SYLLPA AI-native Solutions
Evidence · six measured cases

Applied AI and AI agents

Cases with measured numbers

Custom software with AI and AI automation, built by SYLLPA since 2024.

↓The factory and the six cases
How we work

AI agents execute. Human judgment directs.

Robust, secure and scalable software from day one, with far more testing, automated deployments and shorter delivery times.

Automated test casesMore than 28,900

Sum of five cases with published figures; the 168 test files of WE Captación are a different unit and are not included.

AI agents49
FORGE 24Our software factoryMatías 18Creative direction agentLegolas 7Legal agent

Spread across the software factory (24), the design factory (18) and the legal one (7).

77in-house test suites
4systems in productionIncludes pilots deployed in real environments.
22projects
+1,000hours of measured work
1,727closed tasks

Figures as of November 11, 2026

SYLLPA has been developing software since 2024. The AI-native factory began development in April 2026, reached operational stability in August 2026 and has offered its services to the market since then.

See how our AI-native factory works →
MatíasAI agentCreative direction agent
18specialized roles
16brands worked on
LegolasAI agentLegal agent
26inquiries handled
6areas of law
3companies with a case file
Quality and security

Security controls in our cases

Controls present in our cases; each project's controls are agreed in writing.

  • Security review on every changeBlocks credentials before they are saved
  • Independent reviewBefore every delivery
  • Role-based access
  • Each client's data kept separate
  • Audit log that cannot be deleted
  • Sensitive data encrypted
  • Second factorFor operators
  • Login attempt limits
  • Personal data protectedDetected before they reach the AI
  • Consent and data deletion
  • Automated deployments
  • Continuous improvementTests that watch every change
Verifiable resultPersonal data banks registered with the National Authority for Personal Data Protection, July 2026.
Cases

Our most notable cases with AI

In productionIn testingUnder constructionPlanned
01Education · School management
Case built for SafeSchool

SafeSchoolSchool management platform for Peruvian schools

From about 260 to more than 3,700 automated tests, in 3.5 months instead of about a year.

AI inside
  • New AI features in v2 Planned

HowSYLLPA rebuilt the core with its factory: AI agents under human direction, with tests that watch every change.

+See more

Attendance, assessments, report cards, treasury and communication with families. v1 already had AI features: assessment questions aligned to the curriculum and payments recorded by reading a photo of the receipt.

v1Occasional use of AI
v2 · SYLLPA factoryAI agents with human direction

Automated test cases

v1about 260
v2+3,700

Calendar time

v1about 1 year
v23.5 months

Today it covers the backend and the admin panels; the teacher, student and family apps and the AI features for users are planned.

02Higher education · Pilot
Success case

SadoskyPhysics tutor with AI for university students

The case that shows the speed SYLLPA's factory reached: 3 calendar days.

3dayscalendar days to build it
2,575automated test cases, approx.

Running in production as a pilot since November 4, 2026

AI inside
  • Up to three graded hints before the solution In production
  • Reads the photo of the problem In production
  • Proposes the grade; the teacher decides In production

HowWith SYLLPA's factory: AI agents under human direction.

+See more

Running as a pilot in a physics course at a public university in Lima; the product serves several institutions. Personal data typed in the chat is detected before it reaches the model.

03Civil construction · Payroll
Built for WorkEvolutionIn production since August 2026

WE NóminaWeekly payroll for civil construction

In production. It has begun to incorporate AI, and the agentic part comes next.

111calendar days
275hof measured effective work
13,753test cases, approx.
AI inside
  • Attendance proposed from a photo of the sign-in sheet In testing
  • Payroll assistant on your own PC In testing
  • Compensation written as a message Planned

HowEach criterion from the labor-law specialist was written as an automated test: the rule becomes a test.

+See more

Attendance photo reading, tested with 8 site photos (indicative figures): 150 of 150 rows detected and about 98% accuracy on exit time with the highest-precision model. A person confirms each draft.

04Fashion · Independent brands

Peruvian fashion e-commerceIndependent Peruvian brands in one place

You search the way you speak and find the garment; the brand uploads the photo and the AI does the rest.

HowWith SYLLPA's factory: AI agents under human direction.

+See more

In testing: redrawing the photo in catalog style, keeping color, cut and texture, and matching garments, checked by an automated judge before they are shown.

64calendar days
88hof measured effective work
3,565test cases, approx.
AI inside
  • Natural-language search: "up to 100 soles" In production
  • Description and category from the photo In production
  • Background removal with quality control In production
AaaS · Agentic system

Agents as a service

Complete systems where AI agents run the process end to end and people step in only where strictly necessary.

05AaaS · Agentic systemMarketing · Small businessesIn production

MarkeliaIntegrated marketing with AI

AI agents build your brand DNA, the monthly plan and the content pieces, and review every piece against your brand rules before delivering it. A person approves.

  1. AI agentBrand DNAUnderstands your brand without forms to fill in
  2. AI agentMonth's planContent calendar
  3. AI agentContent piecesImage, text and video in the brand's voice
  4. AI agentReviewAgainst your brand rules
  5. PersonApprovesBefore delivering
121calendar days
189.4hof measured effective work
5,359test cases, approx.

HowWith SYLLPA's factory: AI agents under human direction, with AI from Google and OpenAI.

+See more

You ask for changes in natural language and the system applies them. The system learns your preferences and proposes new rules.

06AaaS · Agentic systemHuman resources · RecruitingUnder construction

WE CaptaciónRecruiting and onboarding with AI agents

AI agents post the job opening, source candidates, pre-screen, interview and handle onboarding through to signature. Irreversible decisions are always made by a person.

  1. AI agentPosts the job openingThe role the hiring manager requestedPlanned
  2. AI agentCollectsApplicationsPlanned
  3. AI agentPre-screensProposes, never rejectsUnder construction
  4. AI agentInterviewsBy voicePlanned
  5. PersonDecidesEvery irreversible decisionUnder construction
  6. AI agentOnboardingThrough to signaturePlanned
9dayscalendar days for the agent platform
7daysactive
168test files

TodayThe platform that makes the agents safe is already built: permissions per capability, minimum data and irreversible decisions always human.

+See more

The recruiting agents plug into that platform in the next wave. No agent can do more than its permission declares or receive data it does not need.

SYLLPA AI-native Solutions

Applied AI, where it adds value and financial return.

Start a project ↗