline
p95 latencystable
482ms
routed paths · 24h
Tool success rate+0.2%
99.4%
across connected systems
Small vs frontier routing · 24h policy-driven · confidence-weighted
Small model · 72.0% Frontier fallback · 18.0% Human approval · 10.0%

Workflow runs · 24h
47
Escalation rate
10.6%
low confidence · sent up
Failed validations
1
schema · 1 connector
Security events
0
policy violations
Human approval queue 2 pending
Audit · every model + tool call
trace · cost · latency versioned prompts ● streaming
// versioned · prompt + model + policy + workflow search · export · replay
08 · GOVERNANCE

Governance from the first token to the final action

Every layer of Fogoarai is built around the controls enterprise security, risk, and compliance teams expect from production infrastructure.

G.01Role-based access controlACTIVE
G.02Tool allowlists and action boundariesACTIVE
G.03Human-in-the-loop approval flowsACTIVE
G.04Full audit logs for model + tool callsACTIVE
G.05Structured output validationACTIVE
G.06Sensitive data redactionACTIVE
G.07Versioned prompts, models, policies, workflowsACTIVE
G.08Evaluation gates before production deploymentACTIVE
G.09Continuous monitoring after deploymentACTIVE
09 · VISION

From one workflow to the whole company

Fogoarai starts with one workflow that matters. Over time, every team can run AI workers on the same private layer — same data controls, same tools, same approvals, same audit. That's how a company moves from AI experiments to AI doing real work, everywhere.

PHASE 01 · WEEK 1–6

Try it

One workflow in production with measurable ROI and a clear success metric.

PHASE 02 · QUARTER 1–2

Production

The first AI workers go live with routing, tool access, approvals, and full audit.

PHASE 03 · QUARTER 3–4

Across teams

Reusable AI workers spread across departments, on the same controls and quality gates.

PHASE 04 · YEAR 2+

Company-wide

AI workers run alongside human teams across the company, on a shared, governed layer.

10 · WHY NOT THE ALTERNATIVES

Why a large company picks Fogoarai over the obvious options

There are four ways a large company usually tries to solve this. Here is where each one falls short for a regulated business.

A · OPEN-SOURCE LIBRARIES

Why not LangChain or CrewAI?

Those are libraries for engineers, not products a company can install. They have no controls, audit, or security review story.

B · CLOUD AI

Why not Google Vertex AI Agent Builder, AWS Bedrock, or Azure AI?

They tie the company's AI to the cloud provider's infrastructure. In regulated industries, that is the deal breaker before evaluation starts.

C · VERTICAL PRODUCTS

Why not Moveworks or Glean?

Those are good at one specific job. A company that buys Moveworks for IT still needs something for finance, legal, document review, and competitive intelligence. That something is Fogoarai.

D · BUILD IN-HOUSE

Why not build it internally?

Most large companies start by assembling LangChain + OpenAI + custom access controls. They end up with brittle systems they can't evaluate, version, or hand to compliance. Fogoarai is the buy option.

11 · GET STARTED

Put your first AI worker in production

Pick one workflow that matters. Connect it to the right internal systems. Add the approvals and audit a regulated business needs. Measure the result inside six weeks.

PILOT WORKFLOWone workflow · scoped together
DURATION6 weeks
DEPLOYMENTyour VPC or on-prem

DATAstays inside your network
APPROVALSyour people, your rules
SUCCESS METRICcost saved · quality held

STATUS● READY TO SCOPE