Glia
Also known as: SaleMove
Banking AI platform for banks, credit unions, and insurers that unifies voice, digital, and video with agentic workflows through a ChannelLess architecture.
Glia is a customer interaction platform built for banks, credit unions, and insurers, unifying voice, digital messaging, video, and AI so a conversation can move between channels without losing context. Founded in 2012 in New York as SaleMove by Dan Michaeli, Justin DiPietro, and Carlos Paniagua, and rebranded Glia in 2019, the company has raised about one hundred fifty two million dollars, reaching a one billion dollar valuation with its forty five million dollar Series D in March 2022 led by Insight Partners. It strengthened its banking AI by acquiring conversational AI company Finn AI in 2022, and now says more than seven hundred financial institutions run on the platform.
The platform comes in three parts. Glia Voice carries phone interactions with voice AI the company markets as the fastest in banking. Glia Digital covers self-service, messaging, video, and the collaboration tools, CoBrowsing among them, that let a representative work alongside a customer on screen.
Glia AI sits above both and is sold as a banking AI workforce: Glia Banker automates routine calls and digital inquiries across more than a thousand banking journeys, Glia Outreach runs outbound voice and SMS campaigns to grow deposits and reduce delinquencies, Glia CoPilot brings approved institutional knowledge and guidance into live interactions, and Glia Analyst handles coaching, quality management, and workforce planning. Tying them together is the ChannelLess architecture, which carries context as a customer shifts from chat to voice to video without reauthenticating or repeating themselves.
The newest addition is Glia Agentic Workflows, which turns an institution's approved procedures into governed AI actions that complete multi-step tasks from inside the interaction, among them unlocking online banking, freezing a card, adding a travel notification, and transferring between accounts. Institutions start from published templates or describe a workflow in natural language, choose which steps run autonomously and which require human approval, and get a full audit trail of every step, decision, and system action.
Everything is engineered for regulated banking, from core and online banking integrations to a knowledge base of institution-approved answers that bounds what the AI is permitted to say. Glia holds SOC 2 Type 2, PCI Level 1, HIPAA/HITECH Type 1, and CCPA compliance, and sells a contractual guarantee against hallucinations and prompt injections.
There is more room for technical teams than the marketing suggests: REST APIs, web and mobile SDKs, event-based data exports that carry structured audit logs into a customer's own systems, and Glia Functions, a serverless platform for deploying custom JavaScript integrations. Pricing is quoted through sales but marketed as predictable, with nothing charged per seat, minute, or token.
For a bank, credit union, or insurer that wants a single AI-powered interaction platform tuned to financial services, Glia is a strong fit. A buyer who needs to choose the underlying model, select a hosting region, or deploy on-premise will find a managed, United States-hosted service rather than a configurable runtime, and a general business outside financial services will find it narrower than a horizontal support tool.
Vendor details
Canonical URL
https://www.glia.com
Category
Customer support agent
Subcategory
Financial services customer interaction
Funding status
Independent, headquartered in New York, founded in 2012 as SaleMove by Dan Michaeli, Justin DiPietro, and Carlos Paniagua, and rebranded Glia in 2019. Has raised about one hundred fifty two million dollars across seven rounds, reaching a one billion dollar valuation with a forty five million dollar Series D on March 21, 2022 led by Insight Partners, with Tola Capital, Wildcat Capital Management, and Entrepreneurs Roundtable Accelerator also backing it. Acquired banking conversational AI company Finn AI in 2022, says more than seven hundred financial institutions run on the platform, and has facilitated over ten billion customer interactions.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Ships pre-built connections into banking cores, digital and online banking, card management, payments, and disputes and account services, and embeds across a bank's digital properties, phone lines, and mobile apps. Agentic Workflows execute account-level actions in those systems rather than only reading from them. Assistants are grounded in a knowledge base of institution-approved content. For anything not covered by a connector, Glia publishes REST APIs at api.glia.com, web, iOS, Android and Ionic SDKs, event-based data exports to CRMs and data lakes, Applets that embed external tools in the agent workspace, and Glia Functions, a serverless JavaScript platform with a CLI and an MCP server maintained in Glia's public salemove GitHub organization.
In practice
Members call, then chat, then get bounced to a human and have to reauthenticate and re explain everything. Glia's ChannelLess architecture carries the full context across voice, digital, and video so the conversation never restarts.
Your contact center is buried in routine balance and card questions. Glia's Virtual Assistant automates up to sixty percent of interactions across phone and digital, grounded only in answers your institution has approved.
