Puente OS vs Palantir: The Same Approach, Brought to Latin American Operations

Palantir is the software company that builds platforms to integrate an organization's data, decisions and operations, with engineers who work alongside the customer. Puente OS is the platform that builds the systems that run your business: apps, workflows, databases and agents that read from your systems, cross-reference the information none of them holds in full, decide on your rules and execute the action inside them.

I'm Cris Ugarte, CEO and co-founder of Puente OS. We work with retail, eCommerce and logistics companies in Chile, Peru and Mexico, and when we explain how we work, with a team that sits next to the person who runs each process and builds the system that runs it, a comparison we often hear is with Palantir. It is a comparison we're honored by: Palantir defined this approach and takes it to an admirable scale.

Here I go through what each one does, with Palantir's public sources for everything I say about it, and when each one makes sense.

What is Palantir and what does it solve?

Palantir presents itself with the line "Foundational Software of Tomorrow. Delivered Today.", and its home page sums up its work as software that empowers organizations to integrate their data, their decisions and their operations. Its retail page says it was founded in 2003, got its start in intelligence and defense, and applied what it learned from its frontline deployments to many industries and commercial enterprises around the world. Today, according to its home page, it is in over 50 sectors and industries.

What it offers, according to its website and its documentation:

• Foundry. Its commercial platform presents itself as "The Ontology-Powered Operating System for the Modern Enterprise," and its Foundry page sums it up as "Run your business as code." There, Palantir explains that for over a decade it has embedded alongside its customers to build Foundry backwards, starting from the most critical operational decisions.
• The Ontology. It is the heart of Foundry. Its documentation defines it as an operational layer for the organization, and its core concepts build it from clear pieces: object types, which represent real-world entities or events; properties, which describe their characteristics; link types, which represent the relationship between two objects; and action types, which group the changes a person can make at once. According to its why create an Ontology page, the Ontology represents the decisions in an enterprise, and its objects, properties and links evolve in real time.
• Actions that write to your systems. Actions change objects in a single transaction, with checks on who can run them and side effects like notifications. With webhooks, an action sends the decision to an external system, like a CRM or an ERP.
• AIP, AI on top of the Ontology. Its Artificial Intelligence Platform connects language models to the enterprise's data, logic and actions: the Ontology anchors AI in the operational truth of the business, synchronizes decisions back to ERPs and other systems, keeps AI-authored proposals subject to human validation and records every action for auditing. Its AIP documentation adds pieces like AIP Logic, a no-code environment for building, testing and releasing functions powered by language models.
• Agents. The Agent Stack is its infrastructure for building, running and observing agents, and AIP Chatbot Studio, formerly AIP Agent Studio, builds assistants equipped with enterprise information and tools. Palantir even has an AI FDE, an agent that operates Foundry through conversational commands: it transforms data, manages code repositories and builds the Ontology.
• Apps and automations. With Workshop, teams build interactive applications for operational users, and with Automate they define conditions, on a schedule or on changes to objects, that run actions, AIP Logic functions and notifications.
• Data and integrations. Its connector catalog covers filesystems, databases, data warehouses, ERPs and CRMs like SAP, NetSuite, Odoo and Salesforce, commerce platforms like Shopify, WooCommerce and Adobe Commerce, streaming and IoT, plus a generic REST APIs connector. With Virtual Tables, AIP works with data where it already lives, in BigQuery, Snowflake, S3 or Azure.
• Apollo and Gotham. Apollo is its continuous deployment platform, with full rollbacks, for public and private clouds, air-gapped networks and the edge. Gotham, according to its home page, is its operating system for global decision making.
• AIP Bootcamps. Its bootcamp is a five-day hands-on workshop where the customer goes "from 0 to use case": it learns how to apply AI to mission-critical operations, develops its first use cases in the software and trains its users for rollout.
• Security. According to its home page, Palantir carries certifications or attestations for demanding standards like FedRAMP and IL5, and AIP governs human and AI activity with role-, classification- and purpose-based access controls, with auditing of every operation.

Palantir's Forward Deployed Engineers

The role the whole AI industry is now hiring for was born at Palantir. Its Forward Deployed Software Engineer posting says Palantir pioneered the position, embedding engineers directly with its customers to tackle their most pressing challenges: they understand the customer's problems, architect and build solutions with its business-critical data and AI, develop custom applications and work with the customer from technical teams to executives, traveling to client sites as needed.

