• Red Oak named 'Visionary' in 2026 Gartner® Magic Quadrant™ for Digital Communications Governance and Archiving.

  • Red Oak Accelerate 2027 is back April 26-29 in Austin, TX. Join us for our annual client summit.

    Register →

Northwind Partners · Text · Item 88211

  1. Captured

    The message, exactly as it was sent.

    • FromDana Reyes
    • ToEllen Calloway

    “Honestly this one can't lose. I'd put the whole rollover in it.”

  2. Mira flag

    Checked against your own policy and handbook.

    • PolicyPromissory statements
    • Handbook clause§4.3

    Representatives may not promise or imply a certain outcome.

  3. Reasoning

    Why Mira flagged it, in plain words.

    Mira's reasoning

    The message promises a result (“can't lose”) and recommends moving the full rollover. §4.3 restricts promissory statements.

  4. Reviewer

    A person makes the call.

    • Reviewer's decisionMarcus Okafor

    Escalated

  5. Recorded

    Kept on your firm's record.

    • WhoMarcus Okafor
    • DecisionEscalated
    • WhenOct 19, 2026, 10:42 ET
  • Policy-grounded
  • Explainable
  • Human-controlled
  • Auditable

EXPERIENCE YOU CAN TRUST

  • 16+

    years serving strictly-regulated firms

  • 84

    Net Promoter Score

  • 1,550+

    leading firms choose Red Oak

  • 3.3x1

    faster content approvals with AI Review

  • 98%

    fewer false positives with Mira

  • 80%

    reduction in communications review time with Mira

1 Red Oak clients using AI Review, compared with firms not on Red Oak. Client-reported.

THE STANDARD

In compliance, an AI decision you can't explain is a liability.

When an examiner asks why a message was cleared, “the model decided” is not an answer. Your name is on the supervisory procedures, and the decisions made under them are yours to defend. AI belongs in compliance only when every flag traces to a policy, every decision to a person, and every change to an approval.

That is the standard we build Red Oak AI to meet.

HOW MIRA FINDS RISK

Reviewers find the risks that matter, and stop clearing false alerts

How do legacy lexicon systems work?

Keyword lists match words, not intent. A list flags “I guarantee this will be the best coffee you've ever tasted” because the word is on it, and passes “I am assured we will see a seven percent return” because it isn't.

How does Mira decide content is risky?

Mira looks at the meaning of the message and compares it to your firm's own policies, so your reviewers spend their time on real risk instead of clearing false alerts, and the promise that mattered doesn't reach a client unreviewed.

  • Your firm's policies, applied to the whole content.
  • Promissory language caught on meaning, not keywords.
  • Ordinary conversations kept out of your review queues.
  • Suggested rewording drawn from your approved language, for a reviewer to accept or reject.
Mira

Contextual AI

Northwind Partners

The same message, read two ways

Choose a message to compare

Dana Reyes to Marcus Okafor

I am assured we will see a seven percent return.

Other platforms · Lexicon list

Not flagged · no term matched

“Guaranteed” isn't in the message, so no term matches and it never reaches the queue.

Red Oak · Mira

Promissory statements · §4.3

It promises a client a specific return, which your policy covers whatever words carry it.

Dana Reyes to Ellen Calloway

I guarantee this will be the best coffee you've ever tasted.

Other platforms · Lexicon list

Flagged · matched “guarantee”

A reviewer clears it by hand. Nothing was ever at risk.

Red Oak · Mira

Not flagged · no policy applies

A remark about coffee has nothing to do with financial advice or a client decision, so none of your policies applies.

EXPLAIN

Mira builds your risk categories from your own policy handbook

Which regulation or firm policy is this finding based on?

Upload your communications policy handbook. Mira turns each supervision rule into a versioned policy program, groups them into risk categories, and sends them to your team for approval.

  • The triggering policy and the handbook clause beside every flag.
  • Risk categories built from your own written procedures.
  • Proposed policies drawn from SEC and FINRA rules and guidance.
  • Nothing runs until your team approves the policy program.
Mira

Where this flag comes from

  1. Your approved policy

    Policy program v12

    Mira builds a versioned policy program from your handbook, groups its rules into risk categories, and your team approves it before anything runs.

    Approved by Priya Natarajan

  2. Your own words

    Handbook §4.3

    Representatives may not promise or imply a certain outcome.

    Every risk category traces back to a clause in your handbook, so a reviewer reads the rule in your firm's own wording.

  3. The flag it raised

    Flag

    Promissory statements · Item 88211

    Each flag carries the policy and the clause it came from, beside the message, so a reviewer sees why it was raised.

EXPLAIN

Reviewers see the policy, the evidence, and Mira's reasoning on every flag

What evidence triggered the finding, and why does it matter?

Each flag is classified by policy area, scored for severity, and routed to the reviewer with authority to act on it. The case summary shows the policy that triggered the flag, the handbook clause it derives from, the context, the rep's history, and Mira's reasoning, then the reviewer's decision.

  • The policy and the handbook clause behind the flag.
  • The thread the message ran in, and the rep's history.
  • Mira's reasoning, written out beside the evidence.
  • The reviewer's decision, kept with what it rests on.
Mira

Northwind Partners · Text · Item 88211

“Honestly this one can't lose. I'd put the whole rollover in it.”

  • Policy

    Promissory statements

  • Handbook clause

    §4.3 · Representatives may not promise or imply a certain outcome.

  • Context

    Dana Reyes → Ellen Calloway · Rep history: annual certification complete, no open cases.

