# Subcore AI (Full Context) > Full markdown export for all public pages. Intended for single-request ingestion by agents. ## Index - [Home](https://subcore.ai/index.md) - [About](https://subcore.ai/about.md) - [Chat](https://subcore.ai/chat.md) - [Console](https://subcore.ai/console.md) - [Contact](https://subcore.ai/contact.md) - [Core0 Framework](https://subcore.ai/core0.md) - [Agents](https://subcore.ai/core0/agents.md) - [Flow Agents](https://subcore.ai/core0/flows.md) - [Developers](https://subcore.ai/developers.md) - [Pricing](https://subcore.ai/pricing.md) - [Privacy Policy](https://subcore.ai/privacy.md) - [Scope](https://subcore.ai/scope.md) - [Terms and Conditions](https://subcore.ai/terms.md) - [Voice](https://subcore.ai/voice.md) ## Home # Subcore AI Canonical: [https://subcore.ai/](https://subcore.ai/) Agentic conversational platform for customers, teams, and operations ## Hero - Agentic conversational platform for customers, teams, and operations. - AI systems that handle real customer conversations across chat, voice, SMS, and email. - Get Started - See Demo - Demo: Appt Scheduling - Incoming Call - Patient calls Atlas Therapy to schedule a new evaluation. - Tool Call - patient_lookup({+1404-123-4567}) - Scope - Policy check passed. All actions within bounds - Intent Detected - Booking appointment request - Tool Call → Scheduling System - check_availability({ date: "next Tuesday" }) - Agent Response - I have openings ... on Monday, May 4th. - Collecting Information - Agent is collecting user preferences. - Adapting to User Preferences - Agent is offering a flexible schedule based on user's preferences. - Visit Policies - Agent reviews required visit details and cancellation policy. - book_appointment({ slot: "10:30 AM", date: "2026-04-22" }) - Confirmation - send_confirmation_sms({ to: "+1 (555) 012-3456" }) - Booking Confirmed - Agent Confirmed the appointment details with the patient. ## Technologies - Google Gemini - ElevenLabs - Twilio - Anthropic ## Capabilities - Subcore creates production-ready AI agents tailored to your business needs. - No generic chatbots, no cookie cutter solutions, just intelligent assistants that understand your customers and scale with your business. - Fully managed service - Live in weeks, not months - Complete transparency and control ## Why Companies Choose Subcore - Subcore builds your complete AI solution with custom agents and advanced multi-agent orchestration that automatically routes conversations to the right AI agent for any task. - See Demo ## Solutions - Chat - Custom chat experiences that go beyond simple Q&A. Intelligent conversation flows with rich interactive elements designed specifically for your business needs. - Voice - Voice agents that understand context and intent. Natural, human-like conversations across your phone systems and voice interfaces. - Scope - Keep every interaction safe, compliant, and on-policy. Scope enforces guardrails, detects fraud, flags out-of-scope requests in real time, and grounds models to approved data and tools. - Console - Your central dashboard for monitoring all AI agents and gaining insights from every customer interaction. Complete tools and data to understand your AI's impact on your business. ## The Subcore Difference - Intelligent AI solutions that adapt to your real business needs, not generic templates. - Custom-Built Agents - Specialized agents tailored to your specific use cases and workflows - Omnichannel Memory - Conversations flow seamlessly. Start on chat, continue on voice, finish via email, and your AI remembers everything - Tailored Experiences - Custom widgets and interfaces designed specifically for your business (seat maps, order tracking, appointment booking) - Full-Service Delivery - From design to deployment to ongoing optimization, everything handled for you ## Platform Overview - Custom AI Solutions. Every Channel. Built for You. - Custom conversational AI solutions built on advanced multi-agent technology. Seamless experiences across all your customer touchpoints with enterprise-grade security and observability. - Get Started - See Demo ## Custom solutions - How We Build Your Solution - Get Started - 1. Discovery & Design - Analyze your business processes, customer journeys, and specific needs to design AI agents that solve real problems. No generic templates. Every solution is built from the ground up for your use case. - 2. Custom Agent Development - Our team creates specialized agents using intelligent orchestration that automatically routes conversations to the right specialist, whether that's for technical support, sales, or complex multi-step workflows. - 3. Tailored Experiences - Custom interfaces that appear exactly when needed: flight seat maps during booking, order status with tracking, approval workflows, all seamlessly integrated into the conversation. - 4. Omnichannel Integration - Every interaction feeds into a unified system. Customers can start on chat, continue on voice, finish via email, and your AI remembers everything and maintains context across all channels. - 5. Complete Visibility - Real-time dashboards show exactly how your AI is performing with conversation analytics, success metrics, and detailed insights. See what's working and what can be optimized. ## Deployment - Enterprise-Ready from Day One - Performance - Sub-second response times with automatic scaling built into every deployment - Security - Enterprise-grade security, compliance, and data protection standards - Analytics - Deep insights into conversation quality, resolution rates, and customer satisfaction - Ongoing Support - Continuous monitoring, optimization, and updates ## The Subcore Advantage - Built for Real-World Business Needs - More than just a platform. Complete AI solutions tailored to your specific business challenges. While others offer generic templates, Subcore creates custom experiences that actually solve your problems. - See Demo ## What Makes Subcore Different - Fully Custom Solutions - Every AI agent designed specifically for your business processes, customer needs, and brand voice. No generic templates or one-size-fits-all approaches. - Complete Transparency - You get full visibility into how your AI performs, what customers are asking, and where improvements can be made. No black boxes, just clear insights and actionable data. - Pricing That Fits the Engagement - You pay for what your agents handle, not for seats. Platform fee, usage and build cost are each scoped to the engagement, and multi year terms waive the build cost. - Full-Service Delivery - From initial consultation to ongoing optimization, our team handles everything. You get the benefits of AI without the complexity of building and