Puente OS vs Nexor AI: Key Differences and When to Use Each

An AI agent for the customer relationship, what Nexor AI calls an AI workforce, is a system that talks with your leads and customers in natural language, by phone or by message, and moves each case forward on your rules: it qualifies, books, reaches out again when nobody replies and logs the outcome in your CRM. 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. At Puente, the agent that talks uses the same data the other three pieces keep up to date.

I'm Cris Ugarte, CEO and co-founder of Puente OS. We work with retail, eCommerce and logistics companies in Chile, Peru and Mexico. Nexor AI is a startup born in Chile that builds AI agents for sales teams, and in this post I compare what each one does and how they can work together.

My short answer is that Nexor AI makes sure every lead and every overdue invoice gets its contact, at any hour and on the channel where the person answers. Puente makes sure that contact is backed by the operation: the invoice being collected is the one still unpaid after checking your bank, the customer being re-engaged is the one who stopped buying according to your sales, and what the agent promises gets delivered in the shipment. Here I go through each one, with public sources for everything I say about Nexor AI, and close with when each one makes sense.

What is Nexor AI and what does it solve?

Nexor AI is an AI agent platform for your customer relationships. Its website is titled "Deploy and run your AI workforce," opens with "Run your customer relationship with AI" and describes itself as "the infrastructure for AI agents that qualify leads and book meetings over Call, WhatsApp, Email and Instagram. One API, one MCP server." Its llms.txt adds that it "builds AI agents that help revenue teams contact leads, follow up, send notifications, sell, support customers, collect documents, and audit conversations across channels."

The use cases on its home page:

Initial qualification. "Every new lead filtered with your questions."
Lost customers. Re-engaged with the recontact rule you set.
Collections. "Overdue invoices collected until they're paid."
Documents. Requested, validated and with the file complete.
Appointments. Confirmed, reminded and rescheduled.
After-sales. Follow-up, reviews and the next visit booked.

Its catalog of flows in production adds eight agents you can clone and adjust: B2B expansion, contactability, recontact, notifications, sales, customer experience, document collector and an auditor that, according to the page, "reviews what the other agents did, spots the promise nobody kept and reopens the thread." The sales agent answers sales questions, handles objections and sends the payment link through Stripe or Mercado Pago once the lead agrees.

How it works, according to its website and documentation:

Channels. Call, WhatsApp, email and Instagram in a single system, and every message lands on the same lead, so the agent opens the conversation with the history. Its llms.txt adds SMS and web chat, and its integrations documentation also lists Messenger and iMessage among its channels. For voice, it provisions numbers in the United States and Chile directly, and in other countries depending on availability (voice calls guide).
Inbound and outbound on the same flows. It answers every incoming lead, on their channel and at any hour, and a scheduled task moves leads from your database into the flow that contacts them. Recontact rules, meetings on your calendar with their reminders and the tools the agent calls mid-conversation are configured from the product or through the API.
Versioned instructions. Changes to a live agent go to a draft first, which you test in a Playground and publish when it's ready, with version history and the option to restore an earlier one (documentation).
For developers. An API, outgoing webhooks like meeting.scheduled and workflow_run.status_changed, a JavaScript SDK, scheduled functions and cloud functions that run your own code in an isolated VM, and an MCP server where, according to its home page, "every API action is also an MCP tool." Its documentation index counts 384 MCP tools, with documentation in Spanish and English.
Named integrations. Its home page shows HubSpot, Salesforce, Pipedrive, Zoho CRM, Kommo, HighLevel, Shopify, Stripe, Slack and Airtable. Its integrations documentation details dedicated cards for HubSpot, Kommo, GoHighLevel, Zoho CRM, Odoo and Google Sheets, Shopify carts as a lead source, and Relbase, Diio and Stripe among others; systems with no dedicated card, like Salesforce, connect through a custom connection using webhooks, the API and HTTP tools. Its motto for that: "If it has an API, we integrate." It also has a listing on the HubSpot Marketplace.
For online stores. Its eCommerce page integrates Shopify natively and WooCommerce, Tiendanube and VTEX through a REST API and webhooks. The agent checks stock live during the conversation, creates the cart in Shopify, sends the payment link over WhatsApp or Instagram and also answers the store's phone.
Implementation. According to its home page, its team sets up the first flow, the channels and the rules with you, and "from there on your team clones, changes and launches on its own"; for teams with no technical staff, it offers to set up tailored integrations. Its security page states SSO with SAML and OIDC, that your data doesn't train general models by default, and SOC 2 Type I in progress with Vanta.
Presence. Infobae described it in July 2026 as a Chilean startup that had closed partnerships in Miami, Singapore and Dubai. Its website is signed by Nexor AI, Inc., in San Francisco, and its llms.txt says it serves businesses in Latin America.

