Models come
and go. SynthWise remains.

AI automations and software for work, life and study

AI moves fast. Providers change week by week, models are superseded before you've learned their names, and whole new technologies appear out of nowhere. Chase every one of them and you spend your time copying and pasting the same context into yet another window.

SynthWise doesn't chase. Its constant is context - the heartbeat of the platform. Your synthesized wisdom - documents, definitions, notes, drafts, the things you've learned - lives in one place, so you can move between providers and services without cloning or re-explaining a thing. Swap models when better ones arrive. Keep a receipt for every call one of them makes. And it stays yours: everything exports in standard formats, at any time.

Book a working session How the receipts work →

What runs here

Chat. Assistants. Automations.
Design. Apps. Images. Prototypes. Video.

Wired into the tools you already use - at work, in study, at home. Every step visible on the run, every model call receipted.

PROVENANCE LOG
synth_wise_ai_operations · live sample
time asset model tokens
20:36:22 arrears_report_q2.csv GPT-4.1 EXTERNAL 420
20:36:20 supplier_po_split Claude Sonnet 4 EXTERNAL 1,151
20:36:19 broker_questionnaire Llama 3.3 INTERNAL 1,882
20:36:17 member_ledger_export Gemini 2.5 EXTERNAL 2,613
20:36:16 loan_book_snapshot Mistral INTERNAL 3,344
20:36:15 dispatch_status_feed GPT-4.1 EXTERNAL 4,075
20:36:13 quote_7731.json Claude Sonnet 4 EXTERNAL 4,806
20:36:12 invoice_final_0419 Llama 3.3 INTERNAL 5,537
20:36:10 arrears_report_q2.csv Gemini 2.5 EXTERNAL 1,068
20:36:09 supplier_po_split Mistral INTERNAL 1,799
20:36:08 broker_questionnaire GPT-4.1 EXTERNAL 2,530
20:36:06 member_ledger_export Claude Sonnet 4 EXTERNAL 3,261
unknown providers classified EXTERNAL by default retention: policy-bound

Why a nervous system

AI as infrastructure,
not a side tool.

Most of the value in AI is trapped where it was made: one person's chat history, one vendor's assistant, one laptop. The context you build up - documents, definitions, drafts, the things you have learned - cannot easily be shared with a colleague, or moved to a better model when one arrives. It never becomes process.

SynthWise is the layer that holds that context and connects it to the systems you already run - accounting, CRM, files, databases - and to whichever models are best this quarter. Automations you can specify, run and trust, with a record of which model processed what data, whether it left your jurisdiction, who authorised it, and what it cost. Everything exports in standard formats, so the next move is yours.

The platform

One substrate.
Five layers.

Start with a chat window. End with assistants running the processes you depend on, unattended. Every layer sits on the same context, the same permissions and the same provenance record.

Five layers in full →

LAYER 04 - AUTOMATION

Automations

Triggers, nodes, conditions and human gates, wired into any system with a REST or webhook API - Xero, Slack, HubSpot, Trello and Google Calendar out of the box, and many more. Quote-to-invoice, month-end reporting, customer onboarding, supplier chasing, approvals, specified once, run forever.

What we do

We build with you.
You keep the system.

We work with organisations to design and ship software, and to put AI into the work itself. The platform underneath is open to anyone with things to run - a business, a course, a household. SynthWise is the scaffolding it all sits on: assistants, agents and custom models, on infrastructure you control, with a provenance record on every call.

01 - ADOPT

Private AI

LLM-agnostic chat and Studios, so your context lives in one place instead of being spread across a different app for every use case - prototyping, video generation, company and personal data stores. Every major vendor and many niche ones, via OpenRouter or direct. Bring your own key, or bring your own model: assistants, agents and custom LLMs deployed on your own hardware, sandboxed, ring-fenced, actually private. It is not all or nothing: model access is set per workspace, so you mix and match - frontier and image models where the work wants them, a private model where the data demands it. We map the processes you depend on and run them here, with permissions and receipts, not a chatbot on the side. No provider lock-in, from day one.