Mira's reasoning

The message promises a result (“can't lose”) and recommends moving the full rollover. §4.3 restricts promissory statements.

Escalated · Marcus Okafor

CONTROL

A person decides every case. The AI never changes itself.

Who can approve, override, escalate, or change the AI's behavior?

A person makes every review decision. The AI flags and suggests; it does not approve. Mira learns from your team's review decisions, every update it drafts waits for a reviewer's approval, and your firm decides who acts on flags and who approves those changes.

  1. Handbook

  2. Approved policy program v12

  3. Flag (item 88211)

  4. Reviewer decision

    Escalated · Marcus Okafor

  5. Drafted refinement

    “Add 'can't lose' to Promissory statements examples”

  6. Reviewer approval

    Approved · Priya Natarajan

    Only path to a policy change

  7. Versioned update

    v12 → v13

Our commitments

  • The AI never approves. A person makes every review decision.

  • Nothing runs until your team approves the policy program.

  • Every policy update Mira drafts waits for a reviewer's approval.

  • Every flag shows the policy, the clause, and the reasoning behind it.

  • Findings stay on your firm's record.

What your team can do

  • ControlDismiss, escalate, or confirm every flag Mira raises.
  • AdjustUse or set aside every rewording AI Review suggests.
  • UpdateApprove or reject each policy update Mira drafts before it ships.
  • OrganizeStart on one channel, review group, or content type, beside your existing archive.
Mira

DEFEND

Designed from day one for regulatory examination

Can we reconstruct the content, the policy, the finding, and the decision later?

Every classification, routing step, escalation, and resolution is logged as it happens. On the supervision platform, every review action lands on one audit trail, with the name of the person who took it.

  • Each classification, routing step, escalation, and resolution logged.
  • Every review action named and timed on one audit trail.
  • The policy, the reasoning, and the decision kept together.
  • The message produced whole, in its native format.
Communications supervision

Decision record · Northwind Partners · Item 88211

Policy Policy-grounded

Promissory statements. Policy-grounded means the flag cites your own policy by name, so a reviewer, and later an examiner, can see which rule it came from.

Handbook clause Policy-grounded

§4.3 · Representatives may not promise or imply a certain outcome.

Message (native format) Auditable

Dana Reyes → Ellen Calloway: “Honestly this one can't lose. I'd put the whole rollover in it.”

Rep history Explainable

Annual certification complete. No open cases.

Mira's reasoning Explainable

The message promises a result (“can't lose”) and recommends moving the full rollover. §4.3 restricts promissory statements.

Reviewer's decision Human-controlled

Escalated by Marcus Okafor to Priya Natarajan, Compliance

Audit line Auditable

Marcus Okafor · Escalated · Oct 19, 2026, 10:42 ET

One record per flag: what was said, which rule, why, who decided, and when.

PROTECT

Where your data goes and who sees it

The AI works on what your firm already holds. Mira reads the text of the messages you capture, and AI Review reads the content your teams submit, and findings stay on your firm's record.

  • Access

    One permission model, single sign-on

  • Storage

    WORM storage

  • Data

    Findings stay on your firm's record

  • Processing

    Decisions refine your own policy program

Our commitments

  • Your reviewers' decisions refine your firm's own policy program.

  • Every policy update Mira drafts waits for a reviewer's approval before it takes effect.

  • Findings and their decisions stay on your firm's record.

Trust center

Compliance-Grade AI™, answered

What is Compliance-Grade AI™?

Compliance-Grade AI™ is AI whose use and outputs a compliance team can explain, control, and defend. It combines intelligent automation with the controls, explanations, and evidence firms need to oversee its use and defend their decisions. In Mira, that means explainable findings, policy-based controls, human oversight, and an auditable decision history. The term describes how the AI is governed. It is not a certification or a regulatory approval. Mira

Why isn't a keyword or lexicon list enough?

A list matches words, not meaning. It flags “I guarantee this will be the best coffee you've ever tasted” because the word is on the list, and passes “I am assured we will see a seven percent return” because it isn't. Mira looks at the meaning of the message and compares it to your firm's own policies, so promissory language is caught whatever wording carries it and ordinary conversation stays out of your reviewers' queue. Mira

How can a firm explain an AI-assisted compliance decision to an examiner?

Keep a decision record that shows the basis for each finding and the person who decided it. In Mira, the case summary shows the policy that triggered the flag, the handbook clause it derives from, the context, the rep's history, and Mira's reasoning, then the reviewer's decision, and its line on the audit trail. Mira

Who makes the final decision when AI flags a message or a marketing piece?

A person makes every review decision. Mira and AI Review flag and suggest; they do not approve. Each firm decides who can act on flags and who approves changes to its policy program. See a demo

How does Mira learn from compliance reviewers?

Mira learns from your team's review decisions, and every update it drafts waits for a reviewer's approval. Each dismiss, escalate, or confirm decision informs Mira. It groups similar decisions across the firm, drafts refined policy logic, and releases a versioned update only after a reviewer approves it. Mira

How is client data used by AI in compliance software?

Red Oak's AI works on what the firm already holds: the messages it captures and the content its teams submit for review. Records are kept in WORM storage, and a team's decisions refine its own firm's policy program, through updates a reviewer approves. Trust center

See the decision record your examiners would see

In a demo, we take one flag from a handbook policy to the reviewer's decision, and its line on the audit trail, and show how a policy update is drafted and approved.

See a demoTake the Mira tour