It publishes named case studies. In the HealthAtom case, the company behind Dentalink, the agents work every lead that enters HubSpot and, according to the case, one in five books a demo. In the Pepperstone case, an Australian broker, the agent reached 1,520 inactive customers by call, WhatsApp and email to ask why they had left, and 10% started trading again.

Making sure every lead and every customer gets an answer and a follow-up on the channel where they respond: that is Nexor AI's job.

What is Puente OS and what does it solve?

Puente OS is the platform that builds the systems that run your business. In your stack it plays the role of the system of action: it reads from your ERP, your invoicing system, your marketplaces and your bank, cross-references what none of them holds in full, decides on your rules and executes the action inside those same systems.

It builds with four pieces, which look like this in the customer relationship:

Databases that bring together, in a single record, the customer, their sales, their invoices, their payments and their shipments.
Workflows with explicit, auditable rules that start on their own when a payment comes in, an invoice falls due or a shipment changes status.
Agents for the steps that take judgment: the conversation with a customer who complains, a payment that doesn't match or an exception nobody anticipated.
Apps where your team reviews, decides on and approves what the system brings them.

With Puente OS you describe in natural language what your operation needs and get software running: apps, workflows, databases and agents. The Growth Ops team maps your processes, builds each one with you, puts it into production in 2 to 4 weeks and maintains it. Puente covers six areas of a commercial operation (catalog, pricing, orders, logistics, customer service and data), and the relationship with your customer runs through all of them: you sell to them at a price, invoice them for an order, ship to them and collect from them.

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 does an agent need to know to follow up or collect?

It needs to know what happened in your operation. A sales agent talks well when it has the full context, and that context is scattered: the CRM knows the lead, the ERP knows the order and its cost, the invoicing system knows the invoice, the bank knows the payment and the courier knows the delivery.

Take any collections run. A distributor closes the month with 40 overdue invoices and wants an agent to call and message each customer. Before the first call, five questions need answers:

Which invoice is it? The invoice number, the due date and the balance, as recorded in your invoicing system.
Have they already paid? The payment, if it arrived, is in your bank statement, and a single transfer often covers several invoices or only part of one. Applying each deposit to the invoice it pays is what separates the customer who owes from the one who already paid.
Is there a credit note? A return or an agreed discount lowers the balance, and it lives in a separate document.
Did the order arrive in good shape? A customer with an incomplete delivery complains before paying, and the conversation changes when the agent knows what happened with the shipment.
Was the promise kept? "I'll pay on Friday" counts when someone checks the bank on Friday to see whether the payment arrived, and writes again if it didn't.

Sales follow-up works the same way: the customer worth re-engaging is the one who stopped buying according to your sales, and the appointment worth confirming is the one with a schedule and stock behind it. When that data is cross-referenced by hand, in a spreadsheet, the cost piles up on your team: it's the coordination tax. Puente builds and runs that cross-referencing, and the agent works on top of it.

How do Puente OS and Nexor AI compare, feature by feature?

Nexor AI covers the customer relationship with agents across all its channels, and Puente OS covers the complete system the relationship is one step of. Dimension by dimension:

