Saient is a local AI desktop app — chat, a real coding agent, image & video generation, text-to-speech, and LoRA training. The 2.0 beta adds native node workflows and inpainting, all running on your own hardware. Local inference needs no Saient account, licence key or hosted-inference key.
[>_<]
/|#|\
/ \
Task-focused local AI tools in a single native window. Language, image, video and speech use model-specific local runtimes.
Talk to a compatible local GGUF model. Renders live HTML artifacts in a side pane — ask it to build a tool or a game and watch it run.
A real agent with a PTY terminal and a tool-use loop (read · ls · write · edit · bash). It reads and edits project files, runs commands and checks its work using your local model. The 2.0 beta improves recovery and keeps track of completed actions. Review its results; coding quality depends on the model and task.
SDXL text-to-image with LoRA support and live progress. In the 2.0 beta, paint a mask to regenerate part of an image with inpainting.
Text-to-video and image-to-video with Wan & CogVideoX — frame control, live progress, and an optional refine & upscale pass.
Ask questions about any image — describe it, read its text, count objects — with a local Moondream2 analyzer. Paste a screenshot and go.
Kokoro American and British voices with speed control and local 24 kHz WAV output.
Train SDXL LoRAs with a dataset cleaner and merge compatible image checkpoints. The 2.0 beta fixes the diffusion-model pickers and removes unsafe checkpoint loading.
Connect prompts, image generation, masks, inpainting and previews in Saient’s own workflow editor. Save graphs, undo edits and rerun your workflow. No ComfyUI installation required.
Local GGUF chat uses the Quartz inference engine. Creative studios use model-specific local runtimes; optional Python tools are installed through setup.
Step-by-step pages with real screenshots, exact model boundaries, hardware notes, proof status, downloads and source links.
The free phone app: local AI chat, Lightning SDXL images and an offline video maker for your music, photos and clips. On Google Play.
Install, add a model, choose CUDA or CPU, tune settings and inspect a real image output.
Text-to-video, image-to-video, actual model routes, memory boundaries and MP4 proof.
GGUF setup, architecture support, RAM/VRAM sizing and a live local response check.
The direct boundary between host-model proposals and Saient's tools, state, memory and control loop.
A full evidence-first paper on durable memory, checkpoints, goal continuity, long-horizon agency and the failures behind the claims.
Dataset preparation, exact trainer scope, hardware unknowns and the current proof boundary.
Compatible image checkpoint merges, the repaired beta model picker and safe loading boundaries.
Kokoro voices, CPU execution, WAV output and current evidence status.
Compare dedicated studios, the new native workflow editor and local agent with ComfyUI’s graph and extension ecosystem.
Open the Agent screen, choose a project and local model, then give it a scoped goal. Saient can run a plan→act→verify loop within the permissions you grant.
[>_<] /|#|\ / \
Get the new agent fixes, native node editor and inpainting. This is a development beta; the previous desktop release remains available below.
Download checksums · Beta source code
Beta requirements: Linux x86_64 with glibc 2.39 or newer (Ubuntu 24.04 or newer). Optional NVIDIA acceleration requires a compatible driver. See the beta manifest for the bundled GPU architectures and tested hardware. Models download separately. Install the beta manually; stable in-app updates remain on 1.0.24.
The 2.0 beta is currently a Linux Debian package. Agent output still needs review: harder local coding tasks can fail. Beta setup and limitations →
Windows signing. The 1.0.24 Windows installer is not Authenticode-signed; Windows may show an unknown-publisher or SmartScreen prompt. Its updater signature is separate. Verify package hashes against the linked release manifest.
Verification. The 2.0 beta has Linux controller, package and native desktop checks, including recovered 7B and 14B agent tasks. This does not establish general coding reliability. The previous 1.0.24 Windows build passed native CI checks; its installed GUI/GPU session remains unverified. Each download’s manifest records its own evidence and limits.
Windows prerequisites. The bundled inference engines require the Microsoft Visual C++ v14 Redistributable (x64). Install it from Microsoft if it is missing. Python 3.10+ is also required for the Saient terminal. NVIDIA GPU acceleration requires a compatible NVIDIA driver. The CUDA runtime library is bundled; the CUDA toolkit is not required.
Thank you. To everyone who has downloaded Saient, reported something broken, or just followed along — genuinely, it helps, and it is the reason the rough edges get found at all.