MICRO · 1–10 PEOPLE
«I want to save hours this week, without hiring.»
Factory kit: templates, prompts and gates ready to use.
Reference from USD 49≈ MXN 900
Ask for the kit → Agentic product factory · Monterrey, MX
Agents drive. You set the direction.
Horme is the product factory where agents listen, prioritize, price, build and prove, and one person decides what to invest in and whether it worked. Real products in production, with quality, security and privacy gates at every step.
For companies of 50 to 500 people already using AI without a system. A 2-week diagnostic at a fixed price, credited to the project.
“3 days”: SERENA case, our own product, November 2026. “Pay per delivery”: each delivery is paid when it passes the quality review and arrives with its evidence.
Almost half of Mexican companies already use AI. Very few have a system.
48%
of companies in Mexico use AI.
13%
use it at an advanced or agentic level.
46%
have no reliable way to measure the return.
40%+
of agentic AI projects will be canceled by 2028 over cost, unclear value or weak controls.
Sources: AWS and Strand Partners, «Unlocking AI's potential in Mexico 2026»; Gartner, June 2025.
Who it's for
A clear first step, at a fixed price, sized to you.
SMALL · 11–49 PEOPLE
«What should I automate first so it actually pays off?»
A governance audit of what you already have, or a 2-week diagnostic with a costed plan.
Reference from USD 1,900≈ MXN 35k
Book 15 min → MID-SIZE · 50–500 PEOPLE
«We already use AI. We need a measurable, secure system.»
An agentic diagnostic in 2 weeks, credited to the project. Then a product in 6 weeks and a monthly cycle.
Reference from USD 2,500≈ MXN 46k
Book your diagnostic → ENTERPRISE · 500+ PEOPLE
«We need governance before scaling agents.»
A governance audit or a scoped pilot alongside your team, with a Trust Pack: governance, contract and code ownership.
Price By scopeScoped pilot
Let's talk → Reference prices, before tax. The final price is confirmed in the proposal based on scope; peso equivalents are approximate. You pay per delivery, not by the hour: each delivery is paid when it passes the quality review and arrives with its evidence.
What you buy
Five services to decide, change and build.
Each service is a delivery with a price and evidence. Below, the order in which they’re used.
DECIDE · strategy and plan
01Agentic diagnosticAgents research, rank and cost your options. You choose one and leave with an approved plan.Step 1 · DiagnosticSee how decisions are made →CHANGE · operations with agents
02Product management with agentsA weekly cycle: signals, a ranked and costed backlog, and you decide what’s next.Step 3 · Continuous cycleSee the cycle →03A factory with 5 reviewsShort specs, continuous delivery and 5 reviews before production. Every delivery arrives with its evidence.Steps 2 and 3 · every deliverySee the 5 reviews →How we start
Diagnostic, pilot and continuous cycle. One step at a time.
- Step 1: 2 WEEKS · FIXED PRICEDiagnosticWe review processes, data, systems and governance, and price every option.You leave with a Decision Brief and the plan for the first pilot.Service 01 Agentic diagnostic
- Step 2: 4–6 WEEKSPilotWe build the option you chose in production and measure the result.You leave with a live product, its evidence and the result against what was expected.Services 04 A product in 6 weeks, 03 A factory with 5 reviews
- Step 3: MONTHLYContinuous cycleOne improvement at a time: every week you decide what’s next, and every month you see whether it worked.You get a ranked backlog every week and measured deliveries.Services 02 Product management with agents, 03 A factory with 5 reviews
Want your team to run it on its own? 05 Your own factory can join at any step.
Solutions
Six products to start with, and one of our own.
Processes that cost hours, errors or customers today, turned into products. Agents do the heavy lifting; a person decides what gets built and checks whether it worked.
- 01Finance
Month-end close without manual entry
Matching invoices, payments and orders takes days and depends on two or three people.
What the agents do Match invoices, payments and orders, and flag only what doesn’t add up.
Man in the loop Finance approves every exception before it’s booked.
- 02B2B sales
Quotes in minutes, not days
Every quote is built by hand with prices and stock from several systems.
What the agents do Build the quote with prices, stock and the customer’s terms.
Man in the loop The sales rep reviews, adjusts and sends it.