Its careers page organizes the work into three overlapping roles: Echos, who decompose the customer's problem into its workflows and solve backwards from each user's workflow; Deltas, the team of Forward Deployed Engineers, who make sure what gets built actually works and delivers tangible outcomes; and Devs, who turn what was found into products. Its engineering blog already explained it in Dev versus Delta: the product engineer builds one capability for many customers, and the forward deployed engineer takes one customer and combines many capabilities. We wrote more about the role in how to become a forward deployed engineer.

Who it serves and where it operates

Palantir works with some of the world's most important institutions. Its Foundry page features cases in aviation, energy, healthcare, supply chain and financial services, and its retail page says top global enterprise retailers use Foundry for their digital transformation: inventory, supply chain, store operations, pricing and omnichannel experiences. In Latin America, its press release with Grupo Globo, the region's largest media company, says Globo has used Foundry since 2019, and that in Brazil Palantir also works with airlines, financial institutions and government agencies.

A model of the whole organization, a platform that turns it into decisions and actions, and engineers who work alongside the customer until it works: that is how Palantir operates, and it does so at a scale few companies in the world reach.

What is Puente OS and what does it solve?

Puente OS is the platform that builds the systems that run your business. Your ERP records what happened. Your BI shows you what happened. Puente makes it happen. In your stack it is the system of action: it reads from your systems, cross-references the information none of them holds in full, decides on your rules and executes the action inside them.

What it builds comes from four pieces:

• Apps where your team reviews, decides on and approves the exceptions.
• Workflows with explicit, auditable rules that start on their own when data changes or at a set time.
• Databases that bring together context from several systems: the order, its receipt, its shipment and its settlement in a single record.
• Agents for the steps that take judgment, like an ambiguous complaint or a discrepancy that needs investigating.

With Puente OS you describe in natural language what your operation needs and get software running: apps, workflows, databases and agents. Puente's Growth Ops team designs each process with you, builds it and maintains it, and each one reaches production in 2 to 4 weeks. Puente acts on your systems, and where the operation lives in spreadsheets, it also keeps the record.

The discipline behind it is autonomous operations design: designing your operation so the repeatable work runs on its own, on your rules, and your team decides what takes judgment. For a closer look at the layer, see what a system of action is.

What do Palantir and Puente OS have in common?

They share two convictions: the software that runs a business is designed alongside the people who do the work, and it is built on a model of how that business works. Each one applies them at its own scale.

A team that works inside your operation

At Palantir they are the Forward Deployed Engineers. At Puente OS, that role is called Growth Ops, and it works in three phases: the Diagnosis maps and prioritizes the processes, the Sprint puts one running in production in 2 to 4 weeks, and the Partnership operates it, measures it and adds the next one.

The mapping starts with two questions, asked next to the person who runs the process: "Show me how you do this, step by step" and "What happens with the exceptions?" The first reveals the real process, and the second, the complexity the written procedure hides.

A model of how your business works

Palantir's Ontology represents the whole organization as objects, links and actions. At Puente, the model is built process by process. For each one, Growth Ops writes down its trigger, the information it cross-references, the rules for the clear cases and the owner of each exception, and classifies it with our process taxonomy, I/T/O/C/BP: where the data comes from (input), what gets done with it (transformation), how the result is delivered (output), which integrations it needs and which type of business process it belongs to, out of ten: orders, pricing, customer communication, catalog, finance, sales, marketing, logistics, inventory and operational intelligence.

We built it from over 200 meetings with commerce companies, and it shows that most processes are variations on the same patterns. That is why each new process starts from what we have already seen at similar companies, and Puente's Databases keep the order, its receipt, its shipment and its settlement together, ready for the process's workflows, agents and apps to read and act on.

When the model, the rules and the execution live in different tools and with different people, coordinating them eats up your team's hours: that is the coordination tax. An autonomous operation meets three conditions: work starts on its own when data changes, your systems talk to each other with no person in the middle, and the system executes the clear cases and brings you what takes judgment. Puente OS delivers all three together, process by process, on a single platform.

How do Puente OS and Palantir compare, dimension by dimension?

