Twin1 AI
by Twin1 AI
A permission-aware AI twin for every professional, networked so expertise moves across the firm
Twin1 AI gives each knowledge worker an AI digital twin trained only on their own work context — email, meetings, documents and connected systems — that answers on their behalf and coordinates with colleagues' twins. It is aimed at professional services firms in law, financial services and energy where the valuable knowledge sits in people's inboxes rather than in a database.
Twin1 AI pairs each professional with an AI digital twin grounded in that person's own work context — email, meeting records, documents and connected systems including Slack, Microsoft Teams, Outlook, Gmail, Google Drive and SharePoint — and then networks those twins so expertise can move across an organisation. The premise is that enterprise knowledge does not live in a database but in individual inboxes, calendars and accumulated judgement, and that a shared corporate assistant flattens exactly the personal context that makes it valuable. Each twin answers complex queries by synthesising across a person's sources, selects and orchestrates tools for multi-step work, and drafts communications in that person's style. Above the individual twins sits the Twin Network, a coordination layer that locates the right colleague, collects permission-aware knowledge and hands work between teams; an enterprise Model Context Protocol server exposes that governed context to other agents and applications. Governance is the core of the pitch: six layers of rules-based and AI-based controls span peer-to-peer and peer-to-AI exchanges, blending corporate policy, inherited permissions and human sign-off, and the user decides which sources their twin reads and who may query it. The company was founded in 2025 by Dr Lewis Z. Liu, Tom Cahn, Huiting Liu and Dr Jonathan Budd — the team behind Eigen Technologies, acquired by Sirion in 2024 — and emerged from stealth on 20 August 2026 with a $20 million seed round co-led by Bessemer Venture Partners, Tribeca Venture Partners and Aramco Ventures. Linklaters, Orrick, Dechert, Customers Bank and Aegis Energy are named customers after more than a year of deployments.
A knowledge-management or innovation lead at a large law firm, bank or professional services firm where partner expertise is the product and it is trapped in individual inboxes
Senior expertise becomes queryable across the firm under each person's own permissions, with customers reporting 30-50% of routine communications work handled by the twin
At a Glance
- Category
- Enterprise Search & Knowledge
- Pricing
- Contact for pricing
- Target Market
- Law Firm CIOs, Knowledge Management Leaders, Chief Innovation Officers, Financial Services CTOs, Heads of Professional Services
- Deployment
- Cloud-first, Hybrid, Self-hosted
- Founded
- 2025
- Headquarters
- San Mateo, United States
- Customers
- Named customers include Linklaters, Orrick, Dechert, Customers Bank and Aegis Energy; the CEO reports a pipeline of 400+ prospects
Key Features
- ✓Personal AI twin
Each twin builds knowledge solely from its user's private work context, so answers carry that individual's judgement rather than a generic corporate voice.
- ✓Twin Network
A coordination layer that locates the right colleague, gathers permission-aware knowledge and hands work between teams across the organisation.
- ✓Six-layer permission model
Rules-based and AI-based controls over peer-to-peer and peer-to-AI exchanges, combining corporate policy, inherited permissions and human sign-off.
- ✓Enterprise MCP server
Lets other agents and enterprise applications pull governed context from a single twin or the whole network without bypassing permissions.
- ✓Tool orchestration
Dynamically selects and coordinates tools to complete multi-step workflows rather than only answering questions in a chat window.
- ✓In-place workplace integration
Runs inside Slack, Microsoft Teams, Outlook, Gmail, Google Drive and SharePoint so professionals do not change where they work.
- ✓Flexible deployment and residency
Full SaaS in US, EU or Singapore, single-tenant, or private cloud, with SOC 2 Type II and Google CASA Level 3 certification.
Capabilities
Use Cases
- •Scaling partner expertise across a law firm
Junior lawyers query a senior partner's twin for judgement on a matter without consuming that partner's time, under the partner's own permissions.
- •Automating routine professional communications
Customers report the twin handling 30-50% of the communications work knowledge workers would otherwise do themselves, such as drafting and follow-ups.
- •Locating internal expertise
The Twin Network finds which colleague has relevant experience on a client, jurisdiction or deal type and routes the request accordingly.
- •Preserving institutional knowledge through turnover
Context accumulated over years of a professional's work remains queryable by the firm rather than leaving when the individual does.