Agents and channels. Nexor AI: agents over call, WhatsApp, email and Instagram, plus SMS and web chat, inbound and outbound on the same flows, voice numbers in the United States and Chile, and an auditor agent that reviews the others. Puente OS: agents on WhatsApp, on the official WhatsApp Business API and with Meta-approved templates, on Instagram, on email and on a web portal chat, working with text, images, files and voice, with handoff to a person on your team when the case calls for it.
Data the agent uses. Nexor AI: the lead with its history across every channel, what it brings in from your CRM and the tools you define, which can call any API mid-conversation; in eCommerce it checks Shopify stock live. Puente OS: the agent reads from a Puente database that has already matched the sale, the invoice, the payment and the shipment, and checks systems like your OMS through an integration.
CRM. Nexor AI: brings in leads from HubSpot, Kommo, Zoho CRM and GoHighLevel, and writes outcomes back to HubSpot, Zoho CRM and GoHighLevel. Puente OS: builds the CRM on real sales from all your channels and keeps it current with every order, every document and every return.
Collections. Nexor AI: a flow that follows up on every overdue invoice until it's paid, with recontact rules and every attempt logged. Puente OS: matches each collection with its document and your bank, applies each deposit to the invoice it pays, deducts credit notes, flags every discrepancy and leaves only the balance still open in collections.
Payments. Nexor AI: a Stripe or Mercado Pago payment link inside the conversation, in its sales flow. Puente OS: reconciles each payment with its order, its document and the deposit that reaches your bank, with the same approach I describe in Puente OS vs Multivende for marketplace settlements.
Invoicing and the SII. Nexor AI: Relbase appears among the cards in its integrations documentation, and its home page shows invoicing and the ERP among the systems it connects through its API and webhooks; on its website and in its documentation, reviewed on October 13, 2026, no connectors appear for the SII (Chile's tax authority), Bsale or Defontana. Puente OS: issues the document in your invoicing system when the process calls for it, and the Growth Ops team connects the SII through its API.
ERP. Nexor AI: a native Odoo connection, currently read-only, and any system with an API through its custom connection and cloud functions. Puente OS: connects to your ERP through its API or by reading its database, brings in the stock, cost and documents each process needs, and writes back.
Stores and marketplaces. Nexor AI: native Shopify, and WooCommerce, Tiendanube and VTEX through the API; Mercado Libre and Falabella don't appear on its website or in its documentation. Puente OS: native connectors for Mercado Libre, Bsale, Multivende and WooCommerce, and the rest through their APIs.
Shipping and after-sales. Nexor AI: after-sales flows with follow-up, reviews and the next visit booked; in eCommerce, the agent answers order-status and returns-policy questions and passes the case to your team when needed. Puente OS: assigns the courier on your rules, generates the label, writes the tracking to the order and measures every delivery against the promise; the agent resolves exchanges and returns with the order it checks in your OMS.
How it's set up. Nexor AI: a self-serve product, configured from its interface or through the API, SDK and MCP, with versioned instructions and tests in a Playground. Puente OS: you describe in natural language what you need and the Growth Ops team builds it with you; every change to the system is logged and can be rolled back to the previous version.
Reporting. Nexor AI: a funnel report and a per-lead report in its API, and conversion events. Puente OS: margin by channel, delivery-promise compliance and the metrics of each process, calculated on your operational data.
Who implements it. Nexor AI: its team sets up the first flow, the channels and the rules with you, then your team clones and launches the next ones, and it offers tailored integrations. Puente OS: the Growth Ops team maps your processes, builds each one with you, puts it live in production and maintains it as your operation changes.

What does it look like in a real operation?

It looks like an agent talking to the right list, because a system built that list from the operation's data. The first two examples are Puente OS customers; the third is a reference design for collections with Nexor AI on the calls.

Example 1: follow-up across an agency network, triggered by its activity

A shipping company in Chile works with a network of agencies that print its shipping labels. Every week, Puente loads the printing activity of 663 points in the network, and the rules flag the ones that stopped printing on their own: in the last week measured, in August 2026, 95.7% of agencies printed, and the 28 inactive ones went straight into a campaign.

The follow-up is done by an agent on WhatsApp, with Meta-approved templates. Between May and August 2026, 3,797 campaign messages went out to 851 points, and 63.7% of the points contacted replied. When an agency answers, the agent replies according to the reason it gives, and of the 316 management notes logged between July and August, the agent wrote 230. The operation decides whom to write to, and the agent talks with the agency that actually stopped printing.

Example 2: the customer relationship, built on real sales

A Chilean retailer with a physical store and an online store had its customers split across the two channels. Puente brought 27,656 orders into a CRM with 9,686 identified customers, and the retailer measured its repeat purchase rate for the first time: 1,702 customers, 17.6%, have two or more purchases. Its loyalty program runs on the same base, with 23.6 million points attributed to 7,685 people.

That's 13 apps in production, with 108 versions deployed, built and maintained by one person on the team. Any repeat-purchase follow-up, whether a Puente agent or another platform's agent runs it, starts from that list: who bought, on which channel and how many times.

Example 3: designing collections that match each payment with its document and your bank

Take a distributor that sells on credit and uses Nexor AI to call and message its customers. This is how you design its collections with Puente OS:

A database links each issued invoice with its customer, its order, its shipment, its credit notes and its deposits.
A workflow pulls the bank statement every day and applies each deposit to the invoice it pays, including when a transfer covers several invoices or only part of one.
The rules sort the cases: whatever is still unpaid after checking the bank goes into collections, and an invoice with an open shipping complaint goes to the account manager.
For each invoice in collections, Puente creates the lead in Nexor AI through its API, with the invoice number, the due date and the balance as metadata, and Nexor's agent calls or messages with that data.
Nexor AI sends back every status change through a webhook, and Puente logs the payment promise. On the promised day, the workflow checks the bank statement: if the payment arrived, it closes the invoice; if it didn't, it sends it back to collections or to a person, with the history in view.

