A few months ago, legal intake tools took a request from email or Slack and put it in a queue. This summer the category became something else. If you evaluated tools earlier this year, your notes are out of date. Ours were too, which is why we rebuilt our comparison page this week.
Here is what moved, in plain English.
The market went to where the work is
Credit where it is due. The vendors in this category made the right moves this summer.
- Streamline AI launched a Slack Intake Agent in July, so an employee can describe what they need in plain language and get a structured request without hunting for a form. The same month they released an MCP Connector so employees can ask about legal matters from inside the AI assistant they already use. In August they published a survey of 467 business professionals. One number stuck with us: 66% said they bypass formal legal channels at least occasionally to keep projects moving.
- ServiceNow opened its platform to outside AI agents in May, so for teams already running legal on ServiceNow, the front door can start to live wherever the employee already is.
- Checkbox shipped AI Agent Actions, AI Corrections that let lawyers improve answers in real time, knowledge capture, and cycle-time reporting in May.
Every one of those says the same thing: the front door to legal is leaving the legal tool and moving into chat, email, and the AI assistant an employee already has open. That is exactly what the market has been asking for, and it deserves it. Nobody wants to learn a new portal to find out whether they can sign an NDA.
What we shipped in the same window
We spent the same three months going the same direction. Since June:
- Iris went to the lawyers. Iris, the AI teammate for the legal team, now works in Slack, Teams, and email, not only the web app. Ask a question, approve a change, generate a document, and it happens in the thread you were already in.
- Every door leads somewhere. Employees can reach legal from the web, Slack, Teams, Zoom Chat, email, or a chat tool your company built itself. The moment a legal expert engages, a tracked request exists. Nobody fills out a form.
- Document review ends in Word. Every key term gets a red, yellow, or green verdict against your playbook, and the findings export into the original .docx as native comments or tracked changes. This is the one lawyers asked for.
- Documents write themselves, mostly. Upload a template, and Iris fills it from the conversation, checks every substitution so no placeholder is left behind, and delivers the finished document back into the channel that asked.
- Restricted Matters. Seal a matter and its title and contents stay out of search, chat, prompts, and integrations.
- Knowledge that builds itself, or not, your call. Every answer the legal team gives can feed the knowledge base automatically, or with a human in the loop. Teams that want to start carefully can have AI answers reviewed by a legal admin before an employee sees them.
- Reporting and insights. A Legal Throughput dashboard and a locked monthly Legal Ops Scorecard, both readable by Iris.
- Open by design. LegalOperator has an MCP Connector, so any MCP-capable tool can use the same tools Iris uses, with the same permissions.
That was 489 merged pull requests. The number is not the point. The list is.
The part that did not move
Features caught up fast this summer. Pricing models did not.
I spent about twenty years helping build and sell eBilling and matter management systems before this, so I say this from the inside. Those systems were priced the way enterprise software was priced: an annual contract, a platform fee, a count of core users, a cap on how many business users could walk through the door, and the newer capabilities sold as add-ons. That model made sense when the product was a system of record that a handful of lawyers logged into.
It makes less sense now. Look at the pricing pages in this category today and you will find annual contracts, four core users included, a business-user cap on the lower plan, and the two things the market just decided it wants most, AI answers and document generation, listed as add-ons. The number itself is usually behind a contact button. None of that is wrong. It is built for a different unit of work.
When the whole company is supposed to use the front door, every cap on business users and every add-on for the AI is a reason for someone not to.
The unit of work changed this summer. It is now a question answered in Slack before the meeting starts, a request created because a lawyer replied, an NDA generated in a thread. When the whole company is supposed to use the front door, every cap on business users and every add-on for the AI is a reason for someone not to.
So here are the three pricing questions we would ask any vendor, including us:
- Can I see the price without a sales call?
- Does the number include every employee who might ask, and the AI that answers them?
- What is sold separately, and would I need it in month one?
Our answers: yes, it is on our pricing page, $200 a month plus $20 per legal seat, unlimited business users, $1,000 one-time setup. Yes, Iris, knowledge capture, document review, and document generation are included. And nothing on the intake side is an add-on. We wrote about why we publish our pricing when we started, and this summer only made the case stronger.
One more thing
The same engine now runs invoice review and legal spend for customers who want it: LEDES upload, invoice approval chains, accruals, and a spend dashboard. It is in early access and it gets its own post. If you are carrying an eBilling contract priced the old way, it is worth a conversation.
Hold us to it
We rebuilt the comparison page this week. Every vendor row is based on that vendor's public site as of September 2026, every claim is dated, and where we could not find something, the page says "not published" rather than "no." If something on it is wrong, tell us and we will fix it. If it is right, you can see how we compare in about two minutes.