maintaining it yourself. ## About # About Canonical: [https://subcore.ai/about](https://subcore.ai/about) Subcore AI was built from firsthand enterprise experience so your customers get fast, consistent answers on every channel, with full visibility and control for your team. Live in days, not months. ## Hero - About - Built for the conversations your customers actually have - Subcore came out of years delivering conversational AI inside large enterprises. It exists so your customers get fast, consistent answers on every channel, and your team keeps full visibility and control. ## 1. Rich Text Section - Subcore grew out of firsthand experience building conversational AI for enterprises at the top of the Fortune list. That work made two things clear. Customers respond to conversations that feel immediate and consistent, and teams move fastest when they can see exactly how their AI is performing and shape it with confidence. - Subcore is a lean platform built around both. Responses stream in real time, so a conversation feels like one. A single assistant carries one voice across chat, voice, SMS and email, so a customer gets the same experience however they reach you. Guardrails are part of the foundation, and every conversation is visible to your team, with feedback that shows how the system handled real customer moments and where to improve. - The result is a customer experience you can stand behind, live in days rather than months, and one your team keeps improving with clear evidence rather than guesswork. ## Who is building this - Portrait of the founder of Subcore AI - 50% 22% - Claudio, Founder - Claudio founded Subcore AI in November 2025. The eighteen months before that were applied AI work at Monogram, for enterprises at the top of the Fortune list and for early stage startups. Enough different codebases, constraints and procurement realities to see what these businesses actually need from a conversational system, and what a generic one cannot give them. Subcore is that pattern turned into a product. ## What it means for your customers - Fast, consistent, and fully in your control. - Everything Subcore does maps to a better customer experience and a more confident team behind it. - Responses in real time - Answers stream as they form, so conversations feel immediate and customers stay engaged. - One voice on every channel - Chat, voice, SMS and email share one assistant with one personality, so the experience is consistent however a customer reaches you. - Complete visibility - Every conversation is searchable and reviewable, so your team always knows what was said and why. - Guardrails by design - Your brand, policy and compliance requirements are enforced in the foundation, not patched on afterwards. - Continuous improvement - Feedback on every interaction shows how the system handled real customers and where to refine it. - Live in days - Integrations that are ready to go mean a working deployment in days, so value arrives while the priority is still fresh. ## Chat # Chat Canonical: [https://subcore.ai/chat](https://subcore.ai/chat) Conversational AI That Actually Understands ## Hero - Solution - Chat - Conversational AI That Actually Understands - Subcore chat conversation - See Demo ## Chat - flicker-on - Custom chat experiences that go beyond simple Q&A. Intelligent conversation flows with rich interactive elements designed specifically for your business needs. ## Components - What We Build - Smart Conversation Management - Agents that seamlessly handle multiple topics and complex customer requests - Custom Interactive Elements - Seat maps, product catalogs, forms, booking interfaces—whatever your business needs - Real-time Responses - Instant, natural conversations that feel human - Seamless Integration - Works with your existing systems and customer touchpoints ## Use cases - Perfect For - Customer support with visual elements (order tracking, product selection) - Sales conversations with interactive demos and configurators - Internal tools with approval workflows and data visualization ## The Subcore Advantage - Built for Real-World Business Needs - More than just a platform—complete AI solutions tailored to your specific business challenges. While others offer generic templates, Subcore creates custom experiences that actually solve your problems. - See Demo ## Console # Console Canonical: [https://subcore.ai/console](https://subcore.ai/console) Your AI Command Center ## Hero - Solution - Console - Your AI Command Center - See Demo - Sign In ## Subcore Console - flicker-on - Oversee every agent and interaction from one place. Subcore Console gives you full visibility into your AI ecosystem—so you can track performance, detect issues, and measure real business impact. ## What You Get - Unified Dashboard - View all AI agents, their activity, and business outcomes in one centralized interface. - Real-Time Performance Metrics - Monitor response times, resolution rates, and satisfaction scores as they happen. - Conversation Intelligence - Analyze intent, sentiment, and conversation quality to uncover trends and opportunities. - Quality and Compliance Monitoring - Automatically evaluate responses for accuracy, tone, and brand or policy adherence. - Team Collaboration - Share insights and coordinate improvements with secure, role-based access. - Custom Reporting - Build reports tailored to your KPIs, from customer experience to ROI and operational efficiency. ## Perfect For - Executives tracking AI performance and ROI - Operations teams optimizing customer experience - Compliance teams ensuring consistency and oversight ## The Subcore Advantage - Built for Real-World Business Needs - More than just a platform—complete AI solutions tailored to your specific business challenges. While others offer generic templates, Subcore creates custom experiences that actually solve your problems. - See Demo ## Contact # Contact Canonical: [https://subcore.ai/contact](https://subcore.ai/contact) We're here for you. Fill out the form below and we'll get back to you as soon as possible ## Contact - Have a question or need assistance? - We're here for you. Fill out the form below and we'll get back to you as soon as possible. - Name - Email - Message - Submit ## Core0 Framework # Core0 Framework Canonical: [https://subcore.ai/core0](https://subcore.ai/core0) Core0. Multi-agent orchestration, tools, memory, and telemetry for production AI. ## Hero - Framework - Core0 Framework - The operating layer for AI agents. Coordination, safety, memory, telemetry, human oversight, wired together in one foundation. - See Demo ## Core0 - Build, trust, and scale AI your team can put in front of customers. Agents that work as one, safety built in, humans in