Change the box.
We reframe a problem before we solve it. The first question on any task: should a person or an agent do this at all?
About Human Managed
We named the company for the part AI can't take over.
Why Human Managed
When we started in 2018, nobody was talking about agents. We named ourselves Human Managed because we believed machines should run the repetitive work on data while humans manage the decisions. Producing intelligence was never going to be the hard part. Deciding what to do with it, and answering for the result, was. AI has dramatically multiplied that problem: organizations are handing it work faster than they are building the accountability to match. Our answer hasn't changed. Agents operate. Humans monitor, engineer and govern.
Our record: same purpose since 2019
- 2019Purpose: empower responsible decision making. Values: Fresh, Open, Present.
- 2022Vision: make enterprise intelligence accessible.
- 2025Turn enterprise signals into intelligence, decisions and autonomous action.
- 2026Agent run, human managed. Agents operate; people monitor, engineer and govern; outcomes prove it.
Purpose
Empower responsible decisions.
We wrote it in 2019, before generative AI. It matters more now. When agents can act, every decision and every action needs an owner.
Vision
Make enterprise intelligence accessible.
- Grounded
- Trusted
- Accountable
Intelligence is only accessible when it is grounded, trusted and accountable. Anything less is just available.
Mission
agent run, human managed
Agents run operations around the clock. People manage them through three planes, shared between you and us at the autonomy level you choose. Every outcome has a live metric.
- Operate01Agent run
Investigate, correlate, recommend and act, every hour of every day.
- Monitor02Human managed
Supervise what agents produce, validate results and handle routine exceptions.
- Engineer03Human managed
Design what agents do, build the tests they must pass and take the complex cases.
- Govern04Human managed
Set the boundaries, decide what should run on its own and own the outcome.
How the work is shared
Agents take on more. People never drop to zero.
Every task is agent run, human run or human managed. As a service matures, the human work moves from monitoring each finding, to engineering how agents work, to governing the limits they work within.
- Agent runAgents01
Explain what changed and why it matters.
- Human managedOur people02
Check every finding before it reaches you.
- Human runYou03
Make every decision and carry out the response.
Agent runHuman managed: people govern what agents doHuman run: people do the workIllustrative †
† Diagrams and worked examples use representative figures. Customer figures such as time to decision, time to contain and alerts handled are measured in a live customer environment and shared anonymized.
What we chose
Always-on outcomes, getting cheaper as they mature.
Operated outcomes for essential services of any size. Agents operate; people monitor, engineer and govern. You pay per agreed outcome and its measured progress—not hours, seats, licenses or lock-in contracts.
What we deliver
Operated outcomesWho we serve
Essential services of any sizeHow it works
Agents operate. People monitor, engineer and govern.How you pay
Per agreed outcome and its measured progressWhat you should feel
Clear enough to decide with confidenceWhy the model compounds
Our costs follow intelligence. Headcount doesn't drive them.
A traditional service provider adds people as it adds customers. A software vendor sells the tool and leaves you to run it. We do the operating, and agents carry the work that used to grow with headcount. Every outcome we run makes the next one cheaper to deliver.
-
events read a day
~5 bn
-
of data processed a day
~50 TB
-
between checks on every outcome
15 min
-
markets across Asia, the Gulf and the US
8
Values and principles
How we think, and how the operation behaves.
What one knows, all can use.
We write things down so a colleague or an agent can use them. You can see how every piece of intelligence was produced.
Every role serves a moment.
We work in your environment, your region and your moment of decision. There are no back-office roles.
- Responsible every consequential decision has a named owner
- Grounded evidence before opinion
- Consistent the same facts give the same answer
- Gated autonomy is earned
- Measured outcomes prove it
- Trusted relationships over transactions
Leadership
The people who manage what agents run.
Where we operate
Eight markets. Someone is always on.
- Singapore
- Philippines
- Cambodia
- India
- Hong Kong
- UAE
- Saudi Arabia
- USA
Teams in Singapore, Manila, Chennai and Phnom Penh, across time zones.
Frequently asked questions
What does "agent run, human managed" mean?
Agents operate: they do the operational work. People manage them through three planes. They monitor what agents produce, engineer how agents work and govern the limits agents work within. You choose how much of that sits with you and how much with us, and you own the consequential decisions. Every outcome is measured.
What is an AI-native service operator?
A company that runs the service for you, with agents doing the work. A software vendor sells you a tool to run. A service provider adds people as it adds customers. We take on the outcome itself, and agents carry the work that would otherwise grow with headcount.
Why "Human Managed"?
We named the company in 2018, before agents were part of the conversation. The hard part of intelligence was always going to be deciding what to do with it and answering for the result. That part stays with people.
Who are your customers?
Enterprises and institutions in essential services, from financial institutions to manufacturing and government. We have run cyber operations for a major ASEAN bank since 2018.
Where do you operate?
Singapore, the Philippines, Cambodia, India, Hong Kong, the UAE, Saudi Arabia and the United States. Our teams sit in Singapore, Manila, Chennai and Phnom Penh, so someone is always on.
What certifications do you hold?
ISO/IEC 27001:2022 and SOC 2 Type II. We are licensed by MISA in Saudi Arabia. Certificates and reports are available under NDA. See Security & compliance.
How do you price?
Per agreed outcome. We agree each outcome with you up front, with how progress is measured, and you can see that progress every 15 minutes. We never charge by gigabyte, log, seat or hour. See how outcome pricing works.
Which AI models do you use?
Whichever fits each task: frontier models, open-source models and classic machine learning. We change them as better ones arrive. Every answer is grounded in your live data and passes through gates before anything acts.
Do you work with MSSPs, MDR providers and other partners?
Yes. A provider can run its service on Human Managed, or bring us in for specific outcomes while it keeps the customer relationship. To talk about partnering, choose "Partnering with Human Managed" on the request form.
Can we buy through a cloud marketplace?
Yes. Human Managed is on Microsoft Azure Marketplace, so you can pay from your existing Azure commitment. See Marketplace.
How do investors, press and candidates reach you?
Investors and candidates: info@humanmanaged.com. Press: the press kit.