Analysis
- The agents
- Eight automated specialists look at the same system from eight angles.
- The person
- Validates or corrects every artifact. The technical committee decides.
The people who sign.
AI is fast, never gets tired and never goes on holiday. But it cannot answer for what it delivers. That is what we are here for.
Five jobs, one rule: nothing ships without a signature.
The one who asks the awkward questions.
Analysis
Before touching a system you must be able to explain it: how it is built, how big it is, where it is at risk, what is worth doing.
- The agents
- They dig fast through code, dependencies and metrics.
- The person
- Picks what truly matters, challenges every conclusion and rewrites it so that a technical committee can decide.
AI prepares the case. People deliberate.
- X-ray of the system
- Risks and debt
- Options and costs
- Not only software
The one who doesn’t trust green tests.
Quality and security
A vulnerability or an anomaly is not closed until you know why it was there.
- The agents
- They find the cause in the code and propose the fix, with its checks.
- The person
- Verifies that the fix really closes the defect, without opening others, before it reaches production.
Tests and builds give evidence, not certainty. That is why there is a signature.
- From detection to cause
- A fix with its proof
- Stop rules
- A trail that stays
The one who gets Haicor talking to your stack.
Architecture and platform
Not a closed box: a platform configured on the customer’s frameworks, technologies and rules.
- The agents
- They work inside the declared boundaries: LOOM enforces them on every request, with any model.
- The person
- Turns what the customer decides into configuration: which systems, which stack, which releases, which models.
Models are swapped like components. The cycle, the rules and the memory stay yours.
- Repository scope
- Technology stack
- Release process
- Models and data
The one who writes the rules, just once.
Rules and knowledge
The rules that matter cannot live in someone’s head, nor in a vendor’s records.
- The agents
- They apply them at every generation, without forgetting them and without interpreting them their own way.
- The person
- Writes, approves and maintains them together with the customer: conventions, constraints, security and domain become Skills.
Software grows. With Skills, what the company can explain grows too.
- Team conventions
- Constraints and security
- Domain knowledge
- History of decisions
The one who keeps production running.
Production
With AI the work runs faster, but it is still a team of people that delivers it: the developers.
- The agents
- They propose changes and commands, in the developer’s repository.
- The person
- Each developer approves actions with effects before they happen. And the team answers for what it delivers.
The keyboard can be the AI’s. The Enter key stays ours.
- The development team
- Same rules for everyone
- Confirm before acting
- What is learnt goes back to LOOM
We don’t work alone.
In every module there are two kinds of intelligence, with different jobs: agents prepare the case, a person decides.
Coding
- The agents
- A team in parallel analyses code and dependencies, and runs the tests.
- The person
- Checks the proposal against the goal and the rules. Then signs, or rejects.
Fixing
- The agents
- They find the cause in the code and propose the fix, with its checks.
- The person
- Verifies that it closes the defect without opening others, and approves.
Shift
- The agents
- They measure versions and risks, upgrade and iterate until everything builds and passes the tests.
- The person
- Decides which jumps to make and in which order, and validates every step.
Skill
- The agents
- They apply the rules at every generation, without forgetting them.
- The person
- Writes them, approves them and keeps them current.
Born at Endow, not on a whiteboard.
Endow is an Italian software consulting and development company, founded in 2010. We design, build and maintain the software that keeps organisations running: business systems, portals, integration layers.
We work for public administrations and companies in regulated industries. These customers hold us to account for every line of code, and they have to: they answer to audits, regulations and citizens.
Haicor was born here, in our innovation lab: from the need to use AI speed without ceasing to answer for what we deliver.
We don’t believe in fully autonomous software. We believe in software someone answers for.
CertificationsUNI EN ISO 9001 qualityISO/IEC 27001 information securityLegality Rating two stars plusGender equality certifiedISO/IEC 42001 AI management · in progress
Talk to a human.
Write to us: a person answers. No bots.
The portraits are illustrations of the people on the team, published with their consent. We swap names on the first call.