control when it counts. Core0 is the operating layer for AI agents — and the foundation behind Chat, Voice, and whatever your team ships next. ## Agents & Flow Agents - Two kinds of agents, one runtime. ## At a glance - Could you draw the conversation on a whiteboard before talking to a single customer? If yes, the path is known and it's a flow agent. If not, reach for an agent. Most production systems use both. - Flow Agents - Explore Agents - Explore Flow Agents - Who owns the steps - The model decides each turn, from the conversation so far. - You define the path up front; the model only fills it in. - An agent hands the structured part to a flow agent. - Best for - Open Q&A, research, and troubleshooting. - Booking, intake, onboarding, checkout, triage. - A greeter out front that routes into procedures. - Tracking what's been said - Works from the conversation history. - Keeps each answer as a structured fact. - One shared session, carried across the handoff. - Reporting a task as done - The model states what it did. - Held back until the real action runs. - The flow's promise holds after a handoff. - Core0 primitive - A plain Agent with tools. - A FlowAgent over a flow() graph. - A FlowOrchestrator routing between them. ## What You Get - Multi-agent orchestration - Orchestrator and handoff modes with bounded loops, cycle detection, and clear agent routing. - Tools and MCP - First-class tools, conditional exposure, and MCP servers wired into each agent. - Memory providers - Pluggable memory (e.g. Firestore, Upstash, Mongo) with a consistent session interface. - RAG and knowledge - Knowledge bases over vector stores with embeddings. Ground answers in your data. - HITL and escalation - Human-in-the-loop coordinators and escalate flows when automation should stop. - Telemetry pipeline - HTTP collectors that ship structured events to Subcore Telemetry API for observability. - Declarative setup - One Core0 constructor for agents, memory, model, and execution mode. Ready for server or edge. ## Perfect For - Platform and backend teams shipping agentic features - Product teams pairing voice or chat UI with a single orchestration core - Enterprises that need audit-friendly telemetry, memory isolation, and tool governance ## The orchestration layer for production AI. One runtime. Every pattern the real world demands: multi-agent routing, safe handoffs, human escalation, memory, retrieval, and telemetry. All wired together, ready to ship. ### One runtime, zero ceremony Describe your agents, your memory, your execution mode. Core0 does the rest. No registries to prime, no boot sequences to manage, no framework tax on every request. #### Details Core0 is the brain behind your AI product. A single orchestration layer that routes requests, runs agents, persists history, and returns a result. One clean surface your team actually wants to use. #### Orchestrator Parallel The runtime picks the best agent per request based on what each one does. Perfect when user intent is varied and routing should be automatic. #### Handoff Sequential Agents route themselves, delegating mid-conversation to specialists. Perfect for multi-step workflows with clear stages. #### Details Engineered for modern infrastructure from day one. Spin up per request on serverless, run it hot on a long-lived service, deploy it to the edge. Core0 stays fast, predictable, and out of your way. ### Agents built for scale Stateless, immutable, concurrency-safe by design. Ship one agent configuration and serve millions of requests without locks, copies, or surprises. #### Details Every Core0 agent is a pure, composable unit. Define it once, reuse it everywhere. No hidden state, no per-request cloning, no race conditions. The same agent handles a trickle or a flood with identical behavior. Adapt per request without touching the agent itself. Inject dynamic instructions based on user context, bolt on one-off tools for a single call, or pass rich caller state into every decision the agent makes. Flexibility without fragility. Observe and intervene at every step. Log tool usage, gate sensitive actions, transform results, or pipe real-time cost data straight into your billing stack. Cleanly layered on top, never tangled up in your agent definitions. ### Tools the model can't misuse Deterministic, schema-enforced workflow control. Your agents get the right capabilities at the right moment, and literally cannot see the wrong ones. #### Details Three tool lifetimes cover every pattern you'll ever need. #### Always-on Every call Declared once on the agent. Available for every interaction, every session. #### Conditional Context-gated Appears only when business rules or runtime context say so. Hidden from the model otherwise. #### Ephemeral Single call Exists for a single call and disappears. No mutation, no leaks, no cleanup. #### Details Model phased workflows like onboarding, checkout, intake, or triage as sequential states with their own instructions and their own toolset. Each phase locks the agent to exactly what it needs. New phases fail safely by default. This isn't prompt-level hope. It's structural enforcement: the model never even sees a tool that shouldn't be available. "Called the wrong tool at the wrong time" becomes impossible, not improbable. The kind of guarantee compliance and security teams actually believe. ### Safe multi-agent handoff Specialist agents pass control to each other mid-conversation. Cycle detection, bounded depth, and frozen context keep every chain predictable. #### Details Build teams of specialists instead of one overloaded generalist. A triage agent greets the customer and hands off to billing. Billing escalates complex cases to a refund specialist. Every handoff carries the full context, and every agent stays focused on what it does best. Every chain is bounded and auditable. Built-in cycle detection stops runaway loops. Sensible depth and time limits cap the worst case. Context is frozen mid-chain, so every agent sees the same inputs. Reproducible, debuggable, safe to run in production. The conversation history stays perfectly consistent across the entire chain. The next agent always reads a clean transcript of everything that came before. No half-written turns, no missing context. ### Evals that ship with your agents Ship with confidence. Core0 includes a first-class eval framework for scripted regressions and goal-completion tests, powered by a simulated user and an LLM judge, so every change is scored before it ships. #### Details Agents that work in a demo and break in production are the single-biggest reason AI projects fail. The core reality Core0 ships with @subcore-ai/evals, a purpose-built evaluation framework wired directly into the same