- •Governed context for other AI agents
The enterprise MCP server lets internal agents and applications draw on permission-aware organisational knowledge instead of ungoverned document dumps.
Ideal For
Best For
- ✓Large law firms preserving and scaling partner-level expertise across offices and practice groups
- ✓Financial services teams where deal and client context lives in email threads rather than a CRM
- ✓Firms that need permission-aware knowledge sharing with inherited entitlements and human sign-off
- ✓Organisations wanting to expose governed internal context to other AI agents via an enterprise MCP server
- ✓Enterprises requiring single-tenant or private-cloud deployment with US, EU or Singapore data residency
Not Ideal For
- ✗Small firms and individual professionals today — the product is sold enterprise-first and the CEO has said a self-service tier is still to come
- ✗Buyers who need published pricing or an independently audited efficiency claim, since the 30-50% communications-automation figure is customer-reported and relayed by the vendor
- ✗Organisations that want a single corporate knowledge assistant rather than per-person twins, where a horizontal enterprise search product is a simpler fit
- ✗Firms whose apprenticeship model depends on juniors doing the drafting and summarising work a twin would absorb — a risk raised in coverage of the launch
Deployment
Market & Ratings
Named customers include Linklaters, Orrick, Dechert, Customers Bank and Aegis Energy; the CEO reports a pipeline of 400+ prospects
Market Analysis
Pros
- ✓Named, verifiable blue-chip customers at launch — Linklaters, Orrick and Dechert — after more than a year of live deployments, which is rare for a seed-stage company
- ✓Governance is architectural rather than bolted on: six layers of controls, inherited permissions and human sign-off suit privileged legal and financial data
- ✓SOC 2 Type II and Google CASA Level 3 with US, EU or Singapore residency and single-tenant or private-cloud options
- ✓Explicit commitment not to train or fine-tune underlying models on customer data without written consent
- ✓The enterprise MCP server makes governed context reusable by other agents rather than locking knowledge inside one product
- ✓Founding team has a relevant prior exit in Eigen Technologies, sold to Sirion in 2024
Cons
- ✗No published pricing and no self-service tier: the CEO has conceded the product cannot stay enterprise-only, but the smaller-firm version does not exist yet
- ✗The headline 30-50% communications-automation figure is customer-reported and relayed by the vendor, with no independent audit or published methodology
- ✗Seed-stage with roughly one year of deployments, so there is no long-run evidence of how twins behave as a person's context and role change
- ✗Academic work in AI & SOCIETY argues the 'digital twin' framing systematically overstates what personalised LLMs can represent, which is a live credibility risk for the category
- ✗Coverage of the launch raised that absorbing junior drafting and summarising work can hollow out the apprenticeship model professional services firms depend on
- ✗No G2, Capterra or TrustRadius listing and no Hacker News discussion, so there is no independent user feedback on accuracy or false-attribution risk
- ✗Per-person twins over privileged data concentrate identity and access risk — a widely documented concern for digital-twin architectures generally
Pricing
Enterprise
Contact for pricing
- ✓Demo-led enterprise sales
- ✓SaaS (US/EU/Singapore), single-tenant or private cloud
- ✓Six-layer permission and governance controls
- ✓Enterprise MCP server
- ✓SOC 2 Type II and Google CASA Level 3
No pricing is published anywhere — the website routes to a demo booking and none of the launch coverage disclosed seat rates, minimums or contract structure. The commercial model is enterprise-first and demo-led, with the CEO describing a pipeline of more than 400 prospects across law firms and banks, so expect firm-wide negotiated agreements rather than per-seat list pricing. He has also stated explicitly that Twin1 cannot remain a product only for the largest institutions and that a self-service version aimed at smaller firms and individual professionals is planned, but that tier does not exist yet, so smaller buyers currently have no entry point. Deployment choice — full SaaS, single-tenant or private cloud — is likely to be a material cost variable given the infrastructure implications of per-user twins.
Security & Compliance
Sources
This page was written from 5 sources, 4 on domains other than twin1.ai.
- 1.twin1.ai — twin1.aivendor
- 2.siliconangle.com — twin1 ai raises 20m to put an ai twin behind every knowledge
- 3.pulse2.com — twin1 ai raises 20 million seed as digital twin platform aut
- 4.techstartups.com — twin1 ai emerges from stealth with 20m in funding to give ev
- 5.artificiallawyer.com — interview with twin1 ceo lewis liu
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
Subscribe