What happened
Bloomberg Law reports that Coinbase General Counsel Molly Abraham is redesigning the company’s legal operations around artificial intelligence. She is hiring a forward-deployed founding engineer for the law department, with a similar role planned for compliance, to evaluate legal technology and build tools around Coinbase’s workflows. Abraham also described using AI agents for memo writing and expects agents eventually to handle functions including contract review and equity grants.
Bloomberg Law published the report on Aug. 25, 2026, identifying Molly Abraham as Coinbase’s new general counsel after chief legal officer Paul Grewal’s departure. Abraham wants to re-engineer the legal group around artificial intelligence and is hiring a forward-deployed founding engineer for the law department, with a similar opening in compliance. The roles will help lawyers assess technical solutions and build automation around internal processes. Online job listings put each role’s salary at $218,000 to $256,500, excluding equity and bonuses.
Abraham told Bloomberg Law that lawyers need embedded technical expertise because they are not always equipped to evaluate legal technology and AI products. The engineer would report to Abraham and focus on understanding legal workflows. The position is funded through savings from outside counsel. Bloomberg Law presents this as an operational change inside Coinbase’s legal and compliance functions, rather than the launch of a public-facing AI product.
The report says Abraham eventually envisions AI agents handling contract review, equity grants, and other legal functions. She currently uses LibreChat, an open-source AI product, for memo writing and built “Better Call Paul” to generate initial perspectives based on Grewal’s approach. Abraham expects to consult the agent more after Grewal’s transition-services agreement ends Oct. 31. The source does not independently test the output, explain its data access, or say it can make final legal decisions.
Bloomberg Law places the strategy amid Coinbase’s regulatory and litigation pressures. The company reduced its workforce by 14% amid market volatility and an AI push and previously reduced its legal team. Abraham oversees litigation, global intelligence, and investigations, while other executives handle policy, compliance, and government relations. She previously worked at Wachtell, Lipton, Rosen & Katz, where machine learning was used for electronic discovery, and has an engineering background from Procter & Gamble. These facts explain her emphasis on translating legal and technical work but do not establish that Coinbase’s systems are effective or widely deployed.
Read the primary source: news.bloomberglaw.com ↗
Why it matters
The move shows how AI adoption is shifting from standalone legal software toward embedded technical roles inside corporate legal and compliance teams. Bloomberg Law reports that Coinbase is using savings from outside counsel to fund the engineering position, linking AI investment to the legal department’s operating budget and its role in supporting new products in a heavily regulated industry.
Bloomberg Law’s report illustrates a change in how large companies may organize AI adoption. Coinbase is placing an engineer inside the legal department instead of relying only on outside vendors or asking lawyers to select tools themselves. The structure could map repetitive work, test systems against actual legal processes, and connect automation to controls required for regulated activities. Its significance is organizational: the legal team is being treated as a place where software can be designed and operated, not only used.
The approach matters because Bloomberg Law describes Coinbase as operating amid legal, policy, and regulatory uncertainty. Abraham said legal innovation can create pathways for new products in developing regulatory fields. AI could affect how quickly lawyers review documents, prepare advice, support product launches, and communicate with regulators or outside counsel. Mistakes could create legal and operational risk, including an incorrect citation, a missed contractual obligation, exposed confidential information, or an uncertain conclusion presented as settled.
Bloomberg Law reports that Abraham wants outside firms to become more efficient through AI while expecting Coinbase’s team to assist them. This raises whether AI will reduce legal spending, redirect work toward higher-value analysis, or move costs into engineering, oversight, data protection, and validation. The report cites Grewal’s statement that Coinbase spent more than $50 million on law firms and other advisers defending a since-dismissed Securities and Exchange Commission lawsuit, but does not quantify AI savings. It also does not establish that AI caused workforce reductions or that the strategy will reduce headcount.
The story has professional-responsibility implications. Abraham told Bloomberg Law that she built an agent that writes and edits like her and that the automated version of Grewal can draft initial perspectives. Such systems may preserve institutional memory or help with recurring tasks, while raising questions about authorship, privilege, supervision, and distinguishing drafts from approved advice. Bloomberg Law does not report the safeguards governing these agents. Traceable, reviewable outputs subordinate to accountable human decision-makers matter more than the novelty of the tools.
What to watch next
The key test will be whether Coinbase can turn experimental agents into reliable, reviewable workflows without weakening legal judgment, confidentiality, compliance controls, or accountability. Bloomberg Law does not independently verify the performance of Abraham’s agents, the expected savings, or any production deployment. Future evidence should include specific use cases, human-review requirements, error rates, security controls, and measurable effects on legal work.
The first indicator will be concrete deployment evidence. Bloomberg Law reports planned uses such as contract review and equity grants but does not say autonomous agents already perform them in production. Follow-up reporting should establish which tasks are automated, which systems are used, what information they can access, and where human approval is mandatory. It should distinguish drafting assistance from systems that can trigger actions or make decisions.
Performance evidence will be important. The source does not independently confirm the quality of Abraham’s writing agent or “Better Call Paul,” and provides no error rates, benchmark results, comparison with human work, or evidence of savings. Future scrutiny should examine hallucinations, missed obligations, inconsistent advice, and failures involving changing regulations, including whether testing uses representative legal matters rather than selected examples.
Governance and security are another watch point. Coinbase’s legal and compliance teams handle sensitive corporate, regulatory, employment, and commercial information. Bloomberg Law does not explain whether the agents run locally or through an external service, whether prompts and outputs are retained, how access is restricted, or how privilege and confidentiality are protected. It also does not describe audit logs, incident response, model evaluation, or updates when laws and company policies change.
Finally, staffing will show whether the embedded-engineer model becomes durable. Bloomberg Law reports that Coinbase is funding the role through outside-counsel savings and has created a comparable compliance position. Follow-up evidence could show additional technical staff, changes in outside-firm spending, new internal controls, or broader deployment across policy and investigations. It remains unknown whether the roles will produce measurable improvements, whether professionals will adopt the tools consistently, and how responsibility will be assigned when an AI-assisted workflow goes wrong.