The step-by-step matching of the bank statement against invoices is in the guide to building your invoicing and collections system (in Spanish). Nexor AI calls and talks, and Puente knows which invoice to collect, how much is still open and whether the promise was kept.

Why choose Puente OS?

Because Puente OS builds and runs the complete cycle of your customer relationship: the sale, the invoice, the payment and the shipment, with the agent that talks working on that same data, on a platform designed from your processes and with a Growth Ops team that builds and maintains it. Six reasons:

The agent works on real sales. Every conversation starts from a record that brings the customer together with their orders, their documents, their payments and their shipments. At the Chilean retailer from Example 2, Puente brought 27,656 orders from the physical store and the online store into a CRM with 9,686 identified customers.
Collections match each payment with its document and your bank. Each deposit in the bank statement is applied to the invoice it pays, credit notes lower the balance, and only what's still open reaches collections. It's the same matching Puente does with each marketplace's settlements, as I explain in Puente OS vs Multivende.
Follow-up starts from what happened in your operation. The rules detect who stopped buying, printing or paying, and the agent writes to that list. At the shipping company from Example 1, the week's 28 inactive agencies entered the campaign on their own, and 63.7% of the points contacted replied.
Everything runs on a single platform. Catalog, pricing, orders, logistics, customer service and data share connections and data, and the agent is one of four pieces. At a pet retailer in Chile, Puente already runs processes for six departments with 10 apps in production.
A team builds it with you and maintains it. Growth Ops maps your processes, connects your ERP, your invoicing system, your bank and your marketplaces, decides with you where each exception goes and maintains each process when a rule changes. At the pet retailer, the time between requesting a new process and having it in production dropped from 105 days to 7.
Results are measured in production. At a multichannel electronics seller in Mexico, 97.9% of the 38,628 orders processed got their shipping label with no human intervention, and monthly order volume grew 40% with the same operations team.

The agent closes the conversation; Puente closes the loop, from the first contact to the payment reconciled in your bank.

When does each one make sense?

Nexor AI fits when your challenge is contacting and following up with a high volume of leads or customers, by phone and by message, at any hour. Puente OS fits when the work lies in what that relationship needs to know and in what it sets in motion. And you can use the two together.

Nexor AI, if you want a self-serve product your team configures on its own, centered on the customer relationship: inbound and outbound agents over phone, WhatsApp, email and Instagram, around the clock, with ready-to-clone flows like recontact, appointments or document collection. Also if your sales team works in HubSpot, Kommo, Zoho CRM or GoHighLevel, or if you have a technical team that wants to configure agents through the API, SDK or MCP, with versioned instructions and its own functions. Its self-serve setup, its ready-to-clone flows and its focus on the customer relationship are real strengths.
Puente OS, if every follow-up or collection forces your team to cross-reference your ERP, your invoicing system, your bank or your marketplaces by hand; if you want the agent to talk with reconciled data; and if you want Puente's Growth Ops team to build and maintain both the customer relationship and the operation behind it, on the same platform the next process gets added to.
Both, when your sales team already works in Nexor AI and you want Puente's complete cycle behind it. Puente decides whom to contact and with what data, creates the lead in Nexor AI through its API, receives each outcome through a webhook and follows the cycle through to the payment in your bank.

How do you start?

With the follow-up that depends most on data your team cross-references by hand today: collecting overdue invoices, re-engaging customers who stopped buying or notifying late deliveries. With Puente OS, each process reaches production in 2 to 4 weeks.

Pick the process. With volume, manual handoffs and a visible cost when it fails: a portfolio with many invoices that someone matches against the bank statement every week is usually a good place to start.
Take the baseline. Average days past due, overdue invoices at each month's close, hours per week your team spends matching payments and the share of payment promises that are kept.
Decide who talks. If you already use Nexor AI or another agent platform, Puente hands it the list and receives the outcomes through the API. If you want the agent inside the same system, Puente builds it on your channels, with text, images, files and voice.
Connect the sources. Your ERP, your invoicing system, your bank statements, your CRM, your marketplaces and your courier.
Write down the rules your team applies from memory. For example: "if the customer has an open shipping complaint, the invoice goes to the account manager and collections wait until the complaint is resolved."
Decide where each exception goes. The system resolves it, or it reaches a named person: a payment that matches no invoice reaches finance with the bank statement and the customer's open invoices in view.
Measure from the first week and add the next process: customer re-engagement, delivery confirmation or margin by customer.