Dimension by dimension:

• Who it serves. Palantir: governments and large organizations in over 50 sectors, with cases in aviation, energy, healthcare, finance, supply chain and global retail. Puente OS: mid-sized retail, eCommerce and logistics companies in Chile, Peru and Mexico.
• What it delivers. Palantir: platforms, Foundry, AIP, Apollo and Gotham, on which the customer and Palantir's engineers build the applications and workflows of its operation. Puente OS: the process working end to end, with its apps, workflows, databases and agents on a single platform, in production.
• The business model. Palantir: the Ontology, with object types, properties, links and actions that represent the whole organization and that apps, automations and agents read and write. Puente OS: a model per process, with its trigger, its information, its rules and its exceptions, classified with the I/T/O/C/BP taxonomy; the Databases keep the context that spans systems, and the process's apps, workflows and agents read and write it.
• Who implements. Palantir: its Forward Deployed Engineers, embedded with the customer, together with Echos and Devs, and the customer's technical team, which builds on the platform with its tools and its SDK. Puente OS: the Growth Ops team builds each process with you and answers for it. You describe in natural language what your operation needs and get software running, and every activation or irreversible change goes through your confirmation.
• How you start. Palantir: a five-day AIP Bootcamp, where the customer goes from zero to a use case and trains its users. Puente OS: mapping alongside the person who runs each process, where your processes are prioritized and the first one reaches production in 2 to 4 weeks; after that, the next one is added on the same platform.
• Integrations. Palantir: a broad connector catalog for databases, data warehouses, global ERPs and CRMs like SAP, NetSuite, Odoo and Salesforce, commerce platforms like Shopify and WooCommerce, and a generic REST APIs connector. In that catalog, as of October 13, 2026, there are no connectors for Mercado Libre, Bsale, Defontana, Nubox, Falabella or the SII (Chile's tax authority); those services connect through their API with the generic connector. Puente OS: native connectors for Mercado Libre (orders and shipping details), Bsale (stock and documents), Multivende, WooCommerce, WhatsApp, Gmail and Google Sheets, among others. The Growth Ops team connects the rest, like the SII, your invoicing system or your ERP, through its API, over HTTP or Python, or by reading the ERP's database.
• Apps. Palantir: Workshop for operational applications on the Ontology, with a suite of out-of-the-box applications and an SDK for developers. Puente OS: web apps tailored to each process, hosted on Puente with their own link, where your team reviews, decides on and approves the exceptions, like a discrepancy inbox for finance or a stalled-shipments dashboard for operations.
• AI and agents. Palantir: AIP connects language models to the Ontology, with AIP Logic for no-code functions, the Agent Stack to build, run and observe agents, AIP Chatbot Studio for assistants and an AI FDE that operates Foundry through conversation. Puente OS: the AI is in how the whole system gets built, in natural language, and agents take the steps inside each process that call for judgment, like reading a complaint or investigating a discrepancy.
• Work that runs on its own. Palantir: Automate runs actions, functions and notifications when a condition is met, on a schedule or on changes to objects, and actions send decisions to external systems through webhooks. Puente OS: workflows that start from a webhook or a schedule, with explicit rules, iterators and per-step failure policies, and that write back to your systems: labels, statuses, prices and documents.
• Versions and deployment. Palantir: Apollo deploys its software continuously to public and private clouds, air-gapped networks and the edge, with full rollbacks. Puente OS: every workflow save creates a new version, and rolling back means activating the previous one; in apps, every publish is saved as a version, with earlier ones kept as history. Irreversible actions, like issuing a tax document, are designed in two steps: a draft and a person's confirmation.
• Security and governance. Palantir: certifications or attestations for standards like FedRAMP and IL5, role-, classification- and purpose-based access controls, and auditing of every operation; AI-authored proposals can be subject to human validation. Puente OS: each app is assigned the tables its process uses, with read, write and delete permissions; and every execution is recorded with the step where it stopped and, per step, the error, the traceback and the logs. Puente manages the platform where everything runs.

What does it look like in a real operation?

It looks like processes running end to end on each customer's systems, on its rules, with a team that maintains them. Two Puente OS customers:

Example 1: a multichannel electronics seller in Mexico, where every order goes out with its label on its own

An electronics seller in Mexico runs without an ERP, and Puente has seven of its sales channels connected. Every order follows its path on rules: the address is validated, the carrier is chosen by zone, cost and insurance, the label is generated and the tracking number is written to the order. What reaches a person goes into a support log, where every incident is recorded with its resolution.

