Skip to content
ThetaLayer

Solutions · Revenue & customer operations

Every CRM update your agents claim, checked against the record

Lead qualification, opportunity, quoting and service agents write to your CRM all day — some supplied by your CRM vendor, some built in house. ThetaLayer checks every claimed update against the CRM's own audit trail and puts cost per qualified lead and per resolved case on one scorecard.

Book a DemoSee the pilot
Evidence from
  • CRM field history
  • Quotes & CPQ
  • Service cases
  • Mailbox audit
  • Agent runtimes
Evidence chain
Lead qualification · CRM agent
Updated lead status
Agent reported
qualified
CRM field history says
status → Qualified
ConfirmedLinked by direct reference · 95%
Logged discovery call
Agent reported
success
CRM activity log says
no activity created
UnconfirmedLinked by time window · 70%
Created quote
Agent reported
discount applied
CPQ audit says
discount above threshold · no approval
Out of mandateLinked by business key · 85%
Quote sent above the discount threshold without approval
Mandate breach · routed to the deal desk owner for review
Illustrative · example run, simulated data

The workflow

One lead, from first email to quote

Lead and opportunity agents write to the CRM in small, frequent updates. Each one is a claim until the CRM's field history confirms it.

  1. 01
    Qualification agent
    Inbound lead read
    Recorded in
    Mailbox or form audit

    Source and arrival time. Message contents are dropped before export.

  2. 02
    Qualification agent
    Lead enriched
    Recorded in
    CRM field history

    Which fields changed, under which identity. Personal values hashed or dropped.

  3. 03
    Qualification agent
    Qualified and routed
    Recorded in
    CRM status & ownership history

    Status change and new owner, linked to the run by record ID.

  4. 04
    Opportunity agent
    Opportunity updated
    Recorded in
    CRM opportunity history

    Stage, amount and close-date changes — claimed versus recorded.

  5. 05
    Deal desk
    Discount approved
    Recorded in
    CPQ approvals

    Discount level against your threshold, and who approved it.

  6. 06
    Account executive
    Quote sent
    Recorded in
    CPQ quote history

    Quote issued to the customer — the outcome the run is measured on.

Service agents follow the same pattern: case opened, agent reply, resolution or escalation — each step linked to the case history.

Direct reference
95%
Business key
85%
Time window
70%

What ThetaLayer shows you

Where CRM agents drift from the record

What ThetaLayer is built to detect in lead, quote and service workflows — the gaps between what an agent says it did and what your CRM shows.

Ledger · Claimed vs. recorded

Updates the agent reported that never reached the CRM

e.g. an agent reports it logged a call and moved an opportunity stage, but the CRM's field history shows no change — so the pipeline and the agent's own success rate both overstate what happened.

Evidence
Runtime: success → CRM field history: no matching change
Mandate · Discount threshold

Quotes above the discount threshold with no approval

e.g. a quoting agent applies a discount above the level that needs deal-desk sign-off, and the quote goes out anyway.

Evidence
CPQ: discount level · Approvals: no matching request
Mandate · Personal data

Agents writing to fields they shouldn't touch

e.g. a qualification agent edits consent or contact-preference fields, or copies personal data into free-text notes, outside its mandate.

Evidence
CRM audit: field-level change by the agent's identity
Scorecard · Resolution quality

Service agents closing cases a customer reopens

e.g. cases marked resolved by an agent and reopened by the customer days later — counted against the agent's resolution rate, not hidden in it.

Evidence
Case history: resolved by agent → reopened
Scorecard · Vendor comparison

Vendor agent and in-house agent on the same work

e.g. a CRM vendor's agent and your in-house agent compared on cost per qualified lead and write accuracy, task type by task type — so routing follows the evidence.

Evidence
Platform credits + token spend · CRM outcomes
Ledger · Identity

Agent actions recorded as a sales rep's

e.g. an agent running under a rep's delegated login, so the CRM credits the rep — and the agent's errors — to a person.

Evidence
CRM audit user vs. agent run identity

Illustrative examples of what ThetaLayer is built to detect — not results from customer deployments.

Mandates that fit this work

Field, data and discount rules for agents that write to your CRM

Revenue agents touch customer personal data and commercial terms. Mandates make the boundaries explicit — and checkable on every run.

