Olark vs Hatz AI
Side-by-side trajectory, velocity, and editorial themes.
Olark rebuilds around v2 — new layout, AI Assistants surface, in-product bot evaluation.
Olark is mid-rollout of v2, a full interface rebuild that landed in mid-2025. The AI Assistants area is now structured around Knowledge / Persona / Evaluation tabs, with bot review and feedback (thumbs up/down on individual answers) happening directly in-product. Knowledge ingestion has expanded to JSON and ZIP files plus much larger website crawls. Smaller v2 quality-of-life touches keep landing — collapsed agent menus, static offline messages, Group ID exposure for API users.
The product is being rebuilt around AI-assisted chat support, not bolted on. The Evaluation tab in particular signals a closed-loop training direction — agents tune the bot from real conversations rather than configuring it abstractly. v2 is also shedding classic settings page by page; expect that migration to keep producing visible incremental wins.
Next moves likely deepen the bot evaluation loop — automatic quality scoring, suggested knowledge updates from low-rated answers — and continue retiring classic surfaces. A pricing/tiering revisit around AI usage is plausible once the v2 migration has run its course.
Hatz is building the governed multi-tenant control plane for MSPs running AI.
Hatz AI is shipping fast on two axes: a multi-tenant MSP control plane (per-tenant integration and custom-MCP enable/disable, provisioning templates that fix a tenant's models, apps, and permissions at creation, usage dashboards, download restrictions) and a broadening model and integration layer (Opus 4.8, an LLM Gateway extended to Anthropic, Gemini 3.5 Flash, Auto model-selection modes, and a steady stream of official MCP integrations). Recent releases emphasize admin control over which capabilities each tenant gets.
Hatz is positioning as the governance and provisioning layer for MSPs delivering AI to many client tenants — not just another chat product, but the control plane that decides which models, tools, and integrations each tenant can touch. Model and integration breadth is table stakes; the differentiation is per-tenant control.
Expect more tenant-governance depth — finer permission and policy controls, more provisioning automation — alongside the continuing cadence of new model and MCP-integration additions.
See more alternatives to Olark →
See more alternatives to Hatz AI →