runtime your agents run on. No separate harness, no brittle mocks, no copy-pasted prompt tests. Two modes, one runner: #### Scripted Single-turn Send an input, assert against the response. Great for deterministic behaviors, tool-call regressions, and prompt refactors. #### Goal completion Multi-turn A simulated-user LLM drives a real conversation while a judge LLM scores the outcome against success criteria, failure conditions, and expected tool calls. #### Details Every run gives you rich, structured judgment: #### Pass / fail + reasoning Per case Overall completion, pass state, and a natural-language explanation from the judge. #### Fields collected Per field Did the agent actually capture destination, date, passengers? Compared expected vs actual. #### Success criteria Per criterion Line-by-line verdict on each criterion you defined, with reasoning per line. #### Failure conditions Per guard Catch hallucinations, premature bookings, skipped confirmations. Each condition is individually evaluated. #### Details Personas, metadata, and maxTurns let you dial in realistic user behavior without recording hours of transcripts. A budget-conscious traveler who changes their mind at confirmation is three lines of config, not a fixture file. Use ephemeral, in-memory providers so every run is isolated and hermetic. Because evals run on Core0, they get everything else on this page for free: real tool calls, real memory semantics, real telemetry. What passes in CI actually reflects how the agent behaves at 3am on Saturday, not a fragile approximation. ### Observability out of the box Every agent decision, tool call, and handoff is automatically instrumented. Ship to Datadog, OpenTelemetry, or your own stack without writing a line of telemetry code. #### Details Stop flying blind in production. Core0 emits a rich stream of structured events automatically: agent runs, tool invocations, routing decisions, handoff chains, memory operations. Each one is correlated across your distributed system and safe for multi-tenant environments. Know exactly where time goes on every request: #### Routing time Per request How long the runtime spent picking the right agent. #### Agent time Per call Execution duration plus time-to-first-token, per call. #### Tool durations Per tool Latency for every tool in the chain, individually tracked. #### Handoff depth Per chain Chain length in handoff mode, for safety and cost budgets. #### Details Ships with collectors for the stacks you already run: console for local dev, Datadog and OpenTelemetry for production, plus a simple integration point for your own. Telemetry is flushed cleanly before serverless functions shut down, so you never lose the last event of a request. ### Streaming that just works Real-time tokens for chat, voice, and any live surface. History is always persisted before you respond, even across multi-agent chains. #### Details Low-latency streaming is a first-class citizen, not an afterthought. Pipe tokens straight into a WebSocket, SSE feed, or voice relay, or iterate them inline for scripts and jobs. Every approach is a one-liner. This is what powers the Subcore products you're already looking at. Voice streams tokens into live phone calls over Twilio. Chat UI pushes them to browsers in real time. Custom surfaces do whatever your product needs, all on the same engine. Even in multi-agent chains, streams stay clean. Every agent's output is fully consumed and persisted before the next one starts. No half-written turns, no race conditions, no corrupted transcripts. Just consistent, production-grade streaming at any scale. ### Human-in-the-loop, built in Escalate to a real person with full context in one step. Hand control back when they're done, without losing a token of conversation. #### Details When the AI reaches the edge of what it should do alone, Core0 hands off cleanly to a human. Agents recognize the moment, trigger an escalation, and route the customer to a real teammate, with the entire conversation history already in front of them. State survives everything: serverless cold starts, deploys, page reloads. The human takes over as long as they need, and when the conversation is ready for automation again, Core0 picks up exactly where it left off with the full human transcript in context. Plug it into the tools your team already lives in. Fire events into Slack, open tickets in your helpdesk, send SMS alerts. Every transition is a simple integration point, not a rewrite. ### Memory that scales with you Start on Upstash Redis. Grow into Firestore. Bring your own provider. Session-isolated conversation history your agents can actually reason about. #### Details Your conversations outlive the request. Memory sits between Core0 and your storage layer. History loads before every agent call, new messages persist right after, so customers pick up exactly where they left off, every single time. #### Upstash Redis Realtime Ultra-low latency with TTL. Built for real-time chat and live voice. #### Firestore Enterprise Multi-tenant storage with strong isolation. Drop-in for enterprise SaaS. #### Bring your own Custom One clean interface. Plug in Postgres, Mongo, DynamoDB, or anything else you already run. #### Details Agents don't just remember what was said. They remember what they did. Every tool call, every result, every decision is preserved. Each session is perfectly isolated, with scoped per-session state for the signals your product actually depends on. ### Grounded answers from your data Wire up retrieval-augmented generation in minutes. Any vector store, any embedding model. Your content, your agents, your truth. #### Details Turn your docs, knowledge base, and proprietary data into a source of truth your agents can search on demand. Index content, retrieve by semantic similarity, filter by namespace, score ruthlessly. Production-ready retrieval without hand-rolling a pipeline. Start small with an in-memory store, scale to Pinecone in production, or drop in your own. Embeddings are just as pluggable. Use any modern embedding model, swap providers without rewriting a line of agent code. Choose how retrieval fits your flow. Let the agent decide when to search and call retrieval as a tool, or inject the freshest context straight into its instructions before the call even starts. Same engine, two patterns, zero vendor lock-in. ### Plug into the MCP ecosystem Any Model Context Protocol server, instantly available. Your agents gain tools, resources, and prompts from the fastest-growing standard in AI. #### Details Connect to the entire MCP ecosystem in one line of configuration. GitHub, Slack, Linear, Postgres, your own custom servers. Drop them in and your agents gain battle-tested capabilities from