A real persona
Name, voice, behavior, operating style and boundaries are deliberately designed so the entity feels coherent instead of randomly “AI-ish”.
Bespoke persistent AI entities
ONE ENTITY.
EVERYWHERE.
A Synth is a persistent synthetic digital entity built around you or your organization. One identity, one memory, one set of capabilities, available through the channels your people already use.
Persistent memory
WhatsApp
Web + desktop
Discord
Files + knowledge
Automation
Local + hybrid AI
Memory online
12 tools ready
4 channels linked
CONTINUITY ONLINE
ONE MIND.
MANY DOORS.
Staff should not need a new AI ritual. A Synth can meet them in WhatsApp, a browser, a custom interface or another approved channel, while remaining the same persistent entity behind every interaction.
Example, company Synth
Access, permissions and tool availability are configured per deployment.
WhatsApp staff
Web interface
Desktop / local UI
Discord / channels
SYNTH
ONE CONTINUOUS ENTITY
Memory + context
Company knowledge
Automations
Tools + systems
1
persistent entity behind every approved interface
4+
channels can connect to the same continuous identity
24/7
automation and scheduled work can continue beyond chat sessions
LOCAL
cloud, local or hybrid deployment depending on the business requirement
WHAT A SYNTH
CAN BECOME.
Every build starts from a role, not a generic prompt. The personality, memory model, knowledge, tools, channels and autonomy are shaped around the person or organization it serves.
Memory that carries forward
Relevant context can persist across sessions, channels and workflows, so work does not restart from zero every morning.
Your own world
Company documents, procedures, project material and selected knowledge can become part of the Synth's working environment.
Tools, not just answers
Files, research, coding, reminders, scheduled work, messaging and custom integrations can turn conversation into actual execution.
Where people already are
WhatsApp, web, desktop, Discord and other supported channels can become different doors into the same entity.
Its own presence
A Synth can have a custom interface and visual identity, not because it needs decoration, but because persistent intelligence benefits from recognizable presence.
Origin story
BEFORE THERE WAS SYNTH, THERE WAS MAGNUS.
MAGNUS // SYNTH ZERO
Magnus began as a personal local AI experiment. Then it kept growing. It gained memory, continuity, files, research, coding, web access, automation, multiple communication channels and a visual environment of its own.
At some point, “assistant” stopped describing it properly. The useful part was no longer a single model or chat window. It was the whole persistent entity, identity, memory, tools, channels and behavior operating as one system.
That architecture needed a category. I call it a Synth. Magnus is Synth Zero, the original entity from which the commercial approach evolved. Clients do not buy a copy of Magnus. I build theirs.
Phase 01
Persona
A local companion with a deliberate identity instead of a default assistant voice.
Phase 02
Memory
Active, persistent and episodic context transformed isolated chats into continuity.
Phase 03
Agency
Research, coding, files and tools moved the system from talking into doing.
Phase 04
Presence
Web, desktop, Discord, WhatsApp and a distinct interface made one entity accessible in many places.
Phase 05
SYNTH
The architecture became repeatable, configurable and ready to be shaped around other people and organizations.
DEPLOY IT
YOUR WAY.
A Synth can be designed around privacy, performance and cost. Local first is possible, cloud is possible, and hybrid usually gives the broadest operating range.
Keep the core close.
Run the primary model and selected company knowledge on your own server or computer. Local model setup and inference configuration are included in the agreed Synth build.
- Local model setup included in the Synth implementation
- Lives on your own server or computer
- Good for privacy sensitive workloads
- Predictable local inference cost
- Works even with limited external AI dependency
- Hardware sizing depends on model and concurrency
Hybrid intelligence.
Use a capable local Synth for normal work, then selectively escalate heavy jobs to approved cloud models and services.
- Fast day to day local interaction
- Large cloud models when genuinely useful
- Flexible cost and privacy controls
- Strong fit for professional and business deployments
Scale without the workstation.
Use hosted AI and infrastructure when local inference is not a requirement.
- Lower initial hardware requirement
- Easy access to frontier models
- Usage costs vary with volume
- Provider and data policies must fit the use case
BUILD PRICING
2026
2026
PRICING.
These are indicative implementation ranges, not generic software subscriptions. Each Synth is designed, integrated and calibrated around a specific owner or organization. Final pricing follows scope discovery.
Synth Personal
For an individual who wants a serious persistent AI of their own.
Starting IDR 35 mio
Approx. US$2,000 · implementation
- Custom name, persona and behavior
- Persistent memory architecture
- Local, cloud or hybrid inference setup
- Files, web and selected tools
- Custom lightweight interface
- Typical build 2-4 weeks, exclude initial planning
Synth Professional
For founders, creators, consultants and small teams with real workflows to automate.