You need to grow deposits and reduce delinquencies without adding headcount. Glia AI Outreach runs automated outbound voice and SMS campaigns to renew certificates, convert borrowers, and reengage members.
Sources & related URLs
Research sources
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool Calling | Full |
|---|---|
|
Named system classes plus authenticated write action, evidenced by the actions themselves. Agentic Workflows ships pre-built connections into banking cores, digital and online banking, card management, payments, and disputes and account services, and the published workflow set is transactional rather than informational: unlock online banking, freeze a card, add a travel notification, transfer between accounts, replace a damaged card. Those are writes into a system of record executed from the interaction. Where a connector does not exist, Glia Functions lets the institution deploy its own JavaScript to build the integration, and a partners and integrations portal is published separately. Sourceglia.com/agentic-workflowsread 2026-09-05 |
|
| Workflow Orchestration | Full |
|
Multi-step execution against a codified process. Glia Agentic Workflows, launched in the current product cycle, turns an institution's approved procedures into governed AI actions that complete multi-step banking tasks inside a single screen. The build path is documented end to end: start from a pre-built template or create a workflow by describing it in natural language, or by uploading internal standard operating procedures and historical interactions. At run time CoPilot scans the live interaction and prompts the agent to run the workflow matching the inquiry, then the workflow fills in information, suggests next steps and completes actions across connected systems while the agent stays in the conversation. Glia names the execution model as AI reasoning under governed execution. Sourceglia.com/agentic-workflowsread 2026-09-05 |
|
| Knowledge Grounding & RAG | Full |
|
A maintained corpus under institutional control. Glia builds its AI assistants on the Glia Knowledge Bank, a single source of answers and content the financial institution has reviewed and approved, so the assistant only ever says what the institution has sanctioned, and a proprietary approvals framework enforces that boundary at runtime. The corpus is fed and maintained from the institution's own material: workflows are generated from uploaded standard operating procedures and internal documentation, and CoPilot surfaces that institutional knowledge into live interactions so an agent can answer complex product questions without leaving the conversation. Phone virtual assistant responses are generated from prior agent interactions through a documented learning loop, so the structure is refreshed from live traffic rather than loaded once. Sourceglia.com/ai-copilotread 2026-09-05 |
|
| Human Oversight & Guardrails | Full |
|
A customer-set autonomy ceiling with a per-step approval gate. Building an Agentic Workflow, the institution chooses the steps and then determines where AI completes actions autonomously and where human approval is required, with the two states named on the page as Human Approval and Fully Autonomous. Glia frames this as configurable control: the customer maintains full control of when AI acts independently and when approval is needed, keeping every workflow aligned to their own policies and risk tolerance. Underneath it, a proprietary approvals framework governs what the AI is permitted to say, carried far enough that Glia sells a contractual guarantee against hallucinations and prompt injections, and assistants answer only from institution-approved content. On the interaction side, the AI recognizes when human help is needed and hands off with context intact. Sourceglia.com/agentic-workflowsread 2026-09-05 |
|
| Security, Identity & Governance | Full |
|
Attestations and customer-facing access controls, on named pages. The security page lists SOC 2 Type 2, PCI DSS, HIPAA/HITECH Type 1 and CCPA, with the Banking AI Workforce page stating PCI Level 1 specifically and positioning it as going beyond SOC 2 Type II. Around that sit scheduled third-party penetration testing, a TDD development process with manual and automated security checks on OWASP principles, a 24/7 intrusion detection system with daily log review, encryption in transit and at rest, PII masking, dedicated security staff holding CISM, CISA and CRISC credentials, employee background checks, and a published bug bounty program. For access, the Extensibility page names customer-managed SSO integrations alongside user and session management in the development toolkit, and the privacy policy states that customers determine how long information is retained and can request modification or deletion as data controllers, which is customer-configured retention. The same security page cites ISO 27001, FedRAMP/FISMA and SSAE-16 SOC 1, 2 and 3 as controls Glia inherits from Amazon Web Services; those belong to the cloud provider rather than Glia. The two-factor, logged, least-privilege access described under the Personnel pillar governs Glia's own employees rather than the buyer's administrators. Sourceglia.com/securityread 2026-09-05 |
|
| Observability & Auditability | Full |
|
