A small generalist firm replaced a patchwork AI-and-research stack with a single Aewita workspace at $99/mo per attorney. Research, drafting, and playbooks in one place. Citations verified. Monthly software spend went down.
Illustrative case study based on early-pilot patterns. Named-customer stories coming as pilots graduate.
A seven-lawyer generalist firm in a mid-sized market. Some business transactional work, some employment matters, estates, a handful of active litigation files. One managing attorney, two associates, a mix of of-counsel and contract lawyers. A paralegal team that runs intake and keeps docket.
Small firm economics. Every software line on the P&L has to justify itself against billable hours.
The firm could not afford what enterprise legal AI vendors were quoting. The cheapest Harvey-class quote the managing attorney had received was, in his description, "a used-car payment per lawyer per month, plus a year-long contract, plus seat minimums I could not hit." CoCounsel was in the same neighborhood. He walked away from both.
What he had instead was a stack. Westlaw for primary research. A drafting tool with a per-seat subscription. A document management system — NetDocuments in their case. ChatGPT Plus for the associates who wanted AI help and understood the confidentiality tradeoffs well enough to redact before pasting. Two other small AI tools that had been tried and half-abandoned.
Three problems with that stack. It cost more than he wanted. The tools did not talk to each other. And he was never confident that the ChatGPT answers his associates were relying on were not fabricated — especially on citations.
All seven lawyers on Aewita for sixty days. Standard workspace. Drop the drafting subscription at the end of month one. Keep Westlaw running in parallel for the full pilot so the firm could compare on the research questions that actually came up in live matters.
They also ran a NetDocuments light integration so that documents drafted in Aewita could land in the right matter folder without copy-paste.
"I was juggling five tools and a guilty conscience about my associates pasting client facts into ChatGPT. Now I have one login and I do not have the guilty conscience. The cost math was nice. The compliance math was what actually sold me."
Monthly software spend dropped. Not dramatically, but meaningfully — the firm estimated a reduction in the mid-hundreds of dollars per month once the drafting subscription and the two small AI tools came off the bill. Aewita at $99 per attorney per month replaced more line items than it added.
Research coverage was effectively complete. Across sixty days and the firm's real matter mix, the associates reported every research question they had was answerable in Aewita. They kept Westlaw running anyway — partly habit, partly for the one or two secondary sources where Westlaw still has utility — but the primary research work moved.
Citation traceability held. Every cited authority in Aewita output traced to primary source in one click. No fabricated citations across the pilot. That under-0.3% hallucination rate at 95% CI held in their real use.
Confidence went up. The associates described the difference between Aewita and a general chatbot in the same terms the litigation-firm partners had used: "I am reading the actual case, not a summary of the case." For a firm that cannot afford a hallucinated citation — nobody can, but a small firm especially — that was the decisive difference.
ABA Rule 1.6 anxiety went down. The managing attorney had been doing firm-wide training on what associates could and could not paste into ChatGPT. He did not need to do that training with Aewita, because the architecture took the decision off the table. Client work stays in a controlled environment; the model is self-hosted, not a third-party API.
The associates ask questions in Aewita first. If an answer needs verification in Westlaw, they do that. The direction of flow reversed — Aewita is now the starting point, not the last-resort tool. The managing attorney started using the drafting features himself on simpler matters where he used to hand the work to an associate.
The paralegal team is using Aewita too, on the administrative-research work they own — court rules, local procedures, filing requirements. The firm did not add a paralegal seat; the managing attorney's seat handles enough of that work for the team to share.
NetDocuments integration is keeping documents out of email attachments and in the matter folder where they belong. It is a small thing. It is also the kind of small thing that accumulates into practice discipline.
The firm still verifies every citation used in a filing. The managing attorney still reads every brief before it goes out. The associates still write — Aewita drafts, they edit. The partner described it as "I got a better first draft. I still write the final."
They did not lay anyone off. They took on a slightly bigger docket with the same team.
The firm dropped Westlaw at the end of month three of regular use, once the associates had confirmed they were not opening it for weeks at a stretch. Their annual software spend went down meaningfully after that. The managing attorney's view is that small firms have been the most poorly served segment in legal AI, because the enterprise vendors are not structured to sell to them — and the general-purpose chatbots should not be used on client matters. Aewita, in his framing, is the first tool priced and built for a firm his size.
Fourteen days free. $720/yr if you prefer annual.