Frequently asked questions

augLab is a no-code platform for building AI agents, multi-agent teams, and automated workflows—you describe outcomes and connect tools instead of shipping custom application code. You bring your own model providers and API keys (BYOK), wire more than three hundred language models and over seventy integrations, and iterate from a visual dashboard built for operators as well as technical owners. Organizations use augLab for customer support and ticket triage, sales and revenue operations, research synthesis, marketing and content creation, and other line-of-business automation where language models need reliable tooling. Teams can start with a single assistant and graduate to coordinated agents or approval-gated pipelines as requirements mature.

You pick a model that fits quality, latency, and budget, attach tools such as email, CRMs, chat, or internal APIs, and describe in plain language what you want your agent to accomplish. augLab uses that brief to draft structured instructions you can edit, test, and version like any other operational asset—not a pile of loose prompts in a scratch document. From there you interact through chat, call agents from your own apps via HTTP APIs, run them on recurring schedules, or embed a hosted widget where customers or employees already work. Execution history, failures, and optional human approvals stay visible in one place so you can tune behavior without redeploying services.

The catalog spans more than three hundred models from OpenAI (including GPT-4o and reasoning models such as o1), Anthropic (Claude), Google (Gemini), xAI (Grok), Mistral, and many additional providers exposed through OpenRouter and first-class connectors. You can swap models when an agent under-performs or when economics shift, typically without rebuilding your tool wiring or knowledge attachments. Because augLab is BYOK, token billing follows each provider’s public rate card rather than a hidden markup layered by us. The dashboard reflects what your workspace can call today so connectivity stays explicit instead of tribal knowledge.

You do not need to write code to use augLab day to day. The product is configured through the web dashboard: point-and-click tool setup, knowledge uploads, scheduling, and a visual workflow builder replace most of what would otherwise be YAML and microservices glue. When you create an agent, AI-assisted flows help turn a short description into a coherent instruction set you refine with tests—not an opaque black box you cannot inspect. Engineers can still integrate via HTTP and extend with MCP or custom connectors, but coding is an accelerant, not a prerequisite. That design lets revenue, support, and operations teams own agents while platform groups retain governance.

augLab uses an enterprise engagement model with custom pricing tailored to each organization. Every engagement includes full platform access with unlimited agents, workflows, and runs. Billing is invoice-based. You pay augLab for the platform and implementation services; large language model usage bills directly to the providers associated with your API keys, so inference economics stay transparent and auditable. We do not mark up provider token pricing — you reconcile model spend with OpenAI, Anthropic, Google, or whichever vendors you enable.

API keys, agent run history, and file uploads are encrypted at rest with AES-256 and are only decrypted in memory when execution needs them, which reduces the chance of plaintext leakage through logs or support dumps. Multi-factor authentication is available for all users and is enforced for HIPAA-enabled organizations. File downloads use short-lived, scoped tokens so access credentials are never exposed in URLs. Comprehensive audit logging records material configuration changes and access patterns administrators can use during security reviews or incident response. BYOK means conversational traffic flows between you and your chosen LLM vendors under their contracts and data policies, with augLab orchestrating rather than silently proxying for resale. We bias defaults toward least retention of sensitive artifacts and operational telemetry that demonstrates reliability without warehousing customer content indefinitely.

HIPAA compliance is available on any plan. When enabled, augLab enforces multi-factor authentication for all organization members, encrypts all data at rest including agent runs, file uploads, and credentials, restricts LLM providers to those with signed Business Associate Agreements, and provides configurable data retention (defaulting to the six-year HIPAA minimum), comprehensive append-only audit trails, and complete data deletion capabilities. The HIPAA flag is toggled per organization and a BAA between your organization and augLab is required before handling PHI. Teams handling PHI should involve compliance reviewers during onboarding so controls map cleanly to your risk assessment.

Teams get access to more than seventy integrations spanning Slack, Gmail, GitHub, HubSpot, Jira, Google Sheets, HeyGen, ElevenLabs, Gamma, compliant web search, managed databases, and generic HTTP bridges for bespoke internal services. MCP server support lets you adopt new tool protocols as vendors ship them without waiting for a dedicated connector tile in our catalog. Most integrations are configured through guided setup flows in the dashboard, with credentials stored using the same encryption posture as model keys. You can mix SaaS connectors with custom endpoints so automations reflect how your company actually works, not a forced lowest-common-denominator stack.

An agent is a single AI assistant with one model configuration, tool permissions, knowledge bindings, and instructions that describe how it should respond or act. A team groups multiple agents that collaborate autonomously—handing subtasks off, reviewing each other’s outputs, or specializing in parallel so complex objectives do not collapse into a brittle megaprompt. A workflow is a structured multi-step pipeline with explicit execution order, branching logic, schedules, and approval gates when humans must sign off before downstream automation proceeds. Choose agents for conversational flexibility, teams when autonomy benefits from specialization, and workflows when repeatability and governance demand a visible control surface.

Yes. Bring-your-own-keys is a core design choice for augLab rather than an aftermarket add-on. You paste provider credentials into secure storage, and runtime traffic for language models goes directly to those vendors under their terms and telemetry. augLab does not mark up token pricing: you see provider invoices or dashboards for model usage distinct from platform fees. Rotation and scoping of keys remain under your control, which matters for least-privilege security reviews and for finance teams forecasting AI spend. If a key is revoked or hits quota, failures surface in execution logs so you can fix forward quickly.

When memory is enabled, agents persistently learn from conversations. They maintain a user profile with structured facts like name, role, and preferences; track unstructured observations and patterns; remember session goals for multi-step tasks; and build up knowledge about external entities like companies or projects. You control learning mode: "agentic" lets the agent decide what to remember, while "always" extracts insights after every response. All stored memories are visible in the Memories tab on each agent page, where admins can review, delete individual entries, or purge all data for GDPR and compliance. For HIPAA organizations, memory data is protected with field-level encryption on top of database encryption at rest. Enable memory from the agent builder with a single toggle—no code required.