the growing open standard the industry is rallying around. MCP tools behave like first-class citizens inside Core0. Same validation, same observability, same governance as everything else. The model sees one unified toolbox. It doesn't know or care where a tool came from. ## The Subcore Advantage - Built for Real-World Business Needs - More than just a platform. Complete AI solutions tailored to your specific business challenges. While others offer generic templates, Subcore creates custom experiences that actually solve your problems. - Book a Demo ## Agents # Agents Canonical: [https://subcore.ai/core0/agents](https://subcore.ai/core0/agents) Core0 agents handle open-ended work: hand one a goal and tools and it works out the steps itself. Great for Q&A, research, and troubleshooting, and it composes with orchestration, handoffs, memory, and retrieval. ## Hero - Core0 · Agents - Agents that find their own path. - Hand an agent a goal and a set of tools, and it works out the steps for itself. Open-ended by design, for the work you couldn't script in advance. - Book a Demo - See Flow Agents ## The idea - You hand an agent a goal and a set of tools. It decides what to do next, and talks like a person the whole way. ## Built for the work you can't script. An agent is the colleague you trust to think on their feet. Give it a goal and the tools to reach it, and it works out the rest, one turn at a time. ### Thinks on its feet Give it a goal and tools, and it decides what to do next from the conversation so far. No script required. #### Details You describe the goal, hand over the tools, and the agent works out the steps for itself. This is exactly what you want when the path can’t be known ahead of time. It’s at its best on open-ended work: #### Q&A and research Open-ended Answer anything about your product, dig through sources, summarize, compare. #### Troubleshooting Branching Each next step depends on the last answer. The agent discovers the path as it goes. ### Reaches for the right tool The model calls the tools it needs, when it needs them. Search, look something up, take an action, then keep going. #### Details Tools are first-class in Core0. You declare what an agent can do, and the model decides when to use each one. It can pull live data, call your APIs, or trigger a real action mid-conversation. Exposure is yours to control. A tool can be always on, gated behind a condition, or offered for a single turn and then withdrawn. The agent only ever sees the tools you allow, so it can’t reach for the wrong one. ### Composes with the whole framework The same primitive sits behind multi-agent orchestration, safe handoffs, memory, and retrieval. One agent or a team of them. #### Details An agent is rarely alone. Core0 can route each request to the best-suited agent, let specialists hand off to one another mid-conversation, and ground every answer in your own data through retrieval. Memory, tools, and telemetry come from the same foundation, so a single agent and a coordinated team behave consistently and stay observable in production. ### Brings in a human when it counts When the agent reaches the edge of what it should do alone, it can hand off to a person with the whole conversation in hand. #### Details Autonomy has limits, and an agent should know them. Core0 lets an agent escalate cleanly to a teammate, passing along the full history so the person picks up exactly where the conversation left off. When automation is ready to take over again, control comes back with the human transcript in context. Nothing is lost in the handoff, either direction. ## When the path is fixed, reach for a Flow Agent - An agent is often the front door. It greets, works out what the caller wants, and routes anything procedural to a Flow Agent. Open-ended where you want range, structured where you want guarantees. - See Flow Agents - The agent greets and routes - It handles the open-ended turns, figures out intent, and decides where the conversation should go next. - The flow runs the procedure - Once a task has a fixed path and required data, control hands off to a Flow Agent that sees it through. ## One foundation - Agents and flow agents, built on the same operating layer for production AI. - Memory, tools, telemetry, and human oversight come from one runtime. Build the open-ended half as an agent, the procedural half as a flow agent, and ship them together. - Book a Demo - Explore Core0 ## Flow Agents # Flow Agents Canonical: [https://subcore.ai/core0/flows](https://subcore.ai/core0/flows) Core0 flow agents run procedural conversations on rails: typed slots, gated steps, and a promise that nothing is reported done until the real action runs. Built for booking, intake, onboarding, and checkout. ## Hero - Core0 · Flow Agents - A conversation on rails. - A flow agent works through a path you designed. It talks like a person but follows your steps in order, and only confirms once the action runs. - Book a Demo - See Agents ## The idea - You draw the path. The flow agent walks it, talking naturally the whole way. ## A conversation that always stays on track. A flow agent is that same sharp colleague, now working from a checklist you designed. It talks just as naturally; the steps simply follow the path you set. ### Runs the path you set You designed the steps; the flow agent walks them in order. It still handles the natural conversation, however the customer phrases things. #### Details Picture the checklist you’d hand a new hire for a booking or an intake. A flow agent works from that, so every step that matters happens, in the order you intended. The AI handles the conversation. The rails handle the process. The whole idea ### Remembers what it's been told It tracks what the customer has already shared as structured facts, so it never re-asks for something it already has. #### Details When the job is to collect specific information, a flow agent keeps each answer as a structured fact rather than leaving it in the running transcript. The moment a detail is captured, the step that needed it is satisfied and the conversation moves forward. For the customer, that’s a conversation with no repeated questions and no going in circles. For you, it’s the assurance that every field you need is filled before the flow advances. ### Says done only when it's done When a flow agent tells a customer something is booked, it’s booked. A step that reports success isn’t reachable until the real action has run. #### Details When the flow says “booked,” the booking exists. The words never run ahead of the work. The guarantee For transactional work like booking, payments, or placing an order, a flow ties the words to the action. The step that confirms waits for the underlying