- 03Customer service
Answers with full context
The team searches three screens for the history before replying.
What the agents do Classify the request, gather the history and draft a reply.
Man in the loop A person approves sensitive cases and refunds.
- 04Operations
The weekly report builds itself
Data lives in the ERP and spreadsheets; the report arrives late.
What the agents do Gather the data, spot deviations and write the summary.
Man in the loop Leadership decides what to do about each deviation.
- 05Product and IT
Modernize a legacy application
No one fully understands the system, and migrating it looks like a year-long project.
What the agents do Analyze the code, inventory the features and map the risks.
Man in the loop Your team decides what moves first and approves every release.
- 06Human resources
Leave and payroll questions answered
HR answers the same questions every day.
What the agents do Answer with your policies and escalate what isn’t clear.
Man in the loop HR reviews exceptions and payroll answers.
- 07Our own solution · Corporate wellbeing
SERENA: wellbeing you can measure
Companies invest in wellbeing without knowing if it works.
What the agents do Guide the live session, collect anonymous arrival and exit surveys, and build the report.
Man in the loop The instructor runs the session; the company only sees averages of 5 people or more.
The first six are examples to guide the conversation; real impact is costed in your diagnostic and proven at the outcome review. SERENA is our own product, built with the factory; its outcome is measured for the first time at the Nov 18 pilot.
How we create products
A product factory: from signal to metric, and back.
Agents listen, prioritize, price, build and prove. One person decides twice: what to invest in and whether it worked. Every step leaves evidence, and what gets measured is listened to again.
Done by agentsDecided by a person
Decide
- 01ListenSix agent roles watch, 24/7:
- Product & growth
- Experience (UX)
- Engineering
- Quality
- Trust & privacy
- Risk
- 02PrioritizeEverything enters one backlog, ranked by value and strategic fit:
- Strategy
- Customers
- Tech debt
- Data
- 03Price the optionsEvery option comes with three numbers:
- Cost · people, agents and infrastructure
- Weeks · when your users see it
- Expected value · the metric that must move
- INVESTMENT DECISIONYou pick one optionA one-page Decision Brief: options, cost, weeks and expected value. You choose; the agents don’t.Man in the loop
Deliver and prove
- 04SpecifyThe chosen option becomes a bounded delivery:
- One-page spec
- Stories
- Acceptance criteria
- 05Build through 5 reviewsTests first and a guardian agent. A person decides two reviews: the spec (G0) and production (G4), with a rollback plan.
- G0
- G1
- G2
- Guardian
- G4
- 06Prove the outcomeThe metric agreed at the decision, measured in production. Everything goes into the evidence file.
- Before and after
- Against expected
- Evidence file
- OUTCOME REVIEWYou review the outcomeReal impact against expected. You continue, adjust or stop.Man in the loop
Step 06 results become step 01 signals. The cycle doesn’t end: every delivery teaches what to build next.
Quality before release
5 reviews before production
Nothing reaches production without passing these five reviews. A person makes two of the calls.
- 1 Idea approved (G0) A person approves the one-page spec: what gets built and how it is measured. Man in the loop
- 2 Tests first (G1) Tests are written before the code, from the acceptance criteria. Reviewed by agents
- 3 Automated quality, security and privacy review (G2) Quality, architecture, UX, accessibility, privacy and security. Reviewed by agents
- 4 Final review by the guardian agent A guardian agent (Gartner’s term) reviews independently. If it says STOP, nothing ships. Reviewed by agents
- 5 Your approval (G4) A person approves the move to production, with a rollback plan. Man in the loop
With every delivery
An evidence file, not a report.
- What was askedDecision Brief, epic, stories and acceptance criteria.
- How it was builtTests, reviews G0 to G4, the guardian agent’s verdict and the rollback plan.
- What happenedMetrics before and after, against what was expected. If it fell short, we say so.
You pay per verified delivery: each one is billed when it passes review and arrives with its evidence file.
Who does what
Three roles, one factory.
- AgentsProduct engineeringShape, build, test and ship every delivery.
- PersonForward-deployed engineeringWorks with you inside your operation: understands the problem, decides at the gates and owns the outcome.
- AgentsAI platformThe agents, their permissions, their gates and their log, under the 7 governance pillars.