A per-run audit record, not just analytics. Agentic Workflows publishes full audit trails in which every step, every decision and every system action is captured and fully auditable, framed as the compliance surface for regulated execution, alongside a built-in end-to-end reporting dashboard that surfaces bottlenecks and process-improvement opportunities. The audit record also leaves the platform on the customer's terms: the Extensibility page describes event-based exports that push structured audit logs to CRMs, data lakes and other external systems to meet compliance requirements, with customizable payloads. Glia Analyst adds interaction-level analysis and on-demand reporting across the estate. This records Glia's own execution rather than logs the customer's core banking systems produce. Sourceglia.com/agentic-workflowsread 2026-09-05 |
|
| Memory & State Persistence | Partial |
|
Continuity is documented and a memory layer is not. The ChannelLess architecture carries full context as a customer moves between chat, voice and video so they do not reauthenticate or repeat themselves, and handoff from AI to a human passes context with it. Beyond the single interaction, Glia personalizes greetings using a caller's previous interactions and call history, which reaches across separate conversations rather than only within one. No persistent agent memory store is published with a scope, lifetime or deletion control a buyer could name, and no memory product is offered; what the platform holds is interaction history rather than retained agent state. Sourceglia.com/ai-chatbot-bankingread 2026-09-05 |
|
| Deployment & Data Residency | Partial |
|
Single-cloud hosting with no buyer selection. The security page states Glia is hosted exclusively on Amazon Web Services, and the privacy policy tells users located outside the United States that their information may be transferred to and processed in the United States. Availability is engineered rather than distributed for residency: a triple-hot redundant architecture is published at 99.99999% uptime with zero planned downtime. No customer-selectable region, no in-region or single-tenant option and no on-premise deployment is documented, and Glia publishes no DPA or subprocessor list, stating instead that compliance with GDPR and CCPA is detailed in customer agreements, which places residency terms at contract level rather than in public documentation. Sourceglia.com/security-compliance/privacy-policyread 2026-09-05 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
A catalog the buyer browses and selects from. The Agentic Workflows page publishes its template set by name and invites customization of each: Unlock Online Banking, Add Travel Notification, Freeze Card, Check Balances, Review Transactions, Transfer Between Accounts, Replace Damaged Card, and Build Your Own. Alongside that sits a named product set a buyer adopts as units, each with its own page: Glia Banker, Glia Outreach, Glia CoPilot and Glia Analyst. Underneath both, the voice agent ships pre-packaged for more than 1,000 banking journeys out of the box with no bot building, integration work or training required. Sourceglia.com/agentic-workflowsread 2026-09-05 |
|
| Triggers & Channel Coverage | Full |
|
Breadth on the inbound side and an initiating path on the outbound one. Inbound work reaches the agent by phone, chat, video and messaging across the institution's digital properties, phone lines and mobile apps, with context carried when the customer moves between them. Outbound is a named product rather than an afterthought: Glia Outreach runs automated voice and SMS campaigns that convert indirect borrowers, prompt CD and share-certificate renewals and reduce delinquencies, then transfers the call to a specialist when the member is ready to act, so the agent starts conversations rather than only receiving them. Callers are greeted with a message personalized to their history. A third trigger class is documented on the developer surface: Glia Functions supports scheduled invocations on cron expressions, so work can also start on a clock with no human or customer involved. Sourceglia.com/glia-outreachread 2026-09-05 |
|
| Model Flexibility & Routing | Not documented |
|
Glia keeps the model layer to itself. No underlying model provider is named anywhere on the platform, no model selection or routing configuration is published, and no bring-your-own-key path is offered. The commercial posture points the same way: Glia sells a contractual guarantee against hallucinations and prompt injections enforced by a proprietary approvals framework, which is a promise a vendor can only make while controlling the model layer itself, and the platform is marketed as turnkey AI that ships pre-packaged for banking journeys rather than as a model-agnostic runtime. Amazon Web Services appears throughout as the hosting environment, which is infrastructure rather than a second model. Sourceglia.com/banking-ai-workforceread 2026-09-05 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Callable from outside on several documented routes. Glia publishes an Extensibility page offering REST APIs, software development kits and data export tools, an off-the-shelf development toolkit that embeds Glia in web and mobile apps and manages users, sessions and SSO integrations, and event-based exports that sync engagement data in real time to CRMs and data lakes. Glia Functions is a customer-facing serverless platform: the buyer deploys their own JavaScript to build integrations and customizations inside the Glia platform. Glia's own public GitHub organization (salemove, the former company name, 90 public repositories) carries glia-functions-tools, holding a Functions CLI, first-party web, iOS, Android and Ionic widget SDKs, and an MCP server exposing roughly two dozen tools so an AI assistant can create, deploy and schedule functions programmatically. The platform API is at api.glia.com on bearer-token authentication, with service-credential API keys documented in the SDK release notes. The developer hub at docs.glia.com sits behind a login. SourceGlia extensibility page, github.com/salemove/glia-functions-tools and the salemove widgets SDK repositories; glia.com/extensibilityread 2026-09-05 |