Of the 38,628 orders processed between November 2025 and August 2026, 97.9% got their shipping label with no human intervention: 37,815 labels. Generating the labels for a cutoff went from 5 to 7 hours to 15 minutes, and monthly volume grew 40% with the same operations team. From August 1 to 12, 12 of 2,051 orders had an incident, 0.59%, with a median resolution time of 10 minutes. In all, that is 20 apps, 75 processes, 7 channels and 19 APIs, built in ten months by one person.

Example 2: a pet retailer in Chile, where every new process arrives faster

A pet retailer sells on four channels. Its OMS (order management system) on Puente processed 56,597 orders between January and August 2026, and every shipment is measured against the delivery promise: 33,192 shipments measured and 85.0% on-time delivery among those not canceled, calculated against the courier's API. A workflow calculates the margin of every sales line each day: as of August 2026 it had covered 247,626 lines, and 3,830 of them, 1.5%, showed a negative margin that nobody saw before. Pricing decisions also go back to the store: 8,362 price changes executed in its online store from Puente.

It runs 10 apps in production for six departments, with 167 versions deployed, built and maintained by one person. The time from requesting a new process to having it in production dropped from 105 days, on the first project, to 7.

Why choose Puente OS?

Because you get the approach Palantir made famous, engineers in your operation and a system that reads, decides and acts on your business, designed for a mid-sized commerce or logistics company in Latin America. Palantir is a global benchmark in data and AI platforms for large organizations; Puente OS delivers your business process working, with the Growth Ops team that designs it with you, builds it and maintains it. Five concrete reasons:

• A close team that designs around your processes. Growth Ops sits next to the person who runs each process, in your language and in your operation, and maps what triggers it, what it decides and what its exceptions are before building. Each process starts from patterns we have already seen in over 200 meetings with commerce companies.
• Each process in production in 2 to 4 weeks. Each new process builds on what is already there: at the pet retailer in Example 2, the time from requesting a process to having it in production dropped from 105 days to 7.
• Connected to the systems of Latin American operations. Puente has native connectors for Mercado Libre, Bsale, Multivende and WooCommerce, and the team connects your invoicing system, your courier or your ERP through its API, over HTTP or Python, or by reading the ERP's database. What gets decided goes back to your systems: the pet retailer has 8,362 price changes executed in its online store from Puente.
• You start process by process, and the operation grows on the same platform. You start with the process that hurts most and add the next one once the first is delivering. That is how the pet retailer reached 10 apps in production for six departments.
• One person sustains a whole customer's software operation. You describe in natural language what your operation needs and get software running: apps, workflows, databases and agents. At the electronics seller in Example 1, one person built 20 apps and 75 processes in ten months, and 97.9% of orders go out with their label with no human intervention.

That way your operation runs on your rules, and your team spends its time deciding what takes judgment.

When does each one make sense?

Palantir fits when you are a large organization, with critical operations at global or public scale, and you want to model the whole organization on a single platform. Puente OS fits when you are a mid-sized retail, eCommerce or logistics company in Latin America and you want your processes working end to end, one at a time, with a team that builds them with you and maintains them.

• Palantir, if you operate at the scale of a government, a defense organization or a large global corporation: supply chains across several continents, hospital networks, airlines, industrial plants or financial services with high regulatory demands. Also if you need to deploy to private clouds, air-gapped networks or the edge, with certifications like FedRAMP or IL5, or if you want an Ontology that unifies data from the whole organization and an in-house technical team that builds on it. Its scale, the depth of its Ontology, AIP and the experience of its Forward Deployed Engineers are real strengths.
• Puente OS, if you sell through marketplaces, an online store and physical channels, work with Mercado Libre, Bsale, your invoicing system or your ERP, and want your order, shipping, reconciliation, pricing or customer service processes running on your rules. Also if you want to start with one process, see it in production within weeks and add the next one on the same platform, with a team that designs it with you, maintains it and responds when something fails.

How do you start?

With a process that hurts, with volume and manual handoffs: generating shipping labels, reconciling a marketplace or matching documents. With Puente OS, each process reaches production in 2 to 4 weeks.

