LoginSign Up
← Back to all sparks
H

Hatz AI

SUPPORT
Velocity6.3

Hatz sells through MSPs, and its roadmap has started to serve the partner more than the end user

msp-channelphone-agentsmodel-selectormulti-tenant-adminshadow-ai
◆Current state
Most of Hatz's window is model-selector maintenance — Claude Opus 5, Kimi K3, GLM 5.2 Fast, DeepSeek V4 Flash, the Gemini 3.x line — plus a steady buildout of the phone agent, which has gained pausing, duplication, reassignment, drafts, silence tolerance, business hours, and caller memory. Against that routine, Hatz Activate stands apart: a rollout console built for the managed service providers who resell the platform.
◆Where it's heading
Two arcs run in parallel. The phone agent is being hardened into a product an MSP can deploy repeatedly across client sites — duplicate it onto a new number, reassign it to a different tenant, cap its minutes — while Activate supplies the partner with the shadow-AI scan and enablement material that starts the sales conversation. Model additions continue but carry no direction; they are table stakes restated every few weeks.
◆Prediction
Expect more tenant- and partner-scoped administration, since renaming tenants and invoicing over the API both landed this window and point at MSPs managing many clients programmatically. The phone agent's usage-based pricing suggests metering and limits will keep expanding ahead of any new end-user capability.

◆Recent moves

  1. 5d ago

    Attachment-Only Messages, Tenant Renaming, Invoices Over the API, Phone Agent Drafts and Silence Controls

    Attachments can be sent with no message and clear after sending so they are not re-billed, draft text stays pinned to its own chat, and phone agents gain save-as-draft on exit plus a configurable silence tolerance. Tenant renaming and invoices over the API continue the multi-tenant administration thread that matters most to resellers.

  2. 12d ago

    DeepSeek V4 Flash 0731, Self-Reporting Credit Usage, Phone Agent Pause/Reassign/Duplicate, Artifact Presentation Mode, Auto-Tool Selection Notice

    DeepSeek V4 Flash 0731 joins the selector with a 1M-token context, but the operational work is more consequential: phone agents can be paused, duplicated onto a new number, or reassigned to another tenant. Those three verbs are what let a partner run the same agent across many clients.

  3. 14d ago

    Hatz Activate: a Shadow AI scan and an MSP rollout console

    ⚡ SPARK

    Against a window otherwise made of model additions and platform polish, Activate is the one release aimed at a different customer entirely — the MSP reselling Hatz. It turns partner enablement into an ongoing product surface rather than a one-time onboarding step.

  4. 19d ago

    AutoTool v2, Optimized Phone Agent Pricing, MSP Workshop Categories for Agents

    AutoTool v2 improves tool selection, mid-conversation discovery, and recovery when tools fail, and defaults to on for new chats. The phone agent also drops its license requirement for usage-based billing with per-number monthly minute limits — a pricing change that lowers the barrier to standing up an agent per client.

  5. 20d ago

    Kimi K3, GLM 5.2 Fast

    Kimi K3 and GLM 5.2 Fast join the model selector for chats, agents, and workflows, one trading latency for depth and the other optimized for real-time work. Selector maintenance of the kind that recurs throughout this window.

  6. 25d ago

    Claude Opus 5

    Claude Opus 5 becomes available across chat, apps, agents, and workflows. A single-model addition carried almost entirely by the vendor's own quoted description.