01 — Founder Perspective
In the founder's words.
Draft — pending founder approval
I was clear about what I wanted to build: a place where a lawyer brings a case to an AI and gets real work back, not a chatbot with a legal skin on it.
What I did not have was the product. First500days built the whole thing — the workspace, the toolkit, the portal behind it — and put it live.
Six months ago this was a thesis. It is now something I can put in front of users and in front of investors.
02 — Executive Summary
A product thesis, and no product.
Insaaf AI is building what it calls, on its own homepage, the first AI-powered legal operating system — a platform where a legal professional brings a case to an AI and gets usable work back rather than a conversation about it.
When First500days was engaged, that thesis was the entire company. Ideation stage, and the starting-position list has exactly one line on it: no product. Nothing to demonstrate, nothing to test, nothing to raise against.
Insaaf AI is also the narrowest engagement of its kind in this portfolio: the brief was product, and only product. No strategy retainer, no brand programme, no go-to-market. The founder knew the category and the thesis; what he was buying was a team that would build the thing.
Six months later the AI-enabled platform was built and launched: a user web application organised around the tools a legal professional actually needs, and an admin portal to run it. The company is live and fundraising.
03 — The Venture
Not another chatbot that knows some law.
The distance between a general-purpose model that can discuss legal questions and a system a practitioner will actually run their work on is the whole of this category. A model answers. A practice needs the answer attached to a matter, to a client, to a document that gets filed, and to a record of who changed what.
Insaaf AI's framing is the useful one: not an assistant bolted onto legal work, but an operating system for it. A case is opened, and everything that happens to it afterwards — drafting, research, compliance checking, building the narrative, dictating a note between hearings — happens inside the same file.
That framing sets the size of the build. An assistant is one interface and one model call. An operating system is a workspace with a matter list, a toolkit whose modules all address the same file, dictation as a first-class input, an account structure that assumes a firm rather than a single user, and an administrative layer behind all of it.
Which is why the engagement was product, singular, and six months long. There was no shortcut version of this that would have proved anything — a demo of a legal chatbot proves the model, not the company.
An assistant answers a question and forgets it. A system keeps the matter, and every tool in it is pointed at the same file.
On sourcing: the positioning line — “the first AI-powered legal operating system” — is Insaaf AI's own, taken from their homepage. The four modules named in the diagram are read off the delivered product's AI Toolkit, not from a roadmap claim.
04 — Where They Started
One line, and it is the only one that matters.
The shortest starting-position list on this site. Not because much was in place — because at ideation stage, with a product-only brief, there is exactly one thing to record.
- Product
- None
For a venture whose entire proposition is software, that single zero is the whole company. There is no interim state in which Insaaf AI half-exists: until a lawyer can open something and do work in it, the thesis cannot be shown to a user, tested against a workflow, or taken to an investor as anything other than a deck.
05 — The Challenge & Our Response
One challenge, and no small version of it.
A single pair. It is the honest shape of this engagement, and the scale of what sits behind it is the point.
Product
Insaaf AI had the idea of an AI-based LegalTech platform where users bring their legal cases to an AI and get their solution back. It was a clear thesis and an entirely unbuilt one — and in a category where every competitor can demo a language model, the thesis on its own proves nothing.
We built the AI-enabled technology solution and launched it: the user web application — matters, the AI toolkit, dictation, authentication and account structure — and the admin portal that runs behind it.
On scope: this engagement was product only. Strategy, branding, go-to-market, hiring and fundraising were not part of it, and nothing on this page claims them — the company is fundraising, and that is Insaaf AI's own work, not ours. No user, revenue or funding figures are claimed either.
06 — What We Built
A workspace, and the portal that runs it.
User web app
The workspace a legal professional signs into: matters, the AI toolkit, dictation, and the account structure a firm needs.
Admin portal
The operational layer behind the workspace — where the platform, its users and its accounts are administered.
Inside the workspace, the AI Toolkit is the front door: Smart Drafting for contracts, petitions and notices; Legal Research for case law and citation analysis; Compliance for checking documents against current regulation; and Storytelling for case timelines and narratives. Below it sits the matter list — document, client, category, when it was last touched — and beside it the voice assistant, because the moment a lawyer most needs to capture something is rarely a moment they are sitting at a keyboard.
The workspace as delivered: the toolkit, the matter list, and dictation as a first-class input rather than a setting.
The left rail carries the same idea in a different shape — Storytelling, Legal Q&A, Case Research, Doc Review and Speech to Text, each reachable from anywhere in the workspace, with a “New Project” action at the top and recent chats beneath. A practitioner does not choose a tool and then find their case; they are already in the case, and the tools come to it.
Two decisions in that build are worth naming. Sign-in supports Google and Slack, because the people this is for already live in one of those two accounts and a new password is a reason not to come back. And the account model carries roles, departments and reporting lines — the product assumes a firm with people in it, not a single practitioner with a subscription.
07 — Business Outcomes
Where the venture stands today.
Insaaf AI entered this engagement with a thesis and left it with a live platform. The product exists, it is in the market, and the company is raising against something an investor can open rather than something they have to imagine.
No user numbers, revenue, retention or funding figures are claimed on this page. The engagement was product, and the verified outcome is the product. What the company raises and who signs up is Insaaf AI's to report.
08 — Before → After
Six months of difference.
09 — The Lesson
What this proves about building in an AI category.
Insaaf AI is a one-line engagement — product, six months — in the most crowded software category of the decade. The interesting part is not that it got built. It is what had to be built for the thesis to mean anything.
The model is not the product
Anyone can put a legal prompt in front of a language model. What cannot be demoed away is the matter, the file, the roles, the record — the unglamorous scaffolding that turns an answer into work a firm can rely on. That scaffolding is the moat, and it is the part that takes six months.
Buy the build, not the advice
This founder did not need a strategy engagement. He knew his category and his thesis, and scoped First500days to the single thing he could not do himself. A client who can name their own gap gets a faster answer and pays for less.
Fundraising follows something openable
The company went out to raise after the product was live, not before. A working platform changes what a conversation with an investor is about: not whether the thing can be built, but what happens now that it has been.
In a category where everyone has the same model, the company is whatever you built around it.
Every founder starts at zero — the first 500 days decide what you are standing on at the end of them.