|
| Testing, Debugging & Optimization | Partial |
|
Measurement after deployment is documented; a gate before it is not. Glia Analyst runs quality management and coaching across interactions, checking whether agents met banking compliance requirements and how they handled conversion and empathy, and Agentic Workflows ships a built-in end-to-end reporting dashboard that identifies bottlenecks and future automation opportunities, so a change can be judged against readable outcomes once it is live. No simulation, scenario set, regression suite or scored pre-release evaluation of a proposed AI agent or workflow is documented. Analyst is published for coaching human agents and planning staffing, so what it scores is the interaction and the representative rather than the customer's own AI configuration. The Glia Functions CLI supports local development of serverless functions, which tests integration code rather than agent behavior. Sourceglia.com/ai-analystread 2026-09-05 |
|
| Browser & Computer Use | Not documented |
|
Glia's co-browsing is a tool for human representatives, not agent computer use. Glia ships CoBrowsing and Live Observation, which let a representative see and co-navigate the customer's own browser during an interaction, with Live Observation extended to capture bubbles, alerts and picker menus so the operator sees the whole screen. That is a human operator working inside an interface with the customer present and consenting, not an agent driving a browser. No hosted browser session, virtual desktop or agent-driven interface control appears anywhere on the platform. The agent acts through programmatic routes instead, with pre-built connections into banking cores, card management and payments, workflows that execute account-level actions in those systems, customer-deployed Glia Functions and a REST API at api.glia.com. Sourceglia.com/high-touch-interactionsread 2026-09-05 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Recent platform changes
Glia introduced three configurable Precision AI Response Modes for its voice and digital agent, Glia Banker. This release allows financial institutions to tune the conversational flexibility of their AI agent on a topic-by-topic basis, utilizing Strict, Rephrase, and Compose modes to balance automated guidance with rigid safety constraints.
Bears on: Agent capability
View sourcePricing
Not public; quoted through sales under a predictable model with no metering on seats, minutes, or AI usage
flat platform subscription, not metered by seats, minutes, or AI
What is public
No list pricing and no dollar figures anywhere on the pricing page. Glia publishes the shape instead: $0 per minute, $0 per seat, $0 per token, with unlimited seats and unlimited usage including API, IVR and chat and voice transcripts. Tiers are sized on supported channels, automation goals, growth stage and industry needs, and are quoted through sales.
Billing mechanics
Fixed-price platform subscription sized to the institution rather than metered per interaction. Tiers are selected on supported channels, automation goals, growth stage and industry. Seats, minutes, tokens, bots and events are explicitly not billable units, and AI is described as embedded and unlimited rather than a metered add-on.
Cost watchouts
Core and online banking integration work and professional services can sit outside the platform subscription.
Variable cost rationale
Glia markets a predictable pricing model with no metering on seats, minutes, or AI usage, so cost stays largely flat as interaction volume grows rather than scaling per action.
Additional watchouts
The predictable pricing claim still requires a quote, so confirm which tier covers what across Glia Voice, Glia Digital and the Glia AI products (Banker, Outreach, CoPilot, Analyst and Agentic Workflows), and how implementation and core banking integration work are priced. Tiers are described as varying by supported channels and automation goals, so channel and product scope is the real price lever even though usage is not.
Sales call required
Yes, required for paid access
Free / trial
Demo and consultation on request; no public free tier
Key ambiguities
No public rate is published despite the transparency positioning, and the tier structure is named but not enumerated, so both the figure and the tier boundaries require a quote.
Related vendors
- Decagon — AI agent platform purpose-built for customer support, with deep…
- 5.Y — Singapore early-stage platform whose GLUCOSE product deploys…
- Ada — Agentic customer experience platform whose AI agents autonomously…
- Aisera — Enterprise agentic AI platform that orchestrates specialized agents…
- ASAPP — Generative AI for enterprise contact centers that plugs into…
- Assembled — AI customer support orchestration platform that unifies autonomous…
Alternatives to Glia
The closest documented capability profiles to Glia among customer support agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Five911.5 / 14Adds documented Model Flexibility & Routing
- Tidio9.5 / 14A lighter documented profile than Glia
- Zingtree9.5 / 14A lighter documented profile than Glia
- Ada11.0 / 14Adds documented Model Flexibility & Routing
- Bland AI12.0 / 14Fuller documented coverage on Memory & State Persistence and Deployment & Data Residency
- Fin11.0 / 14Fuller documented coverage on Deployment & Data Residency and Testing, Debugging & Optimization
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded