×
Twin1 AI
A per-person AI twin for regulated knowledge work
Last verified August 24, 2026 · Updated daily
What Twin1 AI is building
Twin1 pairs each individual professional with a persistent model trained only on that person's own material: their messages, email, documents, and MCP connections. The twin then works where the person works, inside Slack, Microsoft Teams, Outlook, Gmail, Google Drive and SharePoint, plus a web app. It answers substantive questions across those sources, orchestrates tools, and runs multi-step workflows on the person's behalf. The control layer is the product. Twin1 describes six sets of rules-based and AI-based controls that decide what a twin will and will not share, so that information moving between twins respects the permissions and policies the organisation already has. On top of that sits the Twin Network, a coordination layer where twins query each other, which is the mechanism that turns a set of individual assistants into an org-wide knowledge substrate without flattening everyone's data into a single index. Deployment comes in three modes, full SaaS, single tenant, or private client cloud, and the platform is deliberately model-agnostic so no foundation-model vendor gets to own the account. SOC 2 Type II and Google CASA Level 3 are done. The thesis comes straight out of what Lewis Liu concluded from Eigen. His line is that the document is often only the final outcome, and the most valuable knowledge is the context behind it. Eigen extracted from contracts; Twin1 is an attempt to capture the reasoning that produced them. Liu's framing for why it has to be per-person is that in a knowledge organisation the human is the atomic unit of knowledge, so expertise should compound rather than collapse into the average.
Why this matters
The customer list at launch is the argument. Linklaters, Orrick and Dechert on the legal side, Customers Bank in financial services, and Aegis Energy. These are firms whose security review is the moat: getting a per-person AI twin approved inside a magic-circle law firm is a harder problem than the model work, and Twin1 did it before announcing. Orrick went further and made a strategic investment, which is the tell for this company. The buyers are on the cap table. The reported effect is that the platform handles 30 to 50 percent of the communications work those knowledge workers would otherwise do themselves. That is a company-reported figure rather than an independent study, so weigh it accordingly, but note what it is measuring: not drafting quality, not research recall, but coordination overhead. That is a deliberate choice of battlefield. Liu is explicit that he is not fighting for the workflow layer. He expects platforms for legal research, document review, drafting and redlining to keep making enormous progress, and positions Twin1 as complementary to them. The bet is that when every firm has the same excellent drafting tools, the only remaining differentiator is the particular judgement of the particular people, and nobody has built the system that preserves it. Whether that is defensible or merely elegant is the open question, and it is the question a candidate should walk in with.
Open roles at Twin1 AI
4 positions we're tracking. Roles are re-checked daily and removed when filled.
Applied AI / inference engineer (San Mateo or London)
First seen today
Forward-deployed / enterprise deployment engineer (London — law-firm and bank tenancies)
First seen today
Enterprise AE, legal & financial services (London)
First seen today
Security & compliance engineer (single-tenant and private-cloud deployments)
First seen today
Know when Twin1 AI is hiring before anyone else
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Twin1 AI
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Hiring outlook
The stated use of funds is expanding the teams in San Mateo and London, investing in go-to-market, and developing the core technology, said plainly in the company's own press release. But there is no careers page on twin1.ai and no public job board, so the intent is explicit while the process is not. Hiring here runs through people, which is exactly the situation a good cold approach is built for.
Working at Twin1 AI
Founded in 2025, Twin1 AI is A per-person AI twin for regulated knowledge work. They're now people. Working at a AI company at this stage means opportunity to shape your role based on the company stage.
Most openings are based out of San Mateo, USA.
How to actually get hired at Twin1 AI
Why applying the normal way doesn't work
There's an ATS, but at Twin1 AI, referrals get priority. Your cold application competes with sourced candidates and internal recommendations.
Who to contact at Twin1 AI
What to show them
Write the security-review document Twin1 will be asked for by a firm like Dechert, before they ask you to. Take the publicly stated architecture — per-person models, six layers of permission control, a twin-to-twin network, single-tenant and private-cloud deployment — and produce the threat model: how does a twin leak across a Chinese wall between two matters, what does the audit trail look like when a partner's twin answers using material from a matter they are conflicted out of, and what would you actually instrument to prove it did not happen. Ten pages of that beats any demo.
A cold email that works at Twin1 AI
What Twin1 AI screens for
A $20M seed co-led by Bessemer, Tribeca and Aramco Ventures with named enterprise logos already in production is a Series-A-shaped seed, and several of the investors backed Eigen, so a meaningful part of the round is a re-up on this specific team rather than a bet on the category. Orrick invested and is a customer, which is the pattern worth reading: the deals came before the announcement. For a candidate, that determines where the leverage is. This company does not primarily need model researchers; it needs people who can get software through a magic-circle firm's security review and keep it running in a single-tenant deployment. There is no job board, four founders, two offices, and a mandate to spend. A specific, well-aimed note is currently a better route in than an application, because there is nothing to apply to.
Customize your CV for the Twin1 AI role. Matching the job description language helps clear ATS filters.
Don't make these mistakes
Do not pitch Lewis on the opportunity in legal document AI. He already built that company. Eigen Technologies processed over $100 trillion in financial contracts, raised more than $80M, and sold to Sirion in 2024, and his stated conclusion from it is that the document is the leftover and the context behind it is the asset. Selling him his last pitch is the fastest way to look like you read one headline. Second, do not arrive as a wrapper. The founders' whole claim is the six-layer permission model and model-agnosticism, so 'I fine-tuned a model on legal documents' is the wrong artifact to bring. Third, Jonathan Budd built J.P. Morgan's largest RAG system and co-founded Peergos, an encrypted storage protocol; presenting retrieval-augmented generation as a novel idea to him will not go well. If you want to impress that side of the house, bring something about permissioned retrieval across tenants, not about retrieval.
Mistakes that kill Twin1 AI applications
A recycled CV gets rejected fast at Twin1 AI ( people). They notice.
Don't open with what you want. Open with what Twin1 AI is dealing with right now — The customer list at launch is the argument, and what you'd do about it.
The wait-and-hope strategy fails. Follow up on day five — response rates roughly double.
Applying to Twin1 AI? Get the contact, not the form.
The Twin1 AI interview process
4 stages · 14 days typical · take-home: yes · modelled from similar companies
We don't yet have verified candidate reports for Twin1 AI. What follows is the typical process for a -person AI company — treat it as a model, not confirmed detail.
Interview stages
Recruiter Screen
Phone or video · 30 min
Basic qualification and logistics
Recruiter or HR
Hiring Manager Interview
Video call · 45 min
Role fit and experience deep-dive
Hiring manager
Technical/Functional Round
Video call · 60 min
Skills assessment and problem-solving
Team members
Final Round
In-person or video · 60 min
Culture fit and cross-functional alignment
Senior leadership
Twin1 AI take-home assignment
Twin1 AI includes a take-home exercise in their interview process. For AI roles, this typically involves a practical problem that takes 2-4 hours. Focus on clean, working code over premature optimization. They're evaluating how you think and communicate, not just the solution.
Twin1 AI interview timeline
Timeline: ~ days. That's about average than the AI median (14 days at people).
Interviewed at Twin1 AI?
Tell us how it went — stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Submit your Twin1 AI interview experience →Twin1 AI jobs, frequently asked questions
How many jobs does Twin1 AI have open?
As of August 2026, Twin1 AI has 4 open positions.
Does Twin1 AI hire remotely?
Twin1 AI doesn't have remote openings at the moment. All roles are in San Mateo, USA.
What roles is Twin1 AI hiring for?
Twin1 AI is hiring across Engineering, Sales. The most recent opening is Applied AI / inference engineer (San Mateo or London).
How do I apply for a job at Twin1 AI?
Use the apply links above, or check our guide to getting hired at Twin1 AI.
Does Twin1 AI respond to cold emails?
Response rate data for Twin1 AI not yet confirmed.
Who is the hiring manager at Twin1 AI?
At this size, hiring is usually run by the hiring manager.
How competitive is it to get hired at Twin1 AI?
Typical applicant count for AI roles ( people): 100-250 in two weeks. Apply fast.
How many rounds is the Twin1 AI interview?
4 stages: Recruiter Screen, Hiring Manager Interview, Technical/Functional Round, Final Round.
Is the Twin1 AI interview hard?
Expect technical depth and system design, not algorithm trivia. Candidates report Technical Interview as the toughest stage.
Does Twin1 AI give a take-home task?
Yes, Twin1 AI includes a take-home assignment.
How long does Twin1 AI take to get back to you?
Around 14 days across the full process.
What should I prepare for the Twin1 AI interview?
Focus on technical depth and system design. At people, they're testing whether you can operate without process, not whether you memorised algorithms.
Where is Twin1 AI based?
Twin1 AI is headquartered in San Mateo, USA.
Get Twin1 AI roles before they're posted
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching Twin1 AI
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+