Read the full brief →

02 - BUILD

Software

Software that works with you and for you. For organisations: data lakes, warehouses, reporting, integrations and products, designed and shipped through our Studios, sitting under SynthWise or standing on their own - a governed lake for a UK credit union is running on it today. The same Studios are yours for the smaller things: a page for a club, a revision tool for a course, a tracker for the household paperwork, a prototype you want to show someone on Friday. Describe it, watch it take shape, change your mind halfway. The tooling is why it moves at a speed that wasn't available before, and why a small thing no longer needs a big budget.

Read the full brief →

03 - RUN

Platform

A place to manage and deploy agents while software interfaces are still catching up - for a business, a course of study or an ordinary life, on the same substrate. A finance team's month-end, a student's reading list, a family's bills and renewals: each is a process, and each can have an assistant with its own tools, its own bounds and a receipt for everything it did. Data and its provenance will only get more valuable, so you keep both, and export is standard. SynthWise is engineered so natural-language features can replace traditional ones as the experience gets simpler, without throwing the system away.

Read the full brief →

Compliance in the substrate

Automate the work.
Keep the receipts.

Anyone in the UK or EU who sends personal data to a language model inherits GDPR and EU AI Act exposure. In most setups there is no record of what went where. On SynthWise, every data asset that touches a model produces a provenance record, whether you asked for it or not.

  1. 01 Which asset, which model, which provider, internal or external to your infrastructure
  2. 02 Token counts, timestamps, and the actor (user, assistant, automation or agent) that authorised it
  3. 03 Unknown providers classified external by default, because that is the safe assumption
  4. 04 EU AI Act impact-assessment templates that walk you through your actual obligations
  5. 05 Retention policy and full export, leaving is a button, not a negotiation
How the receipts work →
AI_OPERATION_RECORD #81853f12
asset quote_7731.json
workspace -
automation -
model claude-sonnet-4
provider anthropic
classification EXTERNAL
jurisdiction_exit true
tokens_in / out 1,842 / 611
authorised_by -
risk_class unregistered - EU AI Act Annex III: n/a
retention policy: platform-default
recorded_at 2026-09-28T09:38:06Z
exportable · queryable · yours

Two motions. One platform.

Built to hold up under audit.
Context travels wherever you go.

The case files →
REGULATED OPERATIONS LIVE

Transave - UK credit union

Data from multiple vendor systems consolidated into a governed data lake and warehouse, producing live operational reporting - arrears, exposure, member activity - under the audit and data-protection obligations a credit union carries. Built at a speed and price that was not previously available to an organisation this size.

vendor exports → governed lake → warehouse → live reporting

Read the full brief →
CREDIT UNIONS & REGULATED SMES DEPLOYABLE

The governed data lake - built once, for the sector

The vendor-export → lake → warehouse → reporting pipeline built for Transave isn't bespoke to them. The connectors, the schemas, the arrears and exposure reporting and the audit trail are reusable. For the next credit union, this is configuration and onboarding, not a ground-up project.

your vendor exports → same governed lake → your reporting

Read the full brief →
SME OPERATIONS ONBOARDING

299 Lighting - order-to-cash

A quote accepted in Quoter flows through contact, project, item, quote and invoice creation in Xero, a human payment gate, purchase-order splitting by manufacturer, supplier dispatch monitoring, carrier booking, proof of delivery and invoice finalisation. Built once; near-cloned for the next business like it.

Quoter → Xero invoice → ⏸ human gate → PO split → dispatch → POD → final

Read the full brief →

The line we won't cross

We don't want
your data.

It is the last thing we want. Our income comes from services you pay for because they improve your work, so we never need to mine what you store with us, sell it, or make leaving difficult.

Everything auditable. Everything exportable. Every AI operation recorded. Because that is how a system built on these principles has to work.

Next step

Build it with us.
Then run it yourself.

Bring one process, the one everybody dreads, the one that lives in a spreadsheet and three inboxes - at work, in study or at home. We will map it, automate it on SynthWise, and hand you the receipts. If it works, the next process is mostly configuration.

WORKING SESSION - 60 MIN - NO DECK

  1. 01 You describe the process as it actually runs today
  2. 02 We sketch it as an automation, live, on the platform
  3. 03 You leave with a written scope, assumptions and an honest estimate
SYNTHWISE · NETHERLANDS / UK Contact
AI-NATIVE PRIVACY-FIRST SELF-HOSTABLE SW·001