  • Versioned: edit a rule and every historical run is re-checked
  • Detection first: breaches become exceptions for a named owner
  • Enforcement later, only at the tool boundaries you choose
Mandate · Lead Qualification AgentL3 · act within limits
  • Writable fields
    lead.status · lead.score · owner

    The agent may update qualification fields only. Any other write is flagged.

  • Protected personal data
    consent · contact preferences · notes

    No writes to consent or contact-preference fields, and no personal data in free text.

  • Discount approval
    discount > threshold → deal desk

    Quotes above your discount threshold need an approved request before they're sent.

  • Hand-offs to a person
    refunds · complaints · cancellations

    These cases are escalated, not closed by the agent.

  • Spend caps
    per lead · per month

    Caps per lead and per month, with a burn-down the owner can see.

  • Acts under its own identity
    no delegated rep logins

    So the CRM can tell agent from human in every record.

Example mandate · adapted to your policy during the pilot

Metrics on the scorecard

Cost per outcome, not cost per conversation

Metrics anchored in what the CRM recorded — qualified leads, resolved cases, approved quotes — so vendor and in-house agents compare on the same terms.

M01Cost per qualified lead
Agent spend — tokens, platform credits and allocated fees — divided by leads that reached qualified status and were accepted by sales.
M02Cost per resolved case
Agent spend divided by cases resolved and not reopened within a window you choose.
M03CRM write accuracy
Share of updates an agent claimed that the CRM's field history confirms, by field and update type.
M04Escalation rate
Share of leads or cases handed to a person, and why — by agent, version and case type.
M05Discount approval compliance
Share of quotes above your threshold that had an approved request before they were sent.
M06Protected-field changes
Agent writes to personal-data, consent or contact-preference fields outside mandate.

Every metric is defined, versioned and traceable to its source records. Allocated platform fees are labeled as allocated.

Who it's for

For the teams who own pipeline, service and customer data

Chief AI Officer / Head of AI

“Is the CRM vendor's agent worth its cost next to the one we built?”

Cost per outcome for vendor and in-house agents on one scorecard.
VP Revenue Operations

“Can we trust the pipeline data agents write?”

Write accuracy by field, with every unconfirmed update listed and linked.
Head of Customer Service

“Are agents resolving cases — or just closing them?”

Resolution, reopen and escalation rates by agent and case type.
Data Protection Officer

“Which personal data do agents touch, and under which rules?”

Field-level mandates and a record of every protected-field change.

How a pilot works here

Start with one lead or case workflow

Pick lead qualification or case handling, and one agent runtime. We connect read-only to the CRM's audit trail and the agent's traces, baseline cost per qualified lead or per resolved case, and draft field and discount mandates with your owners.

See the pilotHow it works
Pilot scope · Revenue & customer opsSix weeks · read-only
Workflow
Lead qualification or case handling
Agent runtime
One CRM vendor agent or in-house runtime
Systems of record
CRM · CPQ or service desk · mailbox
Access
Read-only. No changes to your agents.
Success criteria
e.g. a target reduction in cost per qualified lead at equal volume, or a defined write-accuracy threshold
  1. Week 0
    Connect
  2. Week 2
    Baseline
  3. Week 4
    Mandates
  4. Week 6
    Decide
Does this work with agents built into our CRM?
Where the platform exposes session or activity exports and the CRM keeps field history, yes. ThetaLayer is built to read from vendor agent platforms such as Salesforce Agentforce and Microsoft Copilot Studio as well as in-house runtimes. Connector availability is confirmed during pilot scoping.
How do you handle customer personal data?
The collector runs in your network and minimizes events before export: personal values are hashed or dropped, and message bodies never leave. The control plane sees which field changed, by whom and under which mandate — not the customer's details.
What counts as a qualified lead or a resolved case?
You define it, and we version the definition. For example, a lead counts once sales accepts it, and a case counts as resolved if it isn't reopened within a window you choose. Every metric links back to the records behind it.
Can ThetaLayer stop an agent from sending an over-discounted quote?
Not at the start. ThetaLayer begins read-only and flags breaches for review. Enforcement can be added later at a tool boundary you choose, such as quote submission.

Book a Demo

See a CRM agent's claims checked against the record.

Book a 30-minute walkthrough. We'll show how claimed updates are reconciled with CRM field history, and scope a six-week, read-only pilot on one of your lead or service workflows.

Book a DemoSee the pilot