Roles follow the evolution of AI software teams described by Gartner (2025).
Case
OWN PRODUCT · CORPORATE WELLBEING
SERENA: our own product, built with the factory.
A bilingual site and SERENA Studio: Practice, QR-based Events with anonymous surveys, and Studio Admin to lead live sessions. All on Cloudflare, private by design: only averages of 5 or more people.
Outcome review pending: the business outcome is measured at the November 18 pilot, and we’ll publish it as it is.
Explore SERENA StudioRead the full case: step by step and delivery metrics (DORA) →5
PRODUCTS
4 already in production
11
AGENTS WITH A ROLE
1 person decides; agents execute
7
PLATFORMS
Web, iPhone, iPad, Android, watch, Mac and Windows
27
PRODUCTION DEPLOYS
In 3 days, each with a rollback
≥ 5
PEOPLE PER AVERAGE
Privacy by design: nothing individual
2
LANGUAGES
Site and apps in Spanish and English
Platforms per product: Studio 7 · Studio Admin 4 · Wellbeing 3 · SERENA Admin 3 · Event screen 3. Today they run as an installable web app (PWA); native apps arrive in stages. Source: SERENA log, Nov 3–8, 2026.
SERENA by product
What the factory built, up close.
Six SERENA products, from the website to the event screen. Screens from a test event, with no personal data.
Screens from SERENA’s visual tests, November 2026. Bracketed fields hide client names. Interface in Spanish.
Command center
Everything in view, in one place.
One person can run the factory because they see, in real time, what to decide, what’s being built, what’s next and what’s at risk.
Command center views with real data from the SERENA project, November 6–8, 2026. Tap a view to enlarge it. Client names and platforms under study are hidden; in Risks, placement on the matrix is illustrative. Interface in Spanish.
Founder
Juan Zacarías
Product leader and digital product architect, experienced in building omnichannel products at scale. I designed Horme so one person with agents can deliver with the discipline of a full team.
People first.
Agents propose; a person decides at every gate.
Build with purpose.
We only build what has an approved success metric.
Grow together.
Your team learns the method and keeps it.
FAQ
What people ask before starting.
Grouped by company size. If your question isn’t here, send us a message or book 15 minutes.
Micro · 1–10 people
How much does it cost per month?
The entry point is the factory kit, with a reference price from USD 49 (≈ MXN 900), before tax. The monthly continuous cycle is quoted by scope.
What result will I see in the first week?
You pick one repetitive task and try it with the kit’s templates. That same week you know how many hours it saves you; if it doesn’t, you know that too.
Small · 11–49 people
What should I automate first?
Whatever costs the most hours with the least risk. The 2-week diagnostic ranks your options by value, cost and time, and you choose one.
What’s the return?
We don’t promise it before measuring. The diagnostic estimates each option’s potential; the pilot measures the actual result against the expected one and shows it to you.
Is my data safe?
Your data stays in your systems and your cloud. Each agent gets only the permissions it needs, and security and privacy are reviewed before every release.
Mid-size · 50–500 people
Does it integrate with our ERP?
Usually yes, if the ERP has an API or exports. The diagnostic reviews your systems, and the integration goes into the plan with its cost and timeline.
Who owns the code?
You do. It lives in your repository and your cloud from day one.
What happens if something fails in production?
Every release has a rollback plan: we go back to the previous version. We also publish our own factory’s change failure rate, so you can see how we measure it.
Do you have clients besides SERENA?
SERENA is our own product and our public case; its first pilot with a client is on November 18, 2026. We’re opening our first diagnostics and publish each case only with the client’s permission.
Enterprise · 500+ people
Do you comply with Mexico’s data protection law (LFPDPPP)?
We work with a privacy notice, minimal data and one database per client. The contract defines who is the data controller and who is the processor. SERENA only shows averages of 5 people or more.
What contract and service level do you offer?
A contract covering scope, deliverables, confidentiality and code ownership, plus the governance Trust Pack. The service level is agreed per project.
If it’s one person, what happens if they’re not around?
The code and documentation are yours, live in your repository and your cloud, and any team can pick them up.
Ready to move faster? See if we’re a good fit.
No pitch, just a conversation.
Someone who has done this before, on your side of the table.