AI first drafts for accounting, consulting, and agency deliverables. Your practice software, your document tools, and a new vertical vendor each month all offer AI. We put an engineer with one practice team to choose what is worth buying and to automate the first draft of one recurring deliverable, such as a monthly client report.
Bring the last 20 examples of one recurring deliverable.
Published results in professional services
Two firms that kept their method in-house: one built on tools it already licensed, the other built on top of a vendor.
A BDO USA senior tax manager built a tax-return review agent in two days, and it saves one to two hours on every return.
Built on AI tools BDO already licensed. It organizes and drafts, and the reviewer validates and decides. 45 colleagues were using it soon after launch.
A&O Shearman's ContractMatrix, built on Harvey, cut contract review time by 30%, an estimated saving of up to 7 hours per review.
A ring-fenced sandbox and governance rules came first. The firm then built its own tool on Harvey's API and Azure OpenAI, grounded in its precedents.
Figures as reported by the vendor or customer at the linked source. They are not OpenNash results.
Where our engineer comes in
BDO's agent was built in two days by a tax manager who knew the review process, on tools the firm already paid for. A&O Shearman bought Harvey, then built its own tool on top, grounded in its precedents. Your templates, review standards, and method are the same kind of asset. Our engineer works inside your practice systems to keep them in-house and buy only the commodity pieces.
| Decision | What we look at | Outcome |
|---|---|---|
| Which deliverable first | The last 20 examples of each recurring deliverable, with time from source data to sent. | The deliverable that uses the most senior hours. |
| Buy or build | AI features in Karbon, Kantata, or your document tools, against what the deliverable needs. | Buy the commodity steps. Build the parts that carry your method. |
| How client data stays separate | Retrieval scope per client and per engagement. | A draft only sees that client's files. |
| Who signs off | Review steps and named reviewers. | Nothing reaches a client without a named reviewer's approval. |
Why forward-deployed engineers
Built for their largest accounts, deploying their own models. OpenAI's deployment company launched with more than $4B. ↗
Configure their own product, priced per seat or per resolution.
A small senior team, ex-Palantir, Meta, and Databricks. Any model, any vendor. Free pilot, then a flat monthly fee.
Even OpenAI closes the gap with engineers in the room. Source: OpenAI, What FDE means at OpenAI, Oct 2026.
The first eight weeks
Written down with your team in week 1. The pilot is free, and it has to pass these before anything goes further.
Example acceptance tests, not results from a customer deployment. Yours are written with your team.
Systems we work in
Rules from day one
FAQ
They are built from your templates and past deliverables, and a senior reviewer scores voice and structure against the rubric before the pilot passes.
Retrieval is limited per client, so a draft for one client cannot see another client's files. Every access is logged.
It prepares a first draft from approved source material. We measure drafting time and review effort during the pilot, and your senior people still sign off.
Why you need one
A senior engineer joins your team on a laptop you issue, tests the options on your data, and puts one professional services workflow into production in 4 to 8 weeks. It starts with a free 14-day pilot judged against test cases you agree in writing.
After launch, we run and improve the workflow on a flat monthly fee, or your team takes it over. The code, tests, and runbook are yours either way.
Bring the last 20 examples of one recurring deliverable.