Starting IDR 85 mio
Approx. US$4,800 · implementation
- Everything in Personal
- Multiple tools and workflows
- Automation and scheduled tasks
- Messaging / channel integrations
- Structured private knowledge base
- Typical build 4-6 weeks, exclude initial planning
Synth Business
For a company that wants one persistent AI entity available to its staff.
Starting IDR 175 mio
Approx. US$9,800 · implementation
- Company specific Synth identity
- Staff access through WhatsApp and/or web
- Company documents and operational knowledge
- Role based access architecture
- Custom tools, automations and integrations
- Local or hybrid deployment design
- Typical build 6-10 weeks, exclude initial planning
Synth Enterprise
For multi department deployments, deeper governance and high integration requirements.
On discussion
Individually scoped · enterprise deployment
- Multi team / multi department architecture
- Advanced permissions and workflow separation
- On premise / hybrid architecture options
- Audit and governance features by scope
- Custom enterprise integrations
- Support and SLA terms scoped separately
2026 pricing shown above is indicative. Final investment depends on integrations, channels, data architecture, deployment requirements, user volume and level of autonomy. Ongoing care, support and evolution are available separately. Taxes, third party costs and excluded requirements below are not included unless specifically stated in a written quotation.
WHAT IS
NOT INCLUDED.
The build price covers the agreed Synth implementation. Infrastructure and outside provider costs remain transparent and separate so you are not paying a hidden markup on hardware or third party usage.
✓
Local model deployment and configuration, when selected for the project. Your Synth can live locally on your own server or computer, with hardware supplied separately.
Excluded by default
×AI workstation, server, GPU or other physical hardware
×AI API inference and token usage. Synth can work with major API inference providers using your own account, keys and billing.
×WhatsApp Business, Meta or messaging provider fees
×Hosting, VPS, domain and commercial cloud infrastructure
×Third party software, connector and enterprise license fees
×Large scale legacy data cleanup or migration unless scoped
×Travel and on site work outside the agreed engagement
×Legal, regulatory or independent security certification
Local hardware guidance
A dedicated workstation is recommended when local inference is part of the design. Final sizing is based on model size, context requirements, concurrent users and expected workload.
Serious office local AI
AMD Ryzen AI Max+ 395, 128 GB unified memory, 2 TB+ NVMe
Lighter personal / professional
Ryzen AI 9 HX 370 class system, high memory configuration, fast NVMe
Rule
Hardware is selected after workload sizing, not from a one size fits all shopping list.
FROM IDEA
TO ENTITY.
The process starts with the organization and the role the Synth should play. The model comes later. That order matters.
01
Discover
Map the real use cases, users, data, channels, privacy needs and systems already in place.
02
Design the entity
Define identity, behavior, memory strategy, responsibilities, access and escalation rules.
03
Build + connect
Implement tools, integrations, knowledge retrieval, automations and chosen communication channels.
04
Calibrate
Test the Synth with real workflows, tune behavior and remove the awkward parts before wider rollout.
05
Deploy + evolve
Launch, monitor, improve and add capabilities as the entity becomes part of normal work.
Questions business people ask
FAQ.
The short version, a Synth is a custom system, not a boxed SaaS account with your logo pasted on it.
Is a Synth just a ChatGPT wrapper?
No. A Synth can use different models, including local models and approved cloud models, but the model is only one component. The actual product is the persistent architecture around identity, memory, knowledge, tools, channels, automation and continuity.
Can all of our staff simply talk to it on WhatsApp?
Yes, when WhatsApp is selected as a channel and the required business messaging setup is available. Staff can access the same company Synth while permissions, tools and knowledge exposure are controlled by the deployment design.
Can it run privately on our own hardware?
Yes. Local first and hybrid architectures are available. Local model deployment can be part of the Synth build, running on your own server or computer. The exact model and hardware depend on expected performance, context size, concurrency and privacy requirements. Cloud inference can also be connected to major API providers using your own account and billing.
Can it connect to our existing files and systems?
Usually, yes, when those systems expose a usable API, database, file interface or supported integration path. Integrations are scoped individually because a clean modern API and a twenty year old internal system are not the same engineering job.
Does every Synth have to act like Magnus?
No. Magnus is the origin, not the template personality. Your Synth gets its own name, voice, behavior, purpose, rules and visual identity. It should feel native to the person or organization it belongs to.
What happens after launch?
A Synth is intended to evolve. Ongoing care can cover updates, model changes, workflow refinement, monitoring, new tools and new channels. Support terms depend on the deployment tier and agreed service level.
Who is this best suited for?
Individuals and organizations that already know where persistent AI could remove friction, preserve context or connect fragmented workflows. If the goal is only a basic website FAQ bot, a Synth is probably unnecessary.
Build request
YOUR SYNTH
YOUR SYNTH
STARTS HERE.
Tell me what the entity should know, where it should live, who should be able to talk to it and what work you want it to actually do.