tool to finish, and the agent commits the action before it speaks. If the action can’t complete, the flow knows, and the customer hears the truth. ### Typed and predictable Under the hood it’s a typed state machine: declared slots, gated transitions, one step at a time. #### Details Each step declares the facts it needs, and a step that performs a real action waits until those facts are in hand. The model only ever sees the step it’s on, written in plain language, with the tools that step allows. In Core0 this is a FlowAgent wrapping the same Agent the open side uses. Same memory, same tools, same telemetry, driven a different way. ## When to reach for a flow agent - The test: could you draw the conversation on a whiteboard before talking to a single customer? If yes, the path is known and only the data varies. That is a flow. - Booking and scheduling - Clinics, salons, restaurants. Collect the details, check availability, and confirm the slot. - Intake and onboarding - Run a form as a conversation. Capture every required field, in order, without the rigid form feel. - Checkout and payments - Walk through the steps and confirm only once the charge has actually gone through. ## Pair them with Agents - A flow agent owns the procedure. Everything open-ended around it, the greeting, the curveball question, the follow-up, belongs to an agent. Core0 routes between the two. - See Agents - An agent handles the open-ended turns - Greeting, questions, and anything off the script stay with an agent that can think on its feet. - A flow agent runs the procedure - The moment the task has a fixed path and required data, the flow agent takes over and sees it through. ## One foundation - Flow agents and agents, built on the same operating layer for production AI. - Memory, tools, telemetry, and human oversight come from one runtime. Build the procedural half as a flow agent, the open-ended half as an agent, and ship them together. - Book a Demo - Explore Core0 ## Developers # Developers Canonical: [https://subcore.ai/developers](https://subcore.ai/developers) Subcore AI's public REST API, OpenAPI specification, and MCP server. Unauthenticated access to pages, products, search, and a demo-request endpoint. ## Hero - Developers - Subcore AI for developers and agents - A public REST API, an OpenAPI spec, and an MCP server — no key required. ## 1. Rich Text Section - Everything on subcore.ai is available programmatically. No API key, no sign-up, no tiers — the same content the site renders is served as JSON, as markdown, and over the Model Context Protocol. - REST API - Base URL: /api/v1. Unauthenticated. - GET /pages — every public pageGET /pages/{slug} — one page, including its markdown bodyGET /products — products and platform componentsGET /search?q= — ranked full-text searchGET /company — company profile and agent interfacesPOST /contact — send a demo request - curl "https://subcore.ai/api/v1/search?q=guardrails&limit=3" - Errors share one envelope. Lists paginate with limit and offset and return a next_url. Search and the contact endpoint are metered per client and send RateLimit-* headers; the cached reads are not metered and send none. POST /contact accepts an Idempotency-Key so a retry is recognised rather than delivered again. - OpenAPI - The full contract, with every schema and example: openapi.json. - MCP server - Model Context Protocol over Streamable HTTP at /api/mcp, also reachable at /mcp. No authentication. - { "mcpServers": { "subcore-ai": { "type": "streamable-http", "url": "https://subcore.ai/api/mcp" } } } - Six tools: list_pages, get_page, search_content, list_products, get_company_info and request_demo. Every page is also exposed as a markdown resource, and two ui:// views let MCP Apps hosts render products and search results inline. - Connection details: mcp.json · Server card: server-card.json - Markdown and agent indexes - llms.txt — curated index, including when to use this site and when not tollms-full.txt — every page in one responseAny page as markdown by appending .md, e.g. /core0/agents.mdard.json — agentic resource catalogagent-skills/index.json — published skills - Versioning and deprecation - The version is in the path. /api/v1 is current and not deprecated. Breaking changes ship as /api/v2; a retired version keeps serving for at least six months and carries Deprecation, Sunset and a successor-version Link header on every response. Their absence means the version is current. - Support - Questions about the API: support@subcore.ai, or the contact form. ## Pricing # Pricing Canonical: [https://subcore.ai/pricing](https://subcore.ai/pricing) Subcore AI engagements are enterprise and custom, priced per engagement rather than by tier, because the integration layer is built for your systems. What shapes a quote, and how to get one. ## Hero - Pricing - Pricing that fits your engagement - Every deployment is built around the systems you already run, so pricing is set per engagement, not from a list. What it can include is below, and which parts apply we decide together. - Book a demo - Talk to us ## 1. Rich Text Section - Subcore engagements are enterprise and custom. The agent itself is the smaller half of the build: the larger half is the integration layer that connects it to your CRM, your ticketing, your telephony, your data and whatever else the work actually touches. That layer is written for your systems, and it does not come in sizes. - So pricing is scoped per engagement rather than published as tiers. The parts an engagement can include are set out below. Which of them apply, and at what level, depends on the engagement, and early or smaller engagements are often structured more simply. ## What an engagement can include - Three parts. Which apply is scoped to the engagement. - Not every engagement carries all three, and the levels are set per engagement. Knowing the parts lets you budget against them before you ever speak to us. - A platform fee - What it costs to run your agents on Subcore: the framework, the guardrails, the telemetry and the console your team works in. - Usage - Charged against what your agents actually handle, so the bill tracks the work rather than the size of your team. - A build cost, waived on multi year contracts - Standing up the integration layer against your systems is real work and is charged as such. Commit for multiple years and it comes off entirely. ## Getting a number - A scoping conversation is the fastest route to a real figure. - Tell us the channels, the systems it has to reach and roughly what volume you expect, and we can scope it from there. - Book a demo - Talk to us ## Privacy Policy # Privacy Policy Canonical: [https://subcore.ai/privacy](https://subcore.ai/privacy) ## Hero - Privacy Policy ## 1. Rich Text Section - Effective Date: 6/20/2025. Last Updated: 6/15/2026 - 1. Overview - Subcore.ai ("Subcore," "we," "us," or "our") provides AI-powered conversational and automation services across digital and voice channels. This Privacy Policy explains how we collect, use, disclose, and protect personal information obtained through our website, platform, SMS campaigns, and other related services (collectively, the "Services"). - By using our Services or opting in to communications from Subcore, you agree to the terms of this Privacy Policy. - 2. Information We Collect - We may collect the following categories of information depending on how you interact with our Services: - Contact Information: Name, phone number, email address, and company details.Account Data: Login credentials, user preferences, and organization settings.Communication Data: Messages sent via chat, voice, SMS, or email, including timestamps and response data.Usage Data: IP address, device identifiers, browser type, operating system, and activity logs.AI Interaction Data: Prompts, responses, and related metadata generated through AI systems, used for system improvement and analytics.Compliance and Consent Records: Opt-in status, timestamps, and proof of consent for A2P SMS or messaging services. - 3. How We Use Information - We use your information to: - Deliver and maintain our Services, including AI agent interactions.Send informational or transactional messages, including SMS updates, alerts, and support responses.Send marketing messages (only with your explicit consent).Monitor and analyze usage to improve performance, accuracy, and reliability of AI models.Ensure compliance with applicable laws, carrier requirements, and industry standards.Prevent fraud, abuse, and unauthorized access. - 4. SMS Messaging - By providing your mobile number and opting in, you consent to receive SMS messages from Subcore. - Message frequency varies by use case.Standard message and data rates may apply.You may opt out at any time by replying STOP, or request help by replying HELP.We maintain logs of opt-in and opt-out events for regulatory compliance. - Subcore complies with all CTIA, TCPA, and carrier-level guidelines governing A2P messaging. - Mobile opt-in information, phone numbers collected for SMS, and SMS consent are never sold, rented, shared, or transferred to third parties or affiliates for marketing or promotional purposes. This information may be shared only with subprocessors that directly support message delivery, such as our messaging provider. Text messaging opt-in data and consent are excluded from all other data sharing described in this policy, including enterprise client data sharing and business transfers. - 5. Information Sharing - We do not sell or rent your personal data. We may share information with: - Service Providers: Partners who support delivery of our Services (e.g., messaging carriers, hosting providers, analytics vendors).Enterprise Clients: If you use Subcore as part of a business relationship, relevant data may be shared with your organization for analysis or compliance purposes.Regulators or Law Enforcement: When required by law, regulation, or court order.Business Transfers: In the event of a merger, acquisition, or sale of assets, where information may be transferred under confidentiality protections. - SMS opt-in information, phone numbers collected for SMS, and SMS consent are excluded from all sharing described in this section, including with Enterprise Clients and in the event of Business Transfers. See the SMS Messaging section above for details. - 6. Data Retention - We retain personal information only as long as necessary to provide our Services, fulfill legal obligations, resolve disputes, and enforce our agreements. - SMS and conversational logs may be stored for limited durations required by compliance and carrier regulations. - 7. Data Security - We implement administrative, technical, and physical safeguards to protect your information, including encryption, access controls, and secure storage. - Despite these measures, no system is completely secure. You acknowledge that transmission of data over networks carries inherent risk. - 8. AI System Data - Conversations with Subcore's AI systems may be processed to improve model performance, reliability, and safety. - We apply data minimization, pseudonymization, and strict access controls to all training or evaluation data. Sensitive or personal information is not used to train models unless explicitly authorized. - 9. Your Rights - Depending on your jurisdiction, you may have the right to: - Access, correct, or delete your personal data.Request restriction or objection to certain data processing.Withdraw consent for communications at any time.Request a copy of data we hold about you. - Requests can be sent to privacy@subcore.ai. - 10. Children's Privacy - Our Services are not intended for individuals under 13 years of age, and we do not knowingly collect information from them. If such data is discovered, it will be deleted immediately. - 11. Changes to This Policy - We may update this Privacy Policy periodically. Updates will be posted on our website with a revised effective date. Continued use of our Services after an update constitutes acceptance of the revised policy. - 12. Contact - For privacy inquiries or data requests, contact: - Subcore AI - 8000 Avalon Boulevard, Suite 100, Alpharetta GA 30009, United States - hello@subcore.ai ## Scope # Scope Canonical: [https://subcore.ai/scope](https://subcore.ai/scope) Guardrails, Grounding, and Risk Controls ## Hero - Solution - Scope - Guardrails, Grounding, and Risk Controls - See Demo ## Scope - flicker-on - Keep every interaction safe, compliant, and on-policy. Scope enforces guardrails, detects fraud, flags out-of-scope requests in real time, and grounds models to approved data and tools. ## What You Get - Policy Guardrails - Role, intent, and content checks with automatic block, redact, or escalate actions - Grounded Responses - Source-bound answers, tool-only actions, and strict retrieval constraints - Out-of-Scope Detection - Live monitoring that interrupts risky or off-policy flows and steers back to approved paths - Fraud and Abuse Signals - Account takeover patterns, social-engineering cues, duplicate attempts, and velocity checks - Jailbreak and Toxicity Defense - Prompt attack detection, safety filters, and controlled fallbacks - PII and Compliance Controls - Redaction, consent checks, audit trails, and retention policies that fit your standards - Human Escalation - Clear handoffs to agents with full context and a reason for the intervention ## Perfect For - Security, risk, and compliance