Portfolio

Systems we've designed and built - from concept to production.

Client Engagement

On-Premise AI · Document Intelligence

Pythia

A self-hosted document intelligence platform for litigation and legal review - built on NVIDIA DGX Spark hardware. Processes entire disclosure sets locally: semantic search, conversational Q&A with citations, interactive timelines, and OCR. No data ever leaves the device, satisfying Legal Professional Privilege by design. The same applies well beyond litigation: any legal, compliance or commercially sensitive material that has to stay on-premise, or that cannot be sent to a cloud model for AI processing.

About Pythia

Document Ingestion at Scale

150,000+ files across all formats (PDF, DOCX, MSG, EML, XLSX, TIF, and more). Checkpoint-based processing survives interruptions.

Semantic Search & RAG Q&A

Ask questions in plain language. Every answer cites its source document and page number - no hallucination from the AI's own knowledge.

Interactive Timeline

Chronological view of case events extracted from metadata. Drill from year to month to individual events, filtered by disclosure side.

Complete On-Premise Privacy

OCR, inference, vector search, and storage all run locally on DGX Spark. No cloud APIs. Satisfies Legal Professional Privilege architecturally.

Example conversations
You

Find all documents referencing the March 2023 board meeting and summarise the key decisions

AI

Found 14 relevant passages across 6 documents. Key decisions recorded: approval of revised share structure (Exhibit C-112), appointment of interim CFO (Email D-0447), deferral of acquisition vote to Q3 2023 (Board Minutes C-089).

You

Show me all emails between the claimant and defendant between January and April 2022

AI

Retrieved 38 emails. Earliest: 4 Jan 2022 (REF: D-0012). Latest: 29 Apr 2022 (REF: D-0891). 12 marked confidential by producing side.

You

Build a timeline of contractual deadlines mentioned across all documents

AI

Extracted 23 deadline references. Key dates: 15 Feb 2022 (payment milestone, Contract §4.2), 1 Jun 2022 (option expiry, Addendum B), 30 Sep 2022 (longstop date, original SPA).

Client Engagement

Model Context Protocol · Consulting Engagement

MCP Bridge - SharePoint & Power Automate

We built an MCP (Model Context Protocol) bridge that gives Claude direct access to a client's SharePoint environment and Power Automate flows. Instead of copy-pasting data into a chat, the team can ask AI to query, create, and update SharePoint records - and diagnose broken automations - through natural conversation.

About the SharePoint Bridge

SharePoint Lists & Structure

Create lists, add columns of any type, model entity relationships, inspect full schemas.

Data Management

Query items with filters and sorting, create, update, and delete records, run bulk transforms.

Power Automate

List flows, inspect definitions, view run history with error diagnostics, trigger manual flows.

Live Documentation

AI searches current Microsoft docs before answering - no outdated or hallucinated API guidance.

Example conversations
You

List all SharePoint lists on my site

AI

Found 8 lists: Tasks, Clients, Projects, Invoices, Contacts, Documents, Assets, Settings.

You

Add a lookup column on Tasks pointing to the Clients list

AI

Column 'Client' (lookup → Clients) added to the Tasks list successfully.

You

Show me the last 5 failed Power Automate runs on 'Invoice Sync'

AI

Found 5 failed runs. Most recent error: 'Connection timeout to Xero API' - occurred 3 times in the last 24 hours.

Client Engagement

Sales Outreach Engine · On-Premise

GTM

A custom system that runs B2B prospecting end to end. It researches prospects, writes personalised emails in the sender's voice, schedules and sends safely, and keeps every account moving - with a human reviewing rather than retyping. It runs on an NVIDIA DGX Spark, not rented cloud.

About GTM

Researches every prospect

Profiles each company and contact, then writes a structured dossier with a personalisation hook. 6,675 prospects profiled across 9 countries.

Writes in the sender's voice

Automated editorial gates check mechanics, facts and voice-match, and strip the tells that get cold email deleted. Anything that fails is held back.

Sends safely and on schedule

Idempotency keys and fail-closed guards mean nothing double-sends or loops. Every send is logged and auditable.

Local by default

Open-weight models on-device do the bulk of the work. A frontier model is called only across a single auditable boundary, with identifying detail held back.

Example conversations
You

Which accounts have gone quiet in the last three weeks?

AI

12 accounts with no inbound since 20 Jul. Highest value: SSAB (licence renewal, Q3), Eidsiva (pilot scoped 2 Jul). Follow-ups drafted for review.

You

Draft the follow-up for SSAB in my voice

AI

Draft ready. Passed mechanics, fact-check and voice-match. Held one claim for review: renewal date cited from CRM, last updated 14 Jul.

entityvault.tutto.one

EntityVault

A privacy-first entity management platform with tokenized data storage, collaboration requests, and entity search. Built for organisations that take data privacy seriously.

Visit EntityVault
EntityVault - Home

tracker.tutto.one

AI ROI Portal

A consulting dashboard for tracking AI adoption ROI, usage metrics, project milestones, and cost savings. Helps teams measure the real impact of automation.

Visit AI ROI Portal
AI ROI Portal - ROI Dashboard