1. Pick the process. With volume, manual handoffs and a visible cost when it fails.
2. Take the baseline. Cases per month, hours of manual work per week and the share of cases that come out right the first time.
3. Map it with the person who runs it. Ask them to show it to you step by step, on their screen, and ask what happens with the exceptions.
4. Connect the sources. Your ERP, your store, your marketplaces, your invoicing system, your courier and your bank statements.
5. Write down the rules your team applies from memory. For example: "if the marketplace settlement differs from the sale by a known commission, it adjusts on its own; if the difference is something else, it goes to finance with the order, the receipt and the payment in view."
6. Decide where each exception goes. The system resolves it, or it reaches a named person with full context in an App.
7. Measure from the first week and add the next process on the same platform.

That is autonomous operations design applied process by process: the repeatable work runs on its own, on your rules, and your team decides what takes judgment. To choose which steps go to a rule and which to an agent, this guide on how to implement AI in your business can help, and if you're weighing a build with your own or an outside development team, see Puente OS vs custom middleware.

We believe every company in Latin America can be operationally autonomous, and that starts with having every process running end to end, with its rules and a team that keeps it running.

Key takeaways

• Palantir builds Foundry, AIP, Apollo and Gotham for governments and large organizations, and its Ontology represents the organization as objects, properties, links and actions that apps, automations and agents read and act on.
• Palantir created the Forward Deployed Engineer role: engineers embedded with the customer who understand its problem and build the solution with its data and AI. Its entry point for new customers is a five-day AIP Bootcamp.
• Palantir and Puente OS share the approach: a team that works inside the operation, and a model of the business that the software decides and acts on.
• Puente OS brings that approach to mid-sized retail, eCommerce and logistics companies in Latin America: Growth Ops designs around your processes, with the I/T/O/C/BP taxonomy, and Puente builds apps, workflows, databases and agents connected to your systems.
• Puente has native connectors for Mercado Libre, Bsale and Multivende, and Growth Ops connects the rest via API, such as the SII, your invoicing system or your ERP: at an electronics seller in Mexico, 97.9% of orders go out with their label with no human intervention.
• Palantir fits large organizations with critical operations at global or public scale. Puente OS fits when you want your processes running end to end, in 2 to 4 weeks per process.

---

Want to see how a Growth Ops team works inside your operation? Let's talk about your processes, and we'll show you Puente OS running with cases like the ones in this article. Book a conversation with our team.

Frequently asked questions

What is Palantir and what is it used for?

Palantir is a software company founded in 2003 that builds platforms to integrate an organization's data, decisions and operations. Its platforms are Foundry, for data and operational applications on the Ontology; AIP, to connect AI and agents to that data and those actions; Apollo, to deploy software continuously; and Gotham, for global decision making. It works with governments and large organizations in over 50 sectors, with engineers who embed with each customer.

What is Palantir's Ontology?

According to its documentation, the Ontology is the organization's operational layer. It represents real-world entities and events as object types with their properties, the relationships between them as link types, and the changes a person can make as action types. Foundry and AIP apps, automations and agents read and write on that model, and actions can send decisions to external systems like an ERP or a CRM.

What is a Palantir Forward Deployed Engineer, and how does it compare to Growth Ops?

It is the engineer Palantir embeds with its customers to understand their problems and build the solution with their data and AI, a role Palantir itself created. At Puente OS, that role is called Growth Ops: it maps the processes alongside the people who run them, builds the system in natural language, puts it in production in 2 to 4 weeks and operates it together with the customer's team. The details of the role and a 30-day plan to get there are in how to become a forward deployed engineer.

Does Palantir integrate with Mercado Libre, Bsale or the SII?

In the connector catalog in its documentation, as of October 13, 2026, there are no connectors for Mercado Libre, Bsale, Defontana, Nubox, Falabella or the SII. The catalog includes a generic REST APIs connector, which connects any service that publishes its API. Puente OS has native connectors for Mercado Libre (orders and shipping details), Bsale (stock and documents) and Multivende, and the Growth Ops team connects the rest via API.

Does Puente OS build a model of my business, like the Ontology?

Puente builds the model process by process. For each one, Growth Ops maps what triggers it, what information it cross-references, which rules it applies and who owns each exception, and classifies it with the I/T/O/C/BP taxonomy. Puente's Databases bring together the context that spans systems, like the order, its receipt, its shipment and its settlement, and the process's apps, workflows and agents read it and act on it. Each new process is added on the same platform.

