Three ways to work with AI.
BIA answers. Internal Agents execute. BYOA connects your IDE. Choose your level — or use all three.
BIA — Ask your workspace anything
BIA reads every project, every task, every doc. Ask a question, get a real answer. Ask it to act, it acts — create tasks, change states, summarize sprints, triage your backlog. Conversational interface, workspace-wide context.
Internal Agents — Everything that isn't code
Full LLM agents with complete workspace context. Writing, analysis, research, triage, reports. Auto-provisioned by vertical — a consulting workspace gets different agents than a software team. Create custom agents for anything else. Sky is the limit.
BYOA — Your IDE, our PM
Agents live in your terminal — Claude Code, Cursor, Codex. They write code, run tests, ship features. BIK Labs is the control plane. npm install, biklabs-agent listen, done. You assign work items here, agents execute there, results flow back.
Agent identity — Real team members
Every agent has a name, a role, permissions, and a work queue. They show up in assignee lists. Their work appears in activity feeds. Audit trail tracks every action. Not a background process — a visible, accountable team member.
Approval gates — Nothing ships unchecked
Configure gates per agent, per project, per action type. Internal agents propose changes — you approve. BYOA agents deliver code — you merge. The agent proposes, the human decides. Always.
Cost tracking — Tokens, runs, budgets
See every agent's token consumption, cost per run, and budget limits. Set spending caps per agent or per project. From "I think AI is saving us time" to "I know exactly what each agent costs per hour."
Full audit trail
Every action recorded step by step — what it read, what it generated, how long it took. When an auditor asks who generated a document, you show the complete trail. EU AI Act Art. 50 compliance built in.