That is autonomous operations design applied to your customer relationships: the repeatable work runs on its own, on your rules, and your team decides what takes judgment. To split each step between rules and agents, see the guide on how to implement AI in your business, and if your customers message you to buy, the same approach is in Puente OS vs Vambe.

Your ERP records what happened. Your BI shows you what happened. Puente makes it happen. We believe every company in Latin America can be operationally autonomous, and in your customer relationships that means every follow-up and every collection starts from what actually happened in your operation.

Key takeaways

Nexor AI is an AI agent platform for the customer relationship: it contacts, qualifies, books, follows up, collects overdue invoices, confirms appointments and handles after-sales over phone, WhatsApp, email and Instagram, connected to your CRM, with an API, webhooks, an SDK and an MCP server.
To follow up or collect, an agent needs data that's scattered: the invoice in your invoicing system, the payment in your bank, the credit note in a separate document and the delivery with your courier.
Puente OS builds and runs the complete system the customer relationship is one step of: the sale, the invoice, the payment and the shipment, connected to your ERP, your invoicing and your marketplaces, with the agent working on that data.
In collections, Puente matches each payment with its document and your bank, and leaves only the balance still open in collections.
If you already use Nexor AI, Puente connects to it through its API: it hands it whom to contact and with what data, and receives each outcome through a webhook.
You start one process at a time, in 2 to 4 weeks each, and Puente's Growth Ops team builds and maintains it.

Want to see what collections look like when they start from reconciled data? 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

Does Puente OS replace Nexor AI?

It depends on where your bottleneck is. If it's in having your sales team set up, on its own, agents that contact and follow up with many leads at any hour, Nexor AI has a self-serve product built for that, and Puente connects to it through its API to hand it the list and the data for each case. If it's in knowing whom to contact, what to collect and whether the promise was kept, Puente OS builds that complete system, along with the agent that talks with your customers on the same data, through text, images, files and voice.

What is Nexor AI and what does it do?

Nexor AI is an AI agent platform born in Chile that presents itself as an AI workforce for your customer relationships. Its agents qualify leads, re-engage lost customers, collect overdue invoices, gather documents, confirm appointments and handle after-sales over call, WhatsApp, email and Instagram, with inbound and outbound on the same flows. It integrates with CRMs like HubSpot, Kommo, Zoho CRM and GoHighLevel, with Shopify and with Stripe, and offers an API, webhooks, a JavaScript SDK and an MCP server. It publishes case studies such as HealthAtom and Pepperstone.

Does Nexor AI integrate with ERPs, the SII or Mercado Libre?

With some systems, yes. Its integrations documentation includes a native Odoo connection, currently read-only, and lists Relbase among its cards, and it connects any system with an API through its custom connection, its cloud functions and the tools you define for the agent. On its website and in its documentation, reviewed on October 13, 2026, no connectors appear for the SII, Bsale, Defontana, Mercado Libre or Falabella. Puente OS has native connectors for Mercado Libre, Bsale and Multivende, and the Growth Ops team connects the SII, your invoicing system and your ERP through their APIs.

Does Puente OS have voice agents?

Yes. Puente's agents work through text, images, files and voice, on WhatsApp, on the official WhatsApp Business API, on Instagram, on email and on a web portal chat. At a shipping company in Chile, the agent that follows up with the agency network wrote 230 of the 316 management notes logged between July and August 2026. If your sales team already works with Nexor AI, Puente connects to it through its API so every contact starts from reconciled data.

How do you automate the collection of overdue invoices?

You start with the matching and follow with the conversation. A workflow pulls your bank statement every day and applies each deposit to the invoice it pays, with rules for payments that cover several invoices or only part of one. Whatever is still unpaid goes into collections with its invoice number, due date and balance, and whatever has a complaint or a pending credit note reaches a person on your team. Only then does the agent talk, whether it's Puente's or one from a platform like Nexor AI, and each payment promise is checked against the bank on the promised day. The step-by-step matching is in the guide to building your invoicing and collections system (in Spanish).

External sources cited

Most of these sources are in Spanish, and quotes from them are our translation. Everything this post says about Nexor AI comes from its website, its public documentation, its HubSpot Marketplace listing and press coverage, reviewed on October 13, 2026. Puente OS customer figures are production measurements as of August 17, 2026; the 105-to-7-day figure cross-references the date each process was requested with the date it went live in production.

Nexor AI: website

Nexor AI, home page: title "Despliega y opera tu fuerza laboral de IA" (deploy and run your AI workforce); headline "Opera tu relación con clientes con IA" (run your customer relationship with AI); description of the infrastructure for AI agents that qualify leads and book meetings over Call, WhatsApp, Email and Instagram, with one API and one MCP server; use cases for initial qualification, lost customers, collections ("Facturas vencidas cobradas hasta que se pagan," overdue invoices collected until they're paid), documents, appointments and after-sales; inbound and outbound on the same flows; versioned instructions; scheduled functions and cloud functions in an isolated VM; "Si tiene API, nos integramos" (if it has an API, we integrate); webhooks like meeting.scheduled and workflow_run.status_changed; the @getnexorai/sdk SDK; every API action is also an MCP tool; integrations with HubSpot, Salesforce, Pipedrive, Zoho CRM, Kommo, HighLevel, Shopify, Stripe, Slack and Airtable; onboarding support for the first flow; tailored integrations for teams with no technical staff; Nexor AI, Inc., San Francisco.
llms.txt (Nexor AI), in English: AI agents that help revenue teams contact leads, follow up, send notifications, sell, support customers, collect documents and audit conversations; serves businesses in Latin America with agents for WhatsApp, phone calls, email, SMS, Instagram and web chat.
Flows in production (Nexor AI): eight flows you can clone (B2B expansion, contactability, recontact, notifications, sales, customer experience, document collector and auditor), built on the same API and the same MCP.
Sales agent (Nexor AI): answers sales questions, handles objections and sends the payment link through Stripe or Mercado Pago once the lead agrees.
Agents for online stores (Nexor AI): native Shopify; WooCommerce, Tiendanube and VTEX through a REST API and webhooks; stock checked live, cart created in Shopify and payment link over WhatsApp or Instagram; the agent also answers the store's phone.
HealthAtom case study (Nexor AI): the company behind Dentalink; HubSpot leads worked across every channel; one in five leads books a demo.
Pepperstone case study (Nexor AI): an FX and CFD broker in Australia; 1,520 inactive customers reached by call, WhatsApp and email; 10% started trading again.
Security (Nexor AI): SSO with SAML and OIDC; data doesn't train general models by default; SOC 2 Type I in progress with Vanta.

Nexor AI: documentation

Documentation index (Nexor AI): API reference (including the funnel report, the per-lead report and conversion events), webhooks, scheduled and cloud functions, 384 MCP tools and documentation in Spanish and English.
Bring in leads and connect your systems (Nexor AI documentation): cards for HubSpot, Kommo, GoHighLevel, Zoho CRM, Odoo and Google Sheets; Shopify carts as a lead source; WhatsApp, Instagram, Messenger, email, phone number and iMessage channels; Relbase, Diio and Stripe among other cards; a custom connection through webhooks, the API and HTTP tools for systems with no dedicated card, like Salesforce.
Connect Odoo (Nexor AI documentation): a native read-only connection with an API key.
Set up voice calls (Nexor AI documentation): numbers in the United States and Chile provisioned directly; other countries depending on availability.
Drafts and versions (Nexor AI documentation): changes to a live agent go to a draft, are tested in the Playground and published, with version history and restore.
JavaScript SDK (Nexor AI documentation) and MCP server (Nexor AI documentation): tools to configure and run Nexor AI from code and from MCP clients.
Create or update leads (Nexor AI documentation): creates a lead through the API and assigns it to a task; extra keys are stored as the lead's metadata.
Event: status changed (Nexor AI documentation): the workflow_run.status_changed webhook when a lead's status changes inside a flow.

Nexor AI: other sources

Nexor AI listing (HubSpot Marketplace): an AI agent platform for sales workflows; syncs leads, contacts, deals, calls, notes, meetings and companies with HubSpot.
Startup chilena de IA Nexor cierra alianzas en Dubai, Miami y Singapur (Infobae, Jul 22, 2026): a Chilean startup building agentic AI technology for sales management, operating over phone, WhatsApp, email and Instagram; partnerships in Miami, Singapore and Dubai.

Puente OS vs Nexor AI: From the Customer Relationship to the Complete Cycle