teams - CX and operations leaders who need safe, reliable automation - Enterprises that require grounding, auditability, and policy enforcement across channels ## The Subcore Advantage - Built for Real-World Business Needs - More than just a platform—complete AI solutions tailored to your specific business challenges. While others offer generic templates, Subcore creates custom experiences that actually solve your problems. - Book a Demo ## Terms and Conditions # Terms and Conditions Canonical: [https://subcore.ai/terms](https://subcore.ai/terms) ## Hero - Terms and Conditions ## 1. Rich Text Section - Effective Date: 6/20/2025. Last Updated: 10/15/2025 - 1. Introduction - Welcome to Subcore.ai ("Subcore," "we," "our," or "us"). These Terms and Conditions ("Terms") govern your access to and use of Subcore's website, platform, and communication services, including SMS, email, chat, and voice-based interactions (collectively, the "Services"). - By using our Services or opting in to receive communications from Subcore, you agree to be bound by these Terms. If you do not agree, do not use the Services. - 2. Messaging and SMS Services - By providing your mobile number and opting in, you consent to receive informational, transactional, or promotional text messages from Subcore. - Message Frequency: Message frequency varies by use case.Message Rates: Standard message and data rates may apply.Opt-Out: You may opt out at any time by replying STOP to any text message. After opting out, you may receive a one-time confirmation message.Help: For assistance, reply HELP or contact support@subcore.ai.Consent: Consent to receive text messages is not a condition of purchase.Carrier Disclaimer: Wireless carriers are not liable for delayed or undelivered messages. - Subcore complies with CTIA, TCPA, and all carrier guidelines for A2P (Application-to-Person) messaging, including consent, content, and opt-out requirements. - 3. Use of the Services - You agree to use the Services only for lawful purposes and in compliance with all applicable laws and regulations. You will not: - Misuse or interfere with the normal operation of the Services.Attempt to reverse engineer, modify, or copy Subcore's software or AI systems.Use Subcore's Services to transmit unlawful, harmful, or unauthorized content.Engage in fraud, phishing, or impersonation activities. - Subcore reserves the right to suspend or terminate access to the Services for violations of these Terms or misuse of messaging systems. - 4. AI-Driven Interactions - Some parts of the Services use artificial intelligence (AI) to automate responses, analyze text, or facilitate workflows. You understand and agree that: - AI-generated responses are based on available data and may not always be accurate.Subcore is not liable for decisions made based on AI-generated content.Conversations may be monitored or logged to ensure compliance, improve performance, or support security audits.All AI usage complies with applicable privacy, data handling, and consent regulations. - 5. Accounts and Security - If you create an account, you are responsible for maintaining the confidentiality of your credentials and for all activity under your account. Notify Subcore immediately of any unauthorized access or security breach. - 6. Intellectual Property - All content, software, and technology within Subcore's Services—including AI models, designs, and workflows—are owned by or licensed to Subcore. You may not reproduce, distribute, or create derivative works without our written consent. - 7. Limitation of Liability - To the fullest extent permitted by law, Subcore and its affiliates will not be liable for any indirect, incidental, or consequential damages arising from: - Your use or inability to use the Services;Delayed, failed, or undelivered messages;Errors or inaccuracies in AI-generated content;Unauthorized access to your data. - Subcore's total liability will not exceed the amount paid for the Services in the six months preceding the claim. - 8. Disclaimer of Warranties - The Services are provided "as is" and "as available." Subcore makes no warranties, express or implied, regarding accuracy, reliability, or fitness for a particular purpose. - 9. Privacy and Data Protection - Your use of the Services is also governed by our Privacy Policy, which describes how we collect, use, and protect your personal information. By using the Services, you acknowledge and agree to our data practices as outlined in the Privacy Policy. - 10. Compliance and Legal Disclosures - Subcore complies with all applicable laws governing data use, messaging consent, and electronic communications, including but not limited to: - Telephone Consumer Protection Act (TCPA)CAN-SPAM ActCTIA Messaging Principles and Best PracticesGeneral Data Protection Regulation (GDPR) (where applicable)California Consumer Privacy Act (CCPA) - 11. Termination - We may suspend or terminate your access to the Services if you violate these Terms or if required to do so by law or carrier policies. - 12. Changes to These Terms - We may modify these Terms from time to time. Updates will be posted on our website with a revised effective date. Continued use of the Services after an update constitutes acceptance of the revised Terms. - 13. Contact Us - For questions about these Terms or our messaging policies, contact: - Subcore AI - 8000 Avalon Boulevard, Suite 100, Alpharetta GA 30009, United States - hello@subcore.ai ## Voice # Voice Canonical: [https://subcore.ai/voice](https://subcore.ai/voice) Voice AI That Sounds Human ## Hero - Solution - Voice - Voice AI That Sounds Human - See Demo ## Voice - flicker-on - Voice agents that handle complex conversations naturally. Advanced voice experiences that process speech, understand intent, and respond with human-like clarity across your phone systems and voice interfaces. ## Components - What We Build - Natural Voice Conversations - Advanced speech processing tailored to your customer's language and accent patterns - Phone System Integration - Seamless integration with your existing telephony infrastructure - Multi-language Support - Voice agents that serve customers in their preferred language - Smart Handoffs - Intelligent transfer to human agents when complex issues require human touch ## Use cases - Perfect For - Customer service phone lines with 24/7 availability - Appointment scheduling and reservation systems - Order status inquiries and account management ## The Subcore Advantage - Built for Real-World Business Needs - More than just a platform. Complete AI solutions tailored to your specific business challenges. While others offer generic templates, Subcore creates custom experiences that actually solve your problems. - See Demo