How does Puente OS get started, and how long does the first process take?

By mapping your processes. The Growth Ops team maps and prioritizes them alongside the people who run them, connects your systems, writes the base rules and puts the first process live in production in 2 to 4 weeks. Each following process reaches production in the same timeframe and builds on the integrations already running.

External sources cited

Everything this post says about Palantir comes from its website, its documentation, its careers page, its job postings, its blog and its newsroom, reviewed on October 13, 2026. Puente OS customer figures are production measurements as of August 2026; label generation time and the request dates for each process come from follow-up with the customer.

Palantir: company and platforms

• Palantir, home page: the line "Foundational Software of Tomorrow. Delivered Today."; software to integrate data, decisions and operations; Foundry as the operating system for the modern enterprise, Gotham as the operating system for global decision making and Apollo as the operating system for continuous delivery; presence in over 50 sectors and industries; certifications or attestations like FedRAMP and IL5; AIP Bootcamps.
• Palantir Foundry: "The Ontology-Powered Operating System for the Modern Enterprise" and "Run your business as code"; the Ontology as the heart of Foundry; over a decade embedded alongside customers to build Foundry from the most critical operational decisions; cases in aviation, energy, healthcare, supply chain and financial services.
• Palantir AIP: the Ontology connects AI to the enterprise's data, logic and systems; synchronization of decisions back to ERPs and other systems; AI-authored proposals subject to human validation and auditing of every action; role-, classification- and purpose-based access controls; application suite, SDKs and Virtual Tables on BigQuery, Snowflake, S3 and Azure.
• Palantir AIP Bootcamp: "From 0 to use case in 5 days"; applying AI to mission-critical operations, developing the first use cases and training users.
• Palantir Apollo: continuous deployment and full rollbacks across multi-cloud, hybrid cloud, private SaaS, air-gapped networks and the edge.
• Palantir for Retail: founded in 2003; got its start in intelligence and defense and applied its frontline deployment learnings to industries and commercial enterprises; top global enterprise retailers use Foundry for inventory, supply chain, store operations, pricing and omnichannel.
• Palantir and Grupo Globo Extend Digital Transformation Partnership (Palantir newsroom): Grupo Globo, Latin America's largest media company, has used Foundry since 2019; in Brazil, Palantir also works with airlines, financial institutions and government agencies.

Palantir: documentation

• Ontology overview: the Ontology as an operational layer for the organization; action types and functions as its kinetic elements.
• Ontology core concepts: definitions of object type, property, link type, action type, function and interface.
• Why create an Ontology?: the Ontology represents the decisions in an enterprise; objects, properties and links that evolve in real time.
• Action types overview: changes to objects in a single transaction, checks on who can run the action and side effects like notifications and webhooks.
• Webhooks (action types): sending a request to an external system, like a CRM or an ERP, before or after the Ontology changes.
• AIP overview: AIP connects AI with data and operations; AIP applications, including AIP Logic, AIP Chatbot Studio and AI FDE.
• AIP Logic overview: a no-code environment for building, testing and releasing functions powered by language models.
• Agent Stack: infrastructure and product suite for building, running and observing agents.
• AIP Chatbot Studio overview: assistants equipped with enterprise information and tools; formerly AIP Agent Studio.
• AI FDE overview: an agent that operates Foundry through conversational commands to transform data, manage repositories and build the Ontology.
• Workshop overview: interactive applications for operational users.
• Automate overview: conditions on a schedule or on object data that run actions, AIP Logic functions and notifications.
• Source type overview (Data Connection): connector catalog, with SAP, NetSuite, Odoo, Salesforce, Shopify, WooCommerce, Adobe Commerce and a generic REST APIs connector. The catalog, reviewed on October 13, 2026, does not include Mercado Libre, Bsale, Defontana, Nubox, Falabella or the SII.

Palantir: careers and Forward Deployed Engineers

• Palantir Careers: the three roles, Echos, Deltas and Devs, and what each one does.
• Forward Deployed Software Engineer, New York (Palantir posting on Lever): Palantir pioneered the position, embedding engineers directly with its customers; responsibilities and travel to client sites as needed.
• Palantir Blog, Dev versus Delta: Demystifying engineering roles at Palantir: the product engineer (Dev) builds one capability for many customers; the forward deployed engineer (Delta) takes one customer and combines many capabilities.