Cognoverge runs competing-firm engagements behind ethical walls. A model adapted from outside counsel conflict-management practice, suitable for any AI-search agency.
Cognoverge engages multiple law firms in the same practice area and jurisdiction. Bennett Klau and Quinn Emanuel are both clients. Halverson and a Big-4 audit firm both run audits on our platform. This is by design — we cannot tell the AI search story for legal and audit if we are not deep in the actual market. It is also a conflicts problem we have to engineer our way around.
Outside counsel relationships handle this with conflicts walls — also called ethical walls, Chinese walls, screening walls. The pattern translates to AI marketing services with some adjustments. This is what we built and what we would recommend any AI-search agency working in regulated markets adopt.
Every engagement has, by default, a conflicts-wall provision in the MSA. The provision commits us to four things:
Contractual provisions are necessary but not sufficient. The wall has to be enforced by the system, not by the strategist's memory.
Every login is scoped to a workspace ID. Cross-workspace data access requires explicit re-authentication and is audit-logged. For competing-firm pairs, the system blocks cross-workspace navigation entirely for users below the partner level.
Our strategist team is sized so that competing firms can be assigned non-overlapping strategist groups. We maintain a conflicts register that the assignment engine consults before assigning a new engagement. Conflicts are detected from firm name, parent-firm relationships, and named-partner overlaps (i.e., one firm's named partner being a former named partner at the other).
The recommendation engine is workspace-scoped — it does not cross-pollinate. This is a technical commitment, not a statistical one. If we cross-pollinated, our recommendations would converge across competing firms and the moats firms are paying to build would collapse. Maintaining true isolation is an engineering cost we accept.
We do not suppress drift alerts on one workspace because another workspace saw the same drift event. Each workspace is independent, end-to-end.
Every recommendation, every lint pass, every publish event, every override, every drift alert is logged in an immutable event log per workspace. For Enterprise-tier engagements, the log is hash-chained and the export is accompanied by a sworn declaration template suitable for litigation production.
The audit trail does two things at once: it gives the firm evidence of work for their own purposes (board reporting, audit-committee documentation, regulator inquiry), and it gives Cognoverge evidence that we maintained the wall on the specific engagement.
Three failure modes we have built for, drawn from outside counsel practice:
A strategist moves from team A to team B. The migration triggers an automatic conflict check across both teams' workspace portfolios. If the migration creates a conflict, the strategist either does not get the new assignment, or one of the workspaces is reassigned.
Firm A acquires Firm B. Both were our clients on non-conflicting engagements. The acquisition closes; we are now potentially conflicted with another client. The contractual provision requires us to notify both affected workspaces within five business days and offer to terminate either engagement at no cost.
Two engagements that were not conflicted at the firm level become conflicted at the matter level — they take opposing sides in a specific case the AI search content might touch. The rulebook linter has a matter-level overlap detector that flags this on draft content. The detection runs against the workspace's stated matter universe (firm tells us at onboarding and updates quarterly).
We do not maintain conflicts walls between non-competing clients in different practice areas — IP boutiques and tax boutiques operate in different RFP markets and can share strategist teams. We do not extend the wall to public-domain information about either client — the wall is about workspace data, not about facts.
If you are running an AI-search consultancy or in-house team engaging multiple competing firms, the conflicts wall is not optional — it is a competitive necessity. Firms making seven-figure decisions about market positioning will not engage you if their competitor is on your client roster without a wall.
Adopt: a written conflicts policy, a contractual provision in every MSA, technical workspace isolation at the row level, and an annual attestation by a senior leader. That covers 80% of the substantive protection. The remaining 20% is the operational discipline of actually enforcing the policy when the strategist team gets tight — which is the test that matters.
If you want a copy of the MSA template we use, ask through our contact form. We will redact the commercial terms and share the conflicts-wall language. It is adapted from outside-counsel practice and is suitable for any AI-search agency working with regulated firms.
The free 24-hour audit shows you specifically how the eight engines describe your firm against 